Socratic Puzzles

......
SOCRATIC PUZZLES
Copyrighted material
SOCRATIC PUZZLES
Robert Nozick
Harvard University Press
Cambridge, MassachusettS London, England ~ 1997 Copyrighted material
Copyright e 1997 by Rob<n Nouck All rights reserved
Primed in the United States of Ameri<:a Designed h)' Marianne Pe.rlak Library of Congrt1s Catalogingin-Publication Data Noz.kk, Roben. Socr;utc puul/ Robert Notick. p. em.
Includes bibliographical references and index. ISBN 0-674-81653-6 (alk. paper) J. Methodology. 2. Socr.ares. 3. Philosophy. 4. Political science-Philosophy. S. Ethics. I. Tide. B945.N683563 !997 J91.- <k21 96-39221
To the memory of my parents, Max Nozick (1906-1990) and Sophie Cohen Nozick (1908- 197 5)
Copyrighted material Copyrighted material
CONTENTS
lntroducrjoo 1 CHOICE AND UTILITY
1 Caercjoo 1 5
2 Newcomb's Problem and Two Principles of Choice 45
3 Reflections on Newcomb's Problem 74 4 Interpersonal Utility Theory 85 5 On Austrian Methodology 110

PHI!.OSOPHY AND METHODOLOGY 6 Socratic Puzzles 14\
7 Experience, Theory, and Language 156 8 Simplicity as Fall-Our 182 9 Invisible-Hand Explanations 191
BTHICS AND POI.! TICS
10 Moral Complications and Moral Structures 201 11 On the Randian Argument 249 12 Weighted Voting and "One-Man, One-Vote" 265
viii ..,. Contents
DISGI!SSIONS AND RBYIBWS
13 Goodman, Nelson, on Merit, Ae.sthetic 273 14 Who Would Choose Socialism? 277 15 Why Do Intellectuals Oppose Capitalism? 280
16 The Characteristic Feature.s of Extremism 296
17 War, Terrorism, Reprisals- Drawing Some Moral Lines 300 18 Do Animals Have Rights? 305
19
20 21
22
PHI!.OSOPHICAI. FICTIONS
Fiction 313 R.S .V.P. - A Story 317 Testament 324 Teleology 329 Notes and References 333 Credits l91 Name Index 393 Subject Index .397
...... SOCRATIC PUZZLES INTRODUCTION
It is disconcerting to be known primarily for an early work. Others have identified me as a "political philosopher, but I have never thought of myself in those terms. The vast majority of my writing and anention has focused on other subjectS. Anarchy, State, and Utopia was an accident. It was written during 1971-72, a year spent at the Center for Advanced Study in the Behavioral Sciences, adjacent to Stanford University, and my plan for that year was to write on the topic of free will. Social and political philosophy was a serious but not a predominant interest of mine. After several months of gening nowhere on the topic of free will, I was invited in early December to give a talk to a Stanford University student group, and I presented some thoughts on how a state would arise out of (individualist) anarchy. Those thoughts seemed worth writing up, and that was done by the beginning of January, when the mail from Cambridge brought john Rawls's long-awaited book, which I had read in manuscript and discussed extensively with Rawls. In lectures at Harvard, I had been developing the entitlement theory of justice, within a libertarian framework, and when I read the final version of A Theory of Justice that January, I was moved to write down the entitlement theory along with my critical thoughts on Rawls's theory, which had developed significantly since our last discussions. By the end of February, I had on my desk an essay on state-of-nature theory, an essay on distributive jus tice, and also an essay on utopia that I had delivered at a meeting of the American Philosophical Association and brought along with me to Stanford. These three pieces fit together and it seemed that with some integration and some additional material they might constitute a book. So, rather than returning ro the intractable problem of free will, I set to
intensive work to complete that book before my stay at Stanford would be up in July.
2 ..,. Introduction I have not responded r.o the sizable literature on Anarchy, State, and Utopia, or followed it closely. I did not want to spend my life writing "The Son of Anarchy, State, and Utopia," "The Return of t:he Son of ... , " etc. I had ot:her philosophical questions to think about: knowledge, the self, why there is something rather than nothing, and-of course-free will. Also, I believed that responding to criticisms of that book or re"'ding them closely would not be a good way for me to gain any critical distance from its positions. My natural tendency would be to defend strongly those positions under attack- and attack it was!and how could I learn that my views were mistaken if I thought about them always with defensive juices flowing? (I returned briefly to express reservations about some of Anarchy, State, and Utopia's views in The Examined Life, pp. 3Q-32, 286- 296.) I have not responded to the literature on my other writing either. I could say t:hat any intellectual contribution I could make was likely to be greater in broaching new ideas on new subjects than in elaborating or defending my previous ideas, and that would be true, but also it is true that what pleases and excites me is to think new t:houghts about new topics. I have learned much from philosophers who stay with one area or one theme, gaining deeper insight, but my bent is elsewhere. Philosophers seek structures telling how things are related and constituted. They find complexity in what appears simple, and simplicity in what appears complex. Philosophy's understanding is structural, its method is clarity of thought and reasoning. Socrates, at western philosophy's beginning, made critical thinking an
explicit. methodical activity. The aim was to bring everything to conscious, explicit, and critical awareness: the definition of important concepts, the articulation of principles of conduct, the methods of reasoning themselves. Philosophers since then have been indebted to Socrates not only for his intellectual daring and incisive trains of thought, but for the vivid and inspiring way he embodies what a philosopher is. Medieval thinkers paid tribute to the scope and depth of Aristotle's theories by referring to him as "the philosopher." But when I think of " the" philosopher, I think of Socrates. Scholars have disagreed over central questions about this sharp and clear thinker. What did he mean when he said he did not know the answers to his questions, what does his method of critical discussion presuppose, and what is his concern in imCopyrighted material Introduction -& 3
proving the souls of those he encounters? In "Socratic Puzzles" (Essay 6), I propose some answers. The intellectual problems and puzzles in the other essays of this volume' are not all of them "Socratic," if that means focusing on the particular questions Socrates investigated. However, the writing here grows out of what Socrates did and is continuous with it. Sometimes an interesting question motivates analyzing a concept, sometimes a central notion leads to a new and puzzling question, and sometimes these links iterate to form a longer chain of reasoning. That is true of much of philosophy, of course, so while the title of this book marks it as philosophy, it does not specifically describe its contents. Rather, it pays
Socrates homage. One of philosophy's central tasks bas been to formulate and justify rules, norms, and principles to help guide us through the welter of possibilities we face, different possibilities of action, of belief, of ways we could be. Let me indicate how the essays here 6t within this philosophical task, and how they connect to my other writing. Normative principles of action wil.l make actions depend upon our beliefs (about what is the case, and what will happen if we act a certain way) and upon our desires or evaluations (what we want to happen or value happening). But what normative rules or principles should our beliefs and desires satisfy? And, given (appropriate) beliefs and desires, what specific principle or rule should then select the action? The most widely discussed rule mandates that our actions maximize expected utiliry. Along with normative conditions on the structure of preferences, this rule is pan of a normative theory of decision that was formulated by economists and mathematicians in the midtwentieth century. Philosophers, with their interest in explicit normative or rational rules, were not slow in taking up this theory. I was drawn seriously into the study of philosophy in Sidney Morgenbesser's philosophy classes at Columbia College, and there I first encountered these decision
theories. I was led to Morgenbesser's classes in particular by my earlier experience in a section that he taught of a required general education course at Columbia on twentieth-century social and political thought. Every time I said something in that course, Morgenbesser raised an objection-[ was overlooking a distinction or ignoring an objection or failing to consider how some structural feature of my point might generalize-and
I am ver)' grateful to Tamar Ge.ndlr.r for her persuasiveness in wging me to bring
together a volume of essays and for her help in assembling this work for press.
Introduction his objection would convince me that I wasn't thinking clearly enough, and that there was a clearer way to think. Thus I was lured to take more courses in the hope of learning to think that cle.arly-soon I had amassed enough courses to "major in Morgenbesser." It was in gradu ate school at Princeton, studying with Carl Hempel, that I took a really close look at decision theory, and a projected dissertation (spurred initially by Hempel's writing) on the confirmation and acceptance of scientific hypotheses, after some months and some work, was transformed into "The Normative Theory of Individual Choice" (1963; published in 1990 by Garland Press). Here, I discussed normative conditions on preference, the theory of choice under certainry, risk and uncertainty, and some issues about game theory. It was in the dissertation that I first discussed Newcomb's Problem, a problem in decision theory thought up by a physicist in California, William Newcomb, and told to me at a parry by a mutual friend and physicist at Princeton, David Martin Kruskal. (It was, for me, the most consequential party I have attended.) This intriguing problem was first published in "Newcomb's Problem and Two Principles of Choice"
4 -<>
(Essay 2); the treatment there differed from that in the dissertation. Martin Gardner wrote a "Mathematical Games" column on the problem in Scientific American, and he asked me to do a guest column (Essay 3) responding to the vast amount of mail he had received. I returned to the problem in The Nature of Rationality (pp. 40-50), where I propose the maximization of decision-value (a weighted sum of causally expected, evidentially expected, and symbolic utility). Although three distinct positions on Newcomb's Problem are a small fraction of the positions in the by-now vast literature, they are more than enough for one writer! Decision theory ("rational choice theory") has received extensive application in the social sciences, and it provides the structure for some of my later philosophical work. It plays a role in Anarch); State, and Utopia's thoughts about how individual action would give rise to a state, and also in its game-theoretic analysis of utopia; in The Nature of Rationality I reformulate decision theory and apply it to issues of rational belief. Decision theory provides the implicit background structure in the essay on "Coercion" (Essay 1), and that part of decision theory concerned with the structure and measurement of preference (utility theory) is one of the models for the theory of moral structures, and the measurement of moral weight, presented in "Moral Complications and Moral Structures" (Essay 10). Another model was Noam Chomsky's Sy11tactic Structures, and I meant "Moral Complications" to begin to uncover the structure underCopyrighted material Introduction ~ 5
lying our moral judgments. The essay presents several possible structure. s (the maximization structure, the deductive structure, and the simple balancing structure) before discussing the more complicated moral balancing structure with its own special governing principles. I added
a bit to the formulation a.nd investigation of alternative moral structures in the presentation and discussion of "side-constraint.s" in Anarchy, State, and Utopia (pp. 26-42), but with rare exceprions, the task has not been pursued very extensively by philosophers since then." One of the new principles in "Moral Complications" (Principle VII) is put to use in a review of Michael Walzer's Just and Unjust Wars (Essay
17)
to sharpen his proposal about what risks should be incurred to protect civilians in warfare. Utility theory, the component of decision theory concerning the structure of rational preferences and of their measurement, has been of interest to normative social theory, especially within the utilitarian tradition which mandates maximizing the total happiness or the aggregate sum
of
utility. That tradition's use of the utility notion, and some other uses as well (hence the interest of the problem, even for critics of utilitarianism), faces the well-known problem of interpersonal comparisons of utility: how can we systematically compare the strength of different people's preference.s or desires? "Interpersonal Utility Theory" (Essay 4) presents a new strategy for attacking this problem and makes specific new proposals. A normative concern about what conditions an individual's choices should satisfy fits in naturally with a concern about the range of choice a person should be free to make. Given my background view that coercion is justified only under very stringent conditions, I was led to investigate the notion of coercion directly (Essay 1 ). Philosophers had nor previously looked closely at this notion, and it presented interesting problems, nor simply of formularing an adequate definition that would handle a range of complicated cases, bur of understanding why the
contours of the notion of coercion rake the particular form they do. Why should threats be an important component of coercion when offers also can establish a significant difference between the utility yielded by two actions? The answer went beyond the utilities involved in the actual
However, the structure of "prim.:a facie .. priociples, investigated in one direction in "Moral Complications, .. has been de\eloped in other intricate and fruitful directions by writers in the field of artifici-a_l intell igence studying defaul1 strucrures; for insrance, in John Holland, Keith Holyoak, Richard Nisben, and Paul Thagard, lndtlction:
Prou.sses
of Inference. Learning, and Discovery (('.ambridge, MA: MIT Press, 1986). Copyrighted material
6 -t> Int roduction choice siruarion (after rhe rhrear or offer has been made); ir lay in the hyporherical choice of entering into rhose rhrear or offer siruarions. A person normally would be willing ro be rhe recipient of an offer yet be unwilling ro be rhe recipient of a rhrear. The difference in willingness ro move iuJo rhe situation makes a difference in the voluntariness of rhe action that is done within rhe situation. A moral evaluation must look beyond the immediate situation and beyond rhe agent's first -level desires within that siruation. (The lirerarure on coercion has grown greatly since my essay appeared. For a discussion, see Alan Wenheimer, Coercion [Prince.ron, Nj: Princeton University Press, 1987].) The issue of how wide a range of factors must be considered by moral or eva luative principles recurs elsewhere. The moral treatment owed ro a person who is mentally retarded, I claim in a review of Tom Regan's bc>ok where he discussed rhis topic (Essay 18), depends not only on rhe traits he or she possesses but also upon rhe typical and normal traits possessed by o1embers of rhe humao species, even when the panicular person being rrcatcd lacks so.me of these rrairs. (The closest-continuer
theory of identity over rime, in Philosophical Explanations, pp. 29-70, also widens a context: ro determine whether two entities are identical, one must look beyond the traits and rela tions of those two entities themselves.) Another principle of normative evaluation is subject to examination in "Goodman, Nelson on Merit, Aesthetic" (Essay 13), where I criticize Goodman's view that a work has aesthetic merit if it is a work of art and changes the way we view the world. That essay contains my only published poem; readers will discover why I want them to think it is a poor poem. "The Characteristic Features of Extremism" (Essay 16) was written for a 1987 conference in Jerusalem (organized by the American j ewish Congress) on pol itical extremism in the United States and Israel, and it was presented in a panel with Geula Cohen, a member of the Israeli Knesset. (Afterwards, some members of the audience asked how, without having met her beforehand, I could so accurately describe rhe other panelist!) The phenomenon of extremism has grown even more widespread and serious since then. A United Stares Supreme Coun decision (Baker v. Carr) formulates the principle that every person's vote is to count equally, and Essay 12 applies forma l apparatus co tl1at decision (and in passing presents a counter-example ro the Shapley-Shubik power index). Preference and decision fall under normative rules, and so does belief. Rules encourage. or mandate some beliefs, and they discourage
Introduction -i> 7 or exclude others. Such rules also might be used to criticize the beliefs of other people-always a satisfying activity. (Surely that was part of the appeal of the verifiability criterion of meaning- now known to be defective-put forth by the logical positivists. Allied with the rule not to believe or pay attention to anything (cognitively) meaningless, it allowed one, without further thought, to dismiss whatever was not verifiable, and to criticize others for indulging in meaningless
talk.) Rules of rational belief, and the justification of these rules, has been a-perhaps the-longtime interest of philosophers. (I investigated such rules recently in The Nature of Rationality.) One rule often proposed is to believe the simplest hypothesis compatible with the evidence. It has been a puz.zle why the simplest hypothesis is most likely to be the nue one. "Simplicity as Fall-Out" (Essay 8) shows how a procedure for fixing belief, without seeking simplicity, might lead one nevertheless to end up believing the simplest hypothesis compatible with the evidence. Thus it proposes an invisible-hand explanation of simplicity. The reviews of Walzer and of Regan are the only two book reviews I have written. Perhaps this surprising paucity of book reviews is explained by my first attempt at one in the late 1960s. I agreed to review H. L. A. Han's book of essays, Punishmerzt arzd Respomibility, for the Harvard Law Review, and after reading this admirable work I set out with what seemed to me a reasonable plan: first, to develop the correct theory of punishment, and then, after doing so, to stand on that platform to state what was right and what was wrong with Hart's views. The review never got written. (The thinking done then, however, found its way into the discussions of punishment in Anarch)\ State, and Utopia and Philosophical Explanations.) My interest i.n the minimization of coercion led me ro investigate "li bertarian" views and arguments. "Who Would Choose Socialism?" (Essay 14) formulates one such argument, looking at the percentage of Israelis who choose life in a kibbutz. "Why Do intellectuals Oppose Capitalism?" (Essay 15) considers why intellectuals are especially resistant to capitalism and to arguments in its favor, and it hypothesizes a new sociological law. The e.ssay on Ayn Rand, "On the Randian Argument"
(Essay 11 ), examines her moral arguments, which ac.adem.ic philosophers previously had neglected, but since it neared them with the standard critical tools of the nade, it brought me vituperation from some of Rand's followers. It is interesting to notice the themes of pan I of Anarchy, State, and Utopia that are first adumbrared in this essay: Copyrighted material 8 ~ Introduction side-constra ints and rights, whether the night-watchman stare of classical liberal theory is redistributive, whether what matters is only people's
experiences.
This aspect of early work holds true of other philosophers, as I discovered once while reaching a graduate seminar where the work of contemporary phi losophers was looked at with the aim of understanding how they generated their ideas and theories. In addition to mature work, the seminar looked also at some early papers, where (I assumed) working methods of thought would be closer to the surface. What surprised me-in the early writing of Quine and Goodman, for examplewas how many of these philosophers' later striking ideas were presenr already in these very early works, for those who had eyes to see. The essay on Ayn Rand was written as a sidenote to my own independent thinking about topics in social philosophy and the moral foundations of capitalism. The e.ssay on Quine, "Experience, Theory, and Language" (Essay 7), however, grew not from thinking independently about topics in the philosophy of language but from thinking about the ideas of Quine himself. Despite its particular points and its discussion of some new questions (such as why Duhem's thesis holds), I was struck, in rereading it soon after its writing, by the difference between thinking independently about a subject and thinking that is focused upon someone
else's ideas. When you approach a topic through the route of someone's theories, that person's mode of structuring the issues limit.~ how far you can stray and how much you can discover. You think within their "problematic." Psychologists have investigated a phenomenon they call "anchoring and adjustment." For example, a subject is asked to estimate a per son's height by estimating how far that height deviates from a fixed benchmark-from, say, five feet tall. If he thinks the person is five foot seven, he says "plus seven inches." The interesting fact is that the expressed estimates of a person's height will differ, depending upon the fixed benchmark. In theory, that particular benchmark should make no difference. Ii it is six feet instead of five feet, then the height of someone who is five foot seven can be said to deviate by "minus five inches" from that taller benchmark. Any given height can equally well be located by its distance from either benchmark. Nevertheless, the estimate of a given person's height by a group of judges who start with the five foot benchmark will be less than the estimate by a group of judges who start with the benchmark of six feet. It is as though the benchmark exerts a gravitational anraction on the estimate, pulling that estimate
Copyrighted material Introduction -o- 9
toward itself. It is similar, I think, when you approach a topic through the thought of another. Even when your own conclusions do deviate, they are "gravitationally" pulled toward those of your source. I cannot say, however, that I have consciously refrained from writing extensively on historical figures in order to avoid being trapped within their mode of thinking. I simply have not felt the urge to do so-except in the case of Socrates. I do not work within the framework of any one philosopher's thought but l do draw upon a k.itbag of intellectual structures (in addition to
the tools of mathematical logic which all contemporary philosophers use). Decision theory (and the associated material of utiliry theory, game theory, and social choice theory) is one structure I keep in mind, and use, and ticker with developing. There also is evolutionary theory, microeconomic theory, and inductive logic. I find such intellectual structures attractive and lovely, and I tend to scout other fields for already developed structures that might find philosophical application or might inspire the development of an analogous new philosophical structure. Another structure I draw upon is presente.d in models of explanation, formulated by philosophers of science. Because of the presence of Carl Hempel at Princeton, the topic of explanation was central in the intellectual life of irs philosophy graduate students. (And it was an admiration for Hempel's work, promulgated in classes at Columbia with Sidney Morgenbesset and with Ernest Nagel, that was the major factor leading me to do graduate study at Princeton.) Hempel's models of deducrivenomological and statistical explanation, their elaborations, difficulties, etc., was the lhtgua franca of study there. Everyone, no matter what his interests-unfortunately, there were no "hers" there then- felt obliged to keep current on that subject. It is no accident that my second book is entitled Philosophical Explanations, and that it focuses upon that notion, even though in ways different from Hempel. (Anarchy, State, a11d Utopia also was concerned with explanation: with fundamental explanarion and with invisible-hand explanations of the state.) Essays in the present volume also are concerned with explanation. "Invisible-Hand Explanations" (Essay 9) investigates that notion further, including new instances beyond those discussed and presented in
Anarchy, State, a11d Utopia. "On Austrian Methodology" (Essay 5) sharpens and then examines the claim of "methodological individualism," made by economists in the Austrian tradition, that all of the institutional structures of a society can and must be explained in terms of the actions of individuals. It also contains a critical examination of their
tO ~ Introduction
thesis that social science should be based upon an a priori theory of hu man choice. Thus the interests in theories of choice and in explanation come together. A philosophical interest in explanation also is normative in part. We want to understand understanding, but we also seek explanations and therefore do not want to believe what could not be an adequate ex planation. An account of what structure an explanation must possess thereby specifies our aim and guides us in our beliefs. " Experience, Theory, and Language" (Essay 7) pursues the theme of explanation in several directions: What explains why we cannot test isolated explanatoC) hypotheses but only large conjunctions of them? What is linguistics needed to explain? The common intellectual interest in explanation was a fruitful stimu Ius for philosophical interchange at Princeton. At Columbia in the late 1950s, C. Wright Mms had just published The Power Elite, and no mat ter what their subject of study, students bad to know what they thought about its thesis of a ruling elite in America. (In Essay 9, I speculate about an invisible-hand explanation of a ruling class.) Thomas Kuhn's Struc tre of Sciet~tific Revoltiot~s held this focal position at many universities in the 1970s, and perhaps sociobiology did for a while afterwards. I sometimes ask students whether there is any serious work of non
fiction which all (or most) of their friends have read or feel they have to read. For the past fifteen years or so, the answer has been no. I do not know why there now is no such common intellectual idea or book that students debate with interest and excitement-have films displaced books as srudents' common focus?-but it is unfortunate for srudents (and also for their teachers, who cannot take for granted the students' passion over any recent ideas). Philosophers seek intellecrual understanding but phi losophy, and what propels it, is not solely intellectual. This volume closes with several philosophical fictions (selections 19-22) that bring forth the emotional content of philosophical problems and ideas. These different aspects of philosophy are explored more fully, and more personally,- in The Examined Life. Philosophy is endlessly interesting and ample; a11ything can be thought about philosophically. The philosopher's deepest urge, though, is to articulate and understand the ultimate basis and nature of things. What is the basis of our beliefs and our ethical principles, our standards of rea
Introduction -& 11 soning and of justification? What is the fundamental nature of mind, matter, identity over time, causaJ;ty, knowledge, free will, truth, and consciousness? And how are all these to be put together into a meaningful whole? What could be more worth thinking about? And what could be more ennobling than thinking about these things? It is, Aristotle said, the most godlike of human activities. Philosophy is not just thought. It is a way of life too, infused by thought and transformed by it. I wrote about these matters in The Examined Life. Ordinary speech rakes "being philosophical" to mean being calm about setbacks, being dispassionate, balanced, and distant
about tragedies and calamities. I certainly bad no desire to cultivate that attitude when, excited by philosophy's intellectual methods and insights, I entered the subject. In late 1994, however, when stomach cancer led to serious surgery followed by months of chemotherapy and radiation treatments, along with dire statistics (never about the most relevant reference class, I thought, and in any case wasn't it an individual probabilistic propensity that I needed to know?), I found, even as I took vigorous steps to avoid what most probably would occur, that I had no complaints and felt no distress. My fifty-five -year lifespan already had been longer than that of most people in human history. (I maintain it was not a complaint when the first words I said to the surgeons upon coming up from anesthesia after seven hours were "I hope we don't have to do this again. I don't have the stomach for it.") And I had no urge to transform my life i.n whatever time remained. I harbored no hidden desire to run off to Tahiti, or to become an opera singer, or a racing car driver or a Dea n. I wanted to continue loving my wife and my children, having fun with them, and doing exactly the things I had been: thinking, teaching, and writing. Only now I needed to do them better. And sooner. It has been simple to accept my own situation calmly; a danger to those I love, however, would prompt all-out warfare. Nietzsche demanded that you should live a life you would be willing to repeat infinitely often. That seems a bit stringent. Philosophy, however, does constitute a way of life that is worth continuing to its end. As Socrates showed us at the beginning.
CHOICE AND UTILITY
1 COERCION
This study of coercion is intended as a preliminary to a longer study of liberty, whose major concerns will be rhe reasons which justify making someone unfree ro perform an action, and the reason why malcing someone unfree to perform an action needs justifying. Though coercion is intimately connected wirh liberty (some writers capsulize freedom as absence of coercion), it does not exhaust the range of nonliberty or unfreedom. In particular, being coerced into not doing an act is neither a necessary nor a sufficient condition for being unfree to do it. That it is not necessary is shown by the following examples: (a) A person robs a bank and is caught and punished. If be knew for sure be would be caught and punished for robbing the bank, he would nor do so, but he does nor know this and so robs the bank. He was unfree to rob the bank, though he was not coerced into not doing so. (b) I was not coerced into not murdering a member of the audience at Columbia when I read this paper, though I was unfree to do so. (c) If I lure you into an escape-proof room in New York and leave you imprisoned there, I do not coerce you i.nto nor going to Chicago though I make you unfree to do so.
Copyrighted material 16 -<> CHOICE AND UTILITY
That being coerced into not doing act A is nor a sufficient condition for being unfree to do A is shown by the following example: You threaten to get me fired from my job if I do A, and I refrain ftom doing A because of th is threat and a.m coerced into not doing A. However, unbeknownst to me you are bluffing; you know you have absolutely no way to ca.rry
out this threat, and would not carry it our if you could. I was not unfree to do A (no doubt I thought I was), though I was coerced into not doing A. But though coercion does not exhaust the notion of unfreedom, iris obviously closely connected to it.1 This paper attemptS to clarify the concept of coercion, and some related conceptS. Though this is an interesting and intriguing task, I do nor pursue ir for its own sake. I am primarily interested in the uses to which such a clarification can be pur; rhe questions one will be in a better position to answer given this clarification (other than questions like: what are the necessary and sufficient conditions for P's coercing Q into not doing A, or for P's threatening Q). l shall nor be able here ro get ro these furrher questions, and shall be engaged largely in tool sharpening rather than in tool use. One final preliminary remark, or warning, or apology. This study of coercion is an exploratory study, and is meant ro raise questions and suggest problems. To many of these questions and problems I propose tentative answers and solutions, bur some are left open. I would be happier if I could answer or solve them all. Bur in philosophy questions and problems often outlive specific proposed answers and solutions. Unfortunately, so it was in the course of writing this paper. Conditions for Coercion I shall begin by considering an account of coercion obtained from combining some things said on this subject in Harr and Honore's Causation in the Law with some remarks of Han in his The Concept of Law. 2 According to this account, person P coerces person Q into nor doing act A if and only if (1) P threatens ro do something if Q does A (and P knows he's making this threat). (2) This threat renders Q's doing A subsranriaUy less eligible as a course of conduct than not doing A.
Coercion -& 17 (3) P makes this threat in order to ger Q nor to do A, intending that Q realize he's been threatened by P.
(4) Q does not do A. (5) P's words or deeds are part of Q's reasons for not doing A .3
Conditions 1-5 do not appear to be sufficient for coercion . . For example, P threatens Q, saying that if Q performs a particular action, a rock will fall and kjlJ him. P thinks Q knows of h.is ( P's) infamous procedure of murdering people, but Q thinks that P is telling him about some strange natural law that holds independently of human action, namely whenever someone performs this action, he gets killed by a falling rock. That is, Q understands what P says, not as a threat but as a warning. If Q refrains from performing the action, P has not coerced him into not doing it, even though the five conditions are satisfied. This suggests that we add as a further condition: (6) Q knows that P has threatened to do the something mentioned in l, if he, Q,does A.4 It is not clear that the conditions thus far listed are sufficient. You threaten to do something if I do A, thinking that I don't want thi.s something done. But in fact, I don't mind it or even slightly want it. However, I realize that you must feel very strongly about my doing A, since you've threatened me, and that you will be very upset if [do A
(rtot
that you will choose to be upset to punish me for doing A). Since I don't want you to be upset, I refrain from doing A. You did not coerce me into not doing A, though it seems that the conditions listed are satisfied; in particular it seems that the relevant conditions 5 and 2 are satisfied. Or, if they can be so interpreted so that they're not satisfied, it would be well to make this interpretation explicit, replacing 5, 1, and 2 by: (5') Part of Q's reasons for not doing A is to avoid (or lessen the likelihood of) the consequenceS which P has threatened to bring about or have brought about.6
(1') P threatens to bring about or have brought about some conse quence if Q does A (and knows he's threatening to do this). (2') A with this threatened consequence is rendered substantially less eligible as a course of conduct for Q than A was without the threatened consequence.
Copyrighted material 18 -& CHOICE AND UTILITY
Must P make the tl:ueat in order to get Q not to do A? Must condition 3 be satisfied? In normal situations it will be satis.fied; e.g., a highwayman says "If you don't give me your money, I'll kill you," making this threat in order to get me to give him my money. But suppose that we are conducting an experiment for the Social Science Research Council, to study people's reactions in the highwayman situation. We don't care how he reacts to our threat (if he gives over the money we must turn it over to the SSRC; if he resistS we are empowered to kill him and, let us suppose, have no moral scruples about doing so). We do not say "your money or your life" in order to get him ro give us h.is money, but in order to gather data. We might even suppose that I think him very brave and have bet with you that he'll resist and be killed. After making the bet, I want him not to hand over the money, and I don't make the threat in order to get him to hand it over. I.n the grip of fear and trembling, he hands over the money. Surely we coerced him into doing so. This suggests replacing 3 by a more complicated condition: (l'art of) P's reason for deciding to bring about the consequence or have it brought about, if Q does A, is that P believes this consequence worsens Q's alternative of doing A (i.e., that P believes
that this consequence worsens Q's alternative of doing A, or that Q
would believe it docs). 7 The SSRC example satisfies this condition, since (part of) the researchers' reason for decidi ng to kiU Q if he doeso't turn over the money is that they believe this consequence worsens Q's alternative of not giving them the money. But the condition formu lated is not broad enough, for we want to
cover cases where P bas not decided to bring about the consequence if Q does A, but is bluffing instead, or neither intends nor intends not ro bring it about if Q does A. This suggests disjoining another condition with the one above: lf P has not decided to bring about the consequence, or have it brought about, if Q does A, then (part oO P's reason for saying he will bring about the consequence, or have it brought about, if Q does A is that ( P believes) Q will bel.ieve this consequence worsens Q's alternative of doing A.s One is tempted ro say that this disjunctive condition is superfluous, because it is built into the notion of th;e.ateni ng, and hence follows from condition 1' . That is, one is tempted to say that if this condition is not
Coercioo ~ 19 satisfied, if P's reasons or motives are not as described, then P bas not threatened Q. I shall nave more to say about mis later. The conditions listed srill do nor appear to be sufficient. Consider cases where Q wants ro do A in order to bring about x, and P says that if Q does A, he (P) will do something which just prevents A from bringing about x. This makes A substantially less eligible as an alternative for Q (Q now, we may suppose, has no reason to do A), and the other conditions may well be satisfied. Yet, at least some cases of this sort ("If you say another word, l shall turn off my hearing aid") are not cases of coercion.9 Cases of this sort sugge.sr the following condition: (7) Q believes that, and P believes that Q believes that, P 's tnreatened consequence10 would leave Q worse off, having done A, than if Q didn't do A and P didn't bring about the consequence.11 In the application of this condition, in deciding how well or poorly off Q is having done A and having had his purpose x thwarted, one must ignore Q's wasted effort, humiliation at having failed to bring about x, and (in some cases) Q's forgone opportunities. Similarly in deciding how well or poorly off Q would be not doing A and not having P bring
about the consequence, one must ignore any regret Q might feel at nor doing A. 12 According to our account of "P coerces Q into doing A," the following c.ases are not cases in which P coerces Q inro doing A.
'
(1) Q mishears P as having said "Your money or your life" and hands over bis money, but P said something else, or said this as a question about something he thought Q said, ere. (2) P doesn't speak English, but has picked up the sentence "Your money or your life" from a movie, though he does not know what it means. To be friendly, P utters this one sentence to Q who is sitting next ro him in a bar (perhaps while showing Q his unusual knife for Q to admire). Q hands over his money. (3) Q walks inro a room, and unbeknownst ro hino there is a rape recorder in the next room playing part of the soundtrack of a movie. Q hears "Put all of your money on the table and then leave, or I'll kill you." Q purs his money on the table, and leaves. I suggest that in these case.s, though Q feels coerced and thinks he is coerced, P does not coerce Q into giving over the money. (In the third case there is no plausible person P to consider.) Those who refuse to
Copyrighted material 20 ~ Cl:iOICE AND UTILITY
accept this might bold the view that though P does not coerce Q into giving over the money, nonetheless Q is coerced into giving over the money. Such a person would reject the view that "Q is coerced into doing A" is equivalent to "there is a P who coerces Q into doing A," and perhaps suggest that Q is coerced into doing A if and only if (1) There is a P who coerces Q into doing A, or (2) Q is justified in believing that there is a P who has threatened to bring abour a consequence which significantly worsens his alternative of not doing A (and that P has the appropriate reasons and intentions), and (part of) Q's reasons for doing A is to avoid or lessen the likelihood of this consequence he believes was threatened. I should mention that a threat need not be verbally expressed; it
may be perfectly clear from actions performed what the threat is, or at least that something undesirable will occur if one aoesn't perform some appropriate action. For example, members of a street gang capture a member of a rival gang and ask him where rhat gang's weapons are hidden. He refuses to tell, and they beat him up. They ask again, he refuses again, they beat him again. And so on until he tells. He was coerced into telling. His captors didn't have to say, "if you don't tell us we will continue to beat you up or perhaps eventually do something worse." This is perfectly clear to aU involved in the situation. In many situations the inAiction of violence is well understood by all parties to be a threat of further infliction of violence if there is noncompliance. Nothing need be said. 13 It may be for reasons such as this that some writers (e.g., Bay) say that all infliction of violence constitutes coercion. But this is, I thin k, a mistake. If a drunken group comes upon a stranger and beats him up or even kills him, this ne.ed not be coercion. For there need have been no implicit threat of further violence if the person didn't comply with their wishes, and it would indeed be difficult for this to be the case if they just come upon him and kill him.14 There is another type of situation very similar to the one we have tbus far been concerned with, for which similar conditions can be offered. I have in mind cases where no one threatens to inflict some damage on Q if he does A, but someone sets things up so that damage is automatically inflicted if Q does A. It's not that if you do A, I will bring about a consequence which you consider to be bad, but rather that I now do something (the doing of which is not conditional upon your doing A)
Coercion ~ 21 which is such that if you do A after I have done this thing, there will
be a consequence which you consider tO be bad.15 Though in such situations a person is deterred from doing something, it is not obvious to me rhar he is coerced into nor doing it. If it is coercion then the account of coercion would say that P coerces Q into not doing A if and only if either of the two sets of conditions is satis6ed.16 I suggest that it is as a case of this sort of situation, rather than the one discussed earlier, that we are to understand the following: some adult's mother says to him, "If you do A I'll have a heart arrack, or the probability = p that I'll have a heart attack." I have in mind a case where the mother does not choose to have a heart attack if her son does A or to do something which will bring on or raise the probability of a bean attack. She just knows she will (or that the probability= p). It seems to me that the mother's statement is nor plausibly construed as a threat to (choose to) do something or bring about a consequence if her son does A. To use a distinction which will be discussed later, what the mother issues is not a thre.at but rather a nonthreatening warning. If we look just at the first sort of situation, we will conclude that the mother did not coerce the son into not doing A. But this example can plausibly be viewed as a case of the second sort of sitoation, in which before Q does A, P does something, making this known to Q, which worsens Q's alternative of doing A. And if this counts as coercion, then the mother may coerce the son. We should look, in this case, at the mother's act, prior to her son's doing A, of telling him that she will or probably will have a heart attack if he doe.s A. We may suppose that without her announcement the consequence of her son's doing A is some probability of her having a heart arrack and some probability of his feeling guilty (a function of the probability of his realizing why she died and the probability that he
will feel guilty anyway because he did something she didn't )jke and then she died) and some probability of A's having quite nice consequences. And we may suppose that after the mother's announcement the consequences of his doing A are changed significantly. For now there is some probability of her dying and his feeling enormously guilty (because he ignored her warning), and even if she won't die because of his doing A, if he does A he will worry over this possibility, feel guilty about doing something he knows upsets her, etc. Her act of making her announcement before he did A worsened the conse4uence of his doing A. If we suppose, furthermore, that one of her reasons for making the announcement was to worsen the consequences, and that one of his reasons for not doing A was this worsening of consequences, then we have a situation of the
second sort. And if this sort of situation counts as coercion, the son was coerced into nor doing A Y Nonceotral Cases of Coercion I have thus far concentrated upon the central part or core of the notion of coercion, and, in order to avoid too many complications all at once, have spoken of necessary and sufficient conditions for coercion (period). However, I believe that there are further cases of coercion which do not themselves satisfy the conditions thus far discussed, but which are cases of coercion by virrue of standing in certain specifiable relations to central cases of coercion.18 It is a task of some intricacy to get these relations just right. The statements which follow are meant to indicate areas in which principles must be formulated. It is 110t claimed that these
stat.ements are the formulations which one would eventually arrive at, nor is it claimed that the statements which follow exhaust the areas for which principles must be formulated. Let me repeat: the statements below are meant to indicate areas for which principles must be formulated and are not put forward as the correct formulation of principles in these areas. I would expect that, after sucb principles are adequately formulated, a re.cursive definition of "P coerces Q into doing A" would be offered, which would begin with the conditions for the central cases discussed earlier. (1) If P coerces Q into doing A, and "A" contains as a proper part the referring expression "r.," and " B" is obtainable by substituting the referring expressions "' r2" for "r1"' in "A," and "r2" and
"rt" have the same reference, and " r 1" occurs transparently in
"Q does A," then P coerces Q into doing B.19 (2) (f P coerces Q into doing A, and it is a necessary truth that if anyone does A he does 8, and it is not a necessary truth that if anyone does anything he does B, then P coerces Q inro doing B. (3) If P coerces Q into doing A, and it is a nomological truth that if anyone does A he does B, and it is not a nomological truth that if anyone docs anything he does B, then P coerces Q into doing B.
(4) If P coerces Q into doing A, and if the only way anyone can do A is by doing either B1 or B2 or, ... , or B., then P coerces Q into doing B 1 or B2 or, ... , B.,. 20
Copyrighted material Coercion ~ 23 (5) If P coerces Q into doing A, and the only way in which Q can do A is by doing either 8 1 or 82 or, .. . , or B., then P coerces Q into doing B1 or B2 or, ... , or 8 . 2 ' (6) If Q can do A only by doing B1 or B2 or, ... , or B., and Q sets our to do A (intending to do it) because of P's threat of a harmful
consequence if Q doesn't do A, and Q does one of the B1 in order
to do A, then Q is coerced into doing 81 or 8z or, ... , or B., even if Q does not do A (whether because be's prevented from doing A or because he's changed his mind about doing it). (7) If P coerces Q into doing 81 or Bz or, ... , or B., and 8 1 is the best of the 81's, the only one of the 8 1's it would be reasonable to do, etc., and Q does 8 1 for this reason, then P coerces Q into doing B1 (8) lf P coerces Q into doing A, and x is a consequence of Q's A, and-, then P coerces Q into bringing about x. (What further conditions are needed in the blank?)22 ln order to avoid concluding that Q was coerced into doing B when Q does A (partly) because of the threat, and does B (which stands in one of the stared relations to A) for some other reason, we must add to the antecedent of each of these statements the qua lification that (part of) the reason Q does B (or, B., or Bz, or ... , or B.) is to avoid or lessen the likelihood of P's threatened consequence (if Q does A). Threats and Offers The notion of a threat has played a central role in what has been said rhus far. In this section we shall consider the differences between threats and offers, and in the next section we shall consider the differences between threats and warnings. lf P offers Q substantially more money than Q is earning at his current job to come to work for P, and Q accepts because he wants to increase his income, has P coerced Q into working for him? Some writers (Hale, Bay) would say that P has; the threat being "come to work for me or I won't give you rhe money.'>23 On this view, every employer coerces his employees, every employee his employer ("give me the money or I won't work for you"), every seller of an object coerces his customers ("give me the money or I won't give you the object"), and
Copyrighted material 24 ~ CHOICE AND UTILITY
every customer the person from whom he buys. It seems clear that
normally these aren't cases of coercion. Offers of inducements, incentives, rewards, bribes, consideration, remuneration, recompense, payment do not normally constitute threats, and the person who accepts them is not normally coerced. As a first formulation, let us say that whether someone makes a threat against Q's doing an action or an offer to Q to do the action depends on how the consequence he says he will bring about changes the consequences of Q's action from what they would have been in the normal or natural or expected course of events. If it makes the consequences of Q's action worse than they would have been in the normal and expected course of events, it is a threat; if it makes the consequences bener, it is an offer.24 The term expected is meant to shift between or straddle predicted and morally reqired. 2.! This handles preny well the clear cases of threats and offers. Let us see how it fares with more difficult examples. (a) P is Q's usual supplier of drugs, and today when he comes to Q he says that he will not sell them to Q, as he normally does, for $20, but rather will give them to Q if and only if Q beats up a
certain person. (b) P is a stranger who has been observing Q, and knows that Q is
a drug addict. Both know that Q's usual supplier of drugs was arrested this morning and chat P had nothing to do with his arrest. P approaches Q and says that he will give Q drugs if and only if Q beats up a cerrain person. In rhe first case, where P is Q's usual supplier of drugs, P is threatening not to give Q the drugs. The normal course of events is one in which P supplies Q with drugs for money. P is threatening tO withhold the supply, to deprive Q of his drugs, if Q docs not beat up the person. In the second case, where P is a stranger to Q, P is not threatening not to supply Q with drugs; in the normal course of events P does not do so, nor is P expected to do so. If P does not give Q the drugs he is not
withholding drugs from Q nor is he depriving Q of drugs. P is o{{eri11g Q drugs as an inducement to beat up the person. Thus in the second case, P does not coerce Q into beating up the person, since P does not threaten Q. (But the fact that P did not coerce Q into beating up the person does not mean that it would not be true for Q to say, in some legitimate sense of the phrase: "I had no choice.")
Copyrighted material Coercion ~ 25
There is a further point to be considered about the fust case in which P is Q's usual supplier of drugs. In addition ro threatening to withhold the drugs if Q doesn't beat up a certain person, hasn't P made Q an offer? Since in the normal and expected course of events Q does not get drugs for beating up the person, isn't this a case in which P then offers Q drugs as an incentive to beat up the person? And if P has made this offer, why do we view the overall situation as one in wh.ich P threatens Q, rather tha.n as one in which P makes Q an offer? We have here a situation in which P takes a consequence viewed as desirable by Q (receiving drugs) off one action (paying $20) and puts it onto another action (beating up the person). Since Q prefers and P believes that Q prefers paying the money and receiving the drugs to beating up the person and receiving the drugs, and since Q would rather not beat up the person, P's statement is a threat to withhold the drugs if Q doesn't beat up the person, and this threat predominates over any subsidiary offer P makes for Q to beat up the person, making the whole situation a threat situation. But instead of subtxacting a desirable consequence from one of Q's actions and tagging the same consequence onto another of Q's actions, P may subuacr a desirable consequence C from one of Q's actions At and add a more desirable consequence C' onto another action A2 available to Q. For example, the dope peddler might say to Q, "I will not give you drugs if you just pay me money, but I will give you a better grade of drugs, without monetary payment, if you beat up this person." It seems plausible to think that as one increases the
desirability of C' to Q, at some point the situation changes from one predominantly involving a threat to deprive Q of C if he does A1 (doesn't do A2l to one which predominantly involves an offer to Q of C' if he does A2 And it seems plausible to claim that this turning point from threat to offer, as one increases the value of C' for Q, comes at the point where Q begins preferring A2 and C' to At and C (stops preferring the latter to the former?).26 The following principle embodies this claim, and also covers the case where it is the same consequence which is switched from one action to another, as in the earlier example. It also is meant to apply to obvious mixtures of threats and offers, e.g., "If you go to the movies I'll give you $10,000.1f you don't go, I'll kill you." If P intentionally changes tbe consequences of two actions A 1 and A2 available to Q so as to lessen the desirability of the consequences
Copyrighted material 26 -b CHOICE AND UTILITY of A I> and so as to increase the desirability of the consequences of A2, and part of P's reason for acting as he does is to so lessen and
increase the desirabilities of the respective consequences, then (a) This resultant change predominantly involves a threat to Q if he does A 1 if Q prefers doing the old A 1 (without the worsened consequences) to doing the new A2 (with the improved consequences). (b) This resultant change predominantly involves an offer to Q to do A2 if Q prefers doing the new A2 (with the improved consequences) ro the old A1 (without the worsened consequences). This principle ties in nicely with something we shall say later. For when the change predominantly involves a threat, Q would normally not be will ing to have this change made (since he'd rather do the old A1 than either of the two alternatives after the change), whereas when the change predominantly involves an offer, Q would normally be willing to have the change made (since he'd rather do the new A2 than the old
A I> and if he prefers doing one of the old alternatives (A 1) to doing
the new A2, he can still do it). I shall claim later that this willingness or unwillingness to make the change marks an important difference berween offers and threats.27 If a statement's being a threat or an offer depends upon bow the carrying out of the statement affects the normal or expected course of events, one would expect that there will be situations where it is unclear whether a person is making a threat or an offer because it is unclear what the normal and expected course of events is. And one would expect that people will disagree about whether something is a threat or an offer because they disagree about what the normal and expected course of events is, which is to be used as a base(jnc in assessing whether something is a threat or an offer. This is indeed the case. Consider the following example. Q is in the water far from shore, nearing the end of his energy, and P comes close by in his boat. Botb know there is no other hope of Q's rescue around, and P knows that Q is the soul of honesty and that if Q makes a promise he will keep it. P says to Q "I will take you in my boat and bring you to shore if and only if you first promise to pay me $10,000 within three days
Copyrighted material Coerc.ion .-& 2 7
of reaching shore with my aid." Is P offering to rake Q to shore if he makes the promise, or is he threatening to let Q drown if Q doesn't make the promise? If one views the normal or expected course of events as one in which Q drowns without P's intervention, then in saying that he will save Q if and only if Q makes the promise, P is offering to save Q. If one views the normal or expected course of eventS as one in which a person in a boat wbo comes by a drowning person, in a situation such as this, save.s him, then in saying that he will save Q if and only if Q makes the promise, P is threatening not to save Q. Whether P's saying
that he will save Q if and only if Q makes the promise is an offer to save Q or a threat not to save Q depends upon what the normal or expected course of events is. Since it is likely to he clear to the reader which course of eventS he wanes to pick out as normal and expected as the background against which to assess whether P's statement is an offer or a threat (namely, the one that makes it a threat) we should sharpen the example. Suppose in addition ro tbe foregoing that P knows that Q has greatly wronged P (or others), bur that Q cannot be legally punished for this (no law covered the wrong, a legal techn ical ity, the statute of limitations has run out, or some such thing). Or P knows that Q will go on to do monstrous deeds if rescued. In some such situations it will be unclear what P is morally expected to do, and hence uncle.ar whether his statement is a threat or an offer. For other such situations it will be dear that P is morally expected ro let Q drown, and hence his statement will be an offer. 28 Thus far we have considered threats as introducing certain deviations from the normal and expected course of events. The question arises as to whether the normal or expecred course of events itself can be coercive. Suppose that usually a slave owner heat.s his slave each morning, for no reason connected with the slave's behavior. Today he says tO his slave, "Tomorrow I will not beat you if and only if you now do A." One is tempted to view this as a threat, and one is also tempted to view this as an offeL I attribute these conflicting temptations to the divergence between the normal course of events, in which the slave is beaten each morning, and the (morally) expected course of events, in which be is not. And I suggest that we have here a situation of a threat, and that here the morally expected course of evenrs takes precedence over the
normal course of events in assessing whether we have a tbreat or an

offer.2 ~
Copyrighted material 28 -f> CHOICE AND UTILITY
One might think that in deciding whether something is a threat or an offer, the (morally) expected course of events always takes precedence over the normal or usual course of events, where these diverge. It is not obvious that this is so. I have in mind particular!)' the example mentioned earlier, where your normal supplier of dope says that he will continue to supply you if and only if you beat up a certain person. Here, let us suppose, the morally expected course of events is that he doesn't supply you with drugs, but the course of events which forms the background for deciding whether he has threatened you or made you an offer is the normal though not morally expected course of events (in which he supplies you with drugs for money); it is against this background tbat we can obtain the consequence that be's threatened you. Thus, in both the slave and the addict examples the normal and morally expected courses of events diverge. Why do we pick one of these in one case, and the other in the other, as the background against which to assess whether we have a threat or an offer? The relevant difference between these cases seems to be that the slave himself would prefer the morally expected tO the normal course of events whereas the addict prefers the normal to the morally expected course of events. 30 It may be that when the normal and morally expected courses of events diverge, the one of these which is to be used in deciding whether a conditional announcement of an action constitutes a threat or an offer is the course of events that the recipient of the action prefers.l 1 I have raised the question of whether the normal and expected course of events itself can be coercive, and was led to consider cases where the normal and (morally) expected courses of events diverged. I now would like to consider this question again, for cases where they do not
diverge. Cao P, by saying that he will bring about a consequence if Q does A (where this consequence is such that if Q does A, P would bring it about in the normal a11d (morally) expected course of events), coerce Q into not doing A? Suppose that in the normal a11d morally expected course of events, people get punished for theft. A.ren 't some people coerced into not st.ealing by the legal apparatus? One might say that if a type of action or consequence is itself parr of the normal and expected course of events if Q does A, one should use the normal and expected course of events minus this type of action or consequence as a background against whicb to assess whether a statement is a threat. If the consequences of an action would be worse, if the statemenr is carried our, than they would be in this new course of events (i.e., the normal and e.xpecred course of events without tbe
Copyrighted material Coercion -& 29
type of action or consequence) then the statement is a threat. But who knows what the world would be like if there was no punishment for crimes? It might well be that things would be so bad that the institution of punishing crimes would improve the consequences of almost all actions, and hence count, according co this suggestion, as making offers to people. An alternative procedure seems more reasonable; namely, to consider the normal and expected course of events, if Q does A, without P's particular act or without the particular consequence P will bring about, and against this background assess whether P's statement that if Q doe-s A he will do a particular act or bring about a particular consequence constitutes a threat (i.e., whether P's statement, if carried out., makes Q's A worse than it would be in this new course of events). There remain some problems about knowing what the course of events would be without this act, but these seem manageable. On this
view, even though in the normal and expected course of events Q gets punished for theft, the statement that he will be punished for theft counts as a threat since the act of punishment, if Q steals, unfavorably affects the consequences of one act of Q's (stealing) against the background of the normal and expected course of events minus this act of punishment. 32 According to the account offered earlier of (the first sort of) coercion, threats are necessary for coercion. One might extend the account to include some offers, if there were dear situations in which Q is coerced into doin.g A even though Q does A because P offered to do 8 if Q did A. Despite my inclination to say that one is never coerced when one does something because of an offer (unless in rhe case discussed earlier, the slave owner is making an offer to his slave), there is one sort of case, where the offer is closely tied to coercion or attempted coercion, which I find it difficult co decide about. Suppose that P knows that Q has committed a murder which the police are investigating, and knows of evidence sufficient to convict Q of chis murdeL P says to Q, "If you give me $10,000 I will nor turn over the information I have to the police." Let us assume that were P unable co contact Q and present his proposal he would turn the informacion over to the police. Furthermore, in chis situation P is (morallj) expected to turn the information over to the police. So in the normal and expected course of events, P turns the information over to the police (whether or nor Q gives him SlO,OOO). It would seem, therefore, that P is offering not co turn the information over to the police, rather chan threatening co turn it over. Yet one is
strongly tempted to say, when Q pays P $10,000 because he accepts the
offer, that Q was coerced by P into paying the $10,000.JJ U the following principle were correct, then this would be a case of
coerc1on:
U P offers to refrain from aiding the threatener of a coercive consequcnce34 for Q's A from bringing about this consequence, in exchange for Q's doing 8, and if the credible threat of this consequencc3; if Q didn't perform 8 would coerce Q into doing 8, then when Q does 8 because of the offer, Q was coerced into doing B. A case similar to the previous one is one in which the police arrest Q for a crime, believing that he has committed it and having sufficient evidence to convict him of it. ln the course of questioning Q, they come co believe that Q knows who has committed some other crime, and they say that Q will not be prosecuted if and only if he tells them who has committed this other crime. Since if the police did not think Q knew who committed the other crime they would have prosecuted, and since they are morally expected to have him prosecuted, the police. have offered not to have Q prosecuted rather than threatened to have bim prosecuted. If Q names the person who commitud the other crime, in order tO escape being prosecuted, some are strongly tempted to say that he was coerced i.nto giving the information. The above principle would yield this consequence. Though I do not deny that one may say, in some legitimate sense of these expressions, "Q was forced to do what he did," "Q had no choice," I am unable to decide whether, in the above cases, Q was coerced into doing so, and I leave this an open question. The two previous cases are cases where P is morally expected or required to do the act, and would normally do so (turn the murderer over to the police, have Q prosecuted ). It is worth mentioning cases where P has a lega l and moral right to do the act, but would not decide to do so (even if Q didn't do A) were he not trying to get Q to do A. For example, P has a righr to build on his land blocking Q's view, or foreclose Q's mortgage, or bring legal action against Q (on a valid and enforceable claim) but would not decide to do so {it's not worth the trouble, P has no pressing need for funds, etc.) were it not for his
wanting Q to do A. P tells Q that unless he does A, be (P) will build on his land, foreclose Q's mongage, bring legal action against Q, etc. Since P's action is not part of the morally expected course of events (not that P is morally expected not to do it) and since in the normal course
Coercion ~ 31 of events P wouldn't do it, the account yields the result thar one would wish: In the.se cases P is threatening to perform his actions rather than offering not to do so. Threats. and Warnings In the section on Conditions for Coercion, because of the example of the SSRC people saying "your money or your life," we rejected the con d.ition that P says or does what he does in order to get Q not to perform some particular action A. We substituted instead a condition requiring that (part of) P's reason for deciding to bring about the consequence if Q does A (or, if he hasn't decided, saying that he has) is that (P believes that) this worsens Q's alternative of doing A (or that Q would believe it does). And we mentioned the view that this is part of the notion of making a threat. This view illuminates the fact that some statements about one's future actions if Q does A, are not threatS even though the acts you've stated you would do if Q does A worsen Q's alternative of doing
A.
Such statementS about one's future actions I shall call nonthreatening warnings (for short, just "warnings"). The distinction between threats and nonthre.atening warnings is crucial to some questions that arise in the law.36 For example, an election is about to be held in a factory to determine whether the employees wi ll be represented by a labor union. The owner of the factory announces to his employees that if the union
wins the election, he will close his factory and go out of business. Has he threatened the employees with the loss of their jobs if the union wins, or merely warned them what will happen if the union wins? If a majority of the employees would have voted for the union if not for the announcement, and the union lost because of the announcement, were the employees coerced by their employer into rejecting the union? We may view this situation as a game, represented by the following matrix: A. Stays in business
EMPLOYER
B. Goes out of business

EMPl-OYEES
I. Union wins II. Union loses (b) (a)

(c) (d)
The employees make the first move (pick a column) and the employer makes the second (picks a row) knowing what move the employees have first made. That is, first the employees choose to be represented by the union or not, and then the employer, knowing what choice his employees have made, decides to stay in business or go out of business. I shall assume that each of the members of some particular majoriry of the employees preferentially ranks the outcomes as follows:
(b) (a) (c) (d)
and shall call th.is the preference ordering of the employees.37 There seem to me to be at least four cases worth considering,38 corresponding to four different preference rankings of these alternatives by the employer. These are:
(I) (a)
(b) (c) (d) (2) (a)
(c)-(b)
(d)
(3) (a)
(c)
(b)
(4) (c)-(a) (b)
I assume, for each case, that the employer knows the preference ranking of the employees. Case 1: The employer, if he were sure that the union would win the election, would not make his announcement, since he prefers continuing in business though the factory is unionized, to going out of business. That is, he prefers (b) to (c). However, he commits himself to going out of business if the factory is unionized, and announces this decision (that is, rules out (b)), in the hope that this will lead his employees to reject the union. By ruling out (b) beforehand he leaves his employees (a), (c), and (d) among which to choose, and since of these (a) is highest in the preference ranking of the employees, they will presumably then act so as to realize (a), and this is the alternative which the employer ra.nks most highly. The employer is committing himself beforehand, for strategic reasons, to do something (B) in a situation (f) such that had he not committed himself to this he would be better off doing something else (A) in that situation when and if it arose. It seems clear that in this case, when the employer announces that he will go out of business if the union wins the election he is threatening his employees.l9 For in
the normal course of events, he does not go out of business, given his
preferences, if the employees vote for the union. Hence he's announcing thar he will deparr from the normal course of events in a way to the detriment of his employees, if they elecr the union.40 And also (parr of) b.is reason for deciding to go out of business if the union wins is that he believes rhar this worsens his employees' alrernarive of electing rhe union (and he thereby hopes ro influence them to reject the union). Hence his announcement is a threar., and he has rhrearened to go our of business if the union wins rhe election. Case 2: The employer announces thar he will go out of business if rhe union wins the election, thus ruling out (b), and leaving the employees a choice between (a) and (c). He makes this announcement hoping that because of it, his employees will reject the union. Note that in this case, unlike the first one, there is not something the employer would rather do than go out of business if the union wins the election. (For he is indifferent between (c) and (b), whereas in the firsr case, the employer preferred (b) to (c) .) Though there aren 't exactly the same straregic considerations as in rhe first case, still strategic considerations are involved. And this employer, too, has threatened his employees. For in the normal or expected course of events be wouldn't have made his announcement, and would, being indifferent between (c) and (b), decide whether or not to stay in business if and when the union won the election. We may suppose that before he decides there is some non.zero probability of his deciding to dose the factory, and some nonzero probability of his staying in business. This being what one would expect in the normal course
of events, his announcement and commitment to going out of business, for sure, if the union wins the election changes the normal or expected course of events in the typical manner of a threat. For his being almost certain to go out of business is worse, from his employee.s' viewpoint, than there being some probability of his going out of business and some probability of his remaining in business. Since part of his reason for deciding to go out of business if the union wins is that (he believes) this worsens his employees' alternative of electing the union, this constitutes a threat. But if in this case the employe.r does nor announce that he will close if the union wins (which would he a threat), bur instead announces truthfully that he is indifferent between closing and staying in business if the union wins and will decide what co do afterwards (and if his employees voting for the union in the face of this announcement would not anger him and yield a larger probability of his closing than if the
Copyrighted material 34 _,. CHOICE AND UTILITY
announcement hadn't been made), then the case may be significantly different. This issue is also raised by case 3, and is treated in the discussion of it. Case 4: The employer tells his employees what his preferences are, and that he will go out of business if the union wins. We may suppose that he doesn't make his decision or teU them this in order to get them to reject the union, since he doesn't care whether they reject it or not. For even if they elect it, he can go out of business, and there is nothing he prefers ro that. No strategic considerations are involved. He makes his announcement solely to inform his employees of what will be the consequences of their action. It was no part of his reason for deciding to go out of business if the u.nion won, that this worsened his employees'
alternative of electing the union, and no part of his reason for making the announcement was to get his employees to reject the union. It seems clear that this is not a case of coercion, and that the employer bas not made a threat (even though his employees might sorrowfully tell the union representative that they bad no choice) but has rather issued a nonthreatening warning. Case 3: I have left case 3 for last because it is the most difficult. In this case, the employer announces what his preference ranking is and says that he' ll go out of business if the union wins the election. And indeed, unlike the employers in cases 1 and 2, but like the employer in case 4, he prefers going out of business if the union wins the election to staying in business with the union representing his employees. In the normal course of events, he would go out of business if the union wins, whether or not he has previously announced that he would do so. However, unlike the employer in case 4 but like the employers in cases l and 2, he prefers staying in business without the union to going our of business if the union wins the election, and makes his announcement in order to get his employees to reject the noion. Has he threa tened his employees, or just warned them? I am inclined to say that he has warned them rather than threatened them. (Note that a teacher can warn a student that he will fail unless his work improves, even if the teacher does so in order to get the student to work harder.) For he does not decide to close if the union wins in order tO worsen the alternative of the union winning, and in making the announcement he does not worsen this alternative but rather makes known what its consequences will be. Furthermore, there seems to be no presumption against this employer's telling his employees that be wiU close if the union wins, whereas there is (normally) a presumption against making threars.41
Coercion -& 35 These cases raise an interesting point relevant to the wider task of deciding what actions people should be free to do, and what actions they should be unfree to do. It may be that, even if one picks out a particular pattern of freedom and unfreedom as optimal, there is no acceptable institutional arrangement available to one which realizes this pattern. One may lack the institutional means to realize exactly the optimal pattern of freedom and unfreedom, or certain wars of realizing this patrem, and publicly distinguishing among persons, may be unacceptable to the society at large. So it may be that, given the available institutional means, the feasible patterns of freedom and unfreedom among which one must choose are suboptimal and nondominated patterns of the following forms: (a) Where some persons are unfree to do acts they should (according to the optimal pattern) be free to do. (b) Where some people are free to do acts they should be unfree to do. (c) Where some people are free to do acts they should be unfree to do, and some people are unfree to do acts they should be free to do. A full theory of freedom would, as well as specifying the optimal pattern of freedom and un.free.dorn, concern itself with choices among such suboptimal patterns. It may be that we have an example of such a choice here. For one may wish to make the case 1 and 2 employers unfree to make their announcements (threats) while leaving the case 3 and 4 employers free to make their announcements (warnings). However, it may be difficult to devise an institutional arrangement which accomplishes this, for it may be difficult to distinguish case 1 and 2 employers from case 3 and 4 employers. (Note that if one attemptS to distinguish them, there will be reason for case 1 and 2 employers to lie when asked about their
preferences.) The actual institutional choice one faces may be to forbid all such announcements, to allow aU such announcements, or to use some condition almost coextensive with the suitable preferences and forbid and allow announcements on the basis of whether this condition is satisfied, regretfully admitting that one cannot handle all the cases in the way one would wish.42 Problems of choice among suboptimal patterns arise also and obviously for paternalistic legislation; legislation wh.ich, in order to prevent
him from coming to harm, or to lessen the chance of this, or to enable him to realize some good, makes someone unfree to perform a particular act .43 The feasible patterns wiU often require either making some persons, who do not need the paternalistic protection and might be better off without it, unfree to perform some acts, or leaving some people free to do acts they should be unfree to do (for their own protection). (I spare the reader examples.) One wiU often have to choose among such patterns because there is no realizable and acceptable instirutional arrangement which divides people up just right (according to the optimal pattern) with respect to particular ac.ts. An important point emerges from this discussion; namely, that the stateme.nt that a particular piece of legislation makes some persons unfree to perform some acts they should not be unfree to perform (according to the optimal pattern of freedom and unfreedom), even if true, is not by itself a conclusive objection to the legislation. For it may be that no feasible and acceptable pattern of freedom and unfreedom is more optimal (or all other realiz.able patterns are less optimal) than the one yielded by this legislation. I might mention the statement, which we might call a tip, which stands to an offer as a nonthreatening warning stands to a threat.; that
is, P's statement which points out that P will bring about some consequence if Q does A which improves Q's alternative of doing A, though P's believing that it improves or that Q would believe it improves Q's doing A is not parr of P's reason for deciding to bring about the consequence if Q does A. Building into the notion of P's making Q an offer to do A the requirement that (part of) P's reason for bringing about the consequence if Q does A is to improve Q's alternative of doing A, illuminates the following example: P comes up to Q and says, "your money or your life." Q resists and beats P up. Had P not said what he did and had he nor confronted Q with a gun, and had Q just beaten P up, Q would have been a bully and people would have scorned him. Bur now the consequences of beating Pup are made far more attractive to Q; Q becomes a hero if he does so. But P hasn't made Q an offer tO beat him up, because though what P did improved the consequences of some action of Q's, bringing about this improvement was not part of P's reason for acring as he did.44 Has P tipped Q off ro something? Highwaymen normally don't go on to say, "And if you resist and beat me up you'll be a hero," but if one did, be would have given his prospective victim a tip, however little appreciated. Note, incidentally, that one should not think for the case where Q beats P up after P makes his threat that Q was coerced i.nto
Copyrighted material Coercion -& 3 7
doing so, even though Q did so to avoid the threatened consequence. For not all of the other necessary conditions for coercion are satisfied. Threats, Offers, and Choices I have claimed that normally a person is not coerced into performing an action if he performs it because someone has offered him something
to do it, though normally he is coerced into performing an action i.f he does so because of a threat that has been made against his not doing so. Writers who count offers as coercive do so, I suspect,, because they accept something roughly like the following statement: If Q has available to him the actions in a set A, and as a result of what P has done Ot will do (a) act A1 is significantly higher in utility to Q than the other
actions in A (b) act Az is significantly lower in utility to Q than the other
actions in A whereas it wasn't before, and Q (a) does A1 because of this (b) refrains from A2 because of this then Q was coerced into (a) doing A 1 (b) not doing Az According to this view, any action of P's which results in A 1 's being significantly greater or Az's being significantly le,ss in utility than the other actions in A may coerce Q. It makes no difference, according to this view, how the difference in urility is brought about; whether in (a) A 1 is absolutely raised in utility or all the other members of A are absolutely lowered in utility, or whether in (b) Az is absolutely lowered in utility, or all the other members of A are absolutely raised in utility. It is only the resulting relative positions, however arrived at, which count. This view is mistaken, and I shall assume that we can all think of examples that show to our satisfaction that this is so. I now want to consider whether anything illuminating can be said about why the notion of coercion isn't so wide as to encompass all bringing about
of actions by the bringing about of difference in relative position. The question f'm asking may seem bogus. After all, there will be some terms which apply both to getting someone to do something via threats, and
to getting someone ro do something via offers, e.g., "getting someone to do something." And there will be some terms which apply to one and not the other. Am I just asking why the word coercion is among those that apply to only one of these and not to both? And why expect the answer, presumably going back to the word's Latin roots, to be philosophically interesting? So let me state my task differently. l would like to make sense of the following claims: when a person does something because of threats, the will of another is operating or predominant, whereas when he does something because of offers this is not so; a person who does something because of threats is subject to the will of another, whereas a person who acts because of an offer is not; a person who does something because of threats does not perform a fully voluntary action, whereas this is normally not the case with someone who does something because of offers; when someone does something because of offers it is his own choice, whereas when he does something because of tl:uears it is not his own choice bur someone else's, or not fully his own choice, or someone else has made his choice for him; when a person does something because of threats he does it unwillingly, whereas this is normally not the case when someone does something because of offers. (There arc other ways to approach this area. One might ask why we say that we accept offers, but we go along with threats rather than accept them.) I would like to make sense of these claims in the face of the following three roughly true statements, which seem to indicate that threat and offer situations are on a par so far as whose will operates, whether the act is fully voluntary, whose choice it is, and so forth. (I ) A person can be gotten to do something which someone else
wants him to do, which he otherwise wouldn't do, by offers as well as by th.reats. (2) A person can choose to do what there is a threat against his doing, just as a person can choose to do what there is an offer for him not to do. ("Just as"?) (3) Sometimes a threat is so great that a person cannot reasonably be expecte.d not to go along with it, but also sometimes an offer is
so great that a person cannot reasonably be expected not to go along with it (that is, not 10 accept it). I shall consider only a partially described person, whom I shall call the Rational Man, and unfortunately shall not get to us. The Rational Mao, being able to resist those temptations which he thinks be should resist, will normally welcome credible offcrs,45 or at any rate not be unwilling to have them be made. For he can always decline to accept the offer, and in this case he is no worse off than he would have been had the offer not been made. (Here I ignore the "costs" of making decisions, e.g., time spent in considering an offer.) Why sho~<ld be be unwiiJing to be the recipient of an offer? On the other hand, the Rational Man will normally not welcome credible th.reats, will normally be unwilling to be threatened, even if he is able to resist going along with them. It is worth mentioning some cases which are or seem to be exceptions to this. A person might not mind threats if he was going to do the act anyway. But, since in this case he (probably) wasn't coerced, he needn't concern us here. A person might welcome threats which restrict the acts he can reasonably be expected tO perform, and therefore improve his bargaining position with a third party, e.g., an employer negotiating with a labor union might welcome publicly known threats against raising wages by
more than n percent. But what he welcomes is not his being coerced into not raising wages by more than n percent (he is not coerced), but its looking to others as though he is coerced. This needn't concern us here. But there are other case.s which are somewhat more difficult. For example, P tells Q that he'll give Q $10,000 if in the next week someone, without prompting, threatens Q. Someone does and Q welcomes the threat. Or, P is jealous of Q's receiving certain sorts of offers and tells Q that if he ( Q) receives another offer before P does, then P will kill Q. Q cringes when the next offer comes. Or Q believes that having at least five threats (offers) made to one in one week brings good (bad) luck, and so is happy (unhappy) at tbe coming of the fifth threat (offer) in a week. Or for tax purposes Q welcomes a threat to illegally take some of his money. And so forth. I want to say that in such cases when threats are welcomed and offers shunned, they are done so for extraneous reasons, because of the special context. (One is tempted to say that in these conte~xts what would normally be a threat (offer) isn't really one.) I find it difficult to distinguish these special contexts from the others, but the claim that threats are normally unwelcome, whereas offers
are not, is not meant to apply to contexts where some special if-then is believed to obtain, where it is believed to be the case that if a threat (offer) is made, resisted (accepted), or carried out then something good (bad) will happen to the recipient of the threat (offer) (where this good (bad) consequence is not "internal" to the threat (offer)), and this belief on the part of the recipient of the threat (offer) overrides other considerat. ions. It is along such lines tbat I suggest viewing the person who welcomes a threat because it affords him the opportunity to prove to others or rest for himself his courage. Finally, let me mention the case where a person is in an ll-person prisoners' dilemma situation. In this case, he may most prefer everyone else's being coerced into
performing a dominated action while he is left free to perform the dominant action. He may also prefer everyone's being coerced into performing a specific dominated action (e.g., paying taxes) to no one's being so coerced. And since he realizes that the policy he most prefers, which treats him specially, isn't a feasible alternative, he may welcome the rhreat to everyone including himself.46 But though he welcomes the system which threatens eveqooe, he might still be coerced into performing his particular action (e.g., paying his taxes). This too seems to me not to be a counterexample to the claim that threats are unwelcome, hut one of the special contexts, with special if-rhens tagged onto tbe making of the threat, to which the claim is not meant to apply. I have said that the Rational Man would normally be willing tO have credible offers made to him, whereas he would not normally be willing to be the recipient of credible threats. Imagine that the Rational Man is given a choice about whether someone else makes him an offer (threatens him). For example, the Rational Man is asked, " Shall I threaten you (make you an offer)?" If be answers "yes," it is done. I am supposing that no offer is made the Rational Man to say "yes" to this question, and no threat is made against bis not saying "yes"; i.e., that no threats of offers arc involved in this choice about whether a threat (offer) is to be made. Let us call the situations before a threat or offer is made the presituarion. (I shall speak of the prethreat and preoffer situations, in anticipation of what is to come.) And let us call the situations after a rhrear or offer is made the threat and the offer situations, respectively. Looking first at offers: (a) The Rational Man is normally willing to go and would be willing to choose to go from the preoffer to rhe offer situation.
Copyrighted material Coercion -& 41
(b) In the preoffer situation, the Rational Man is normally willing to do A if placed in the offer situation. (c) The Rational Man, in the preof.fer situation, is unwilling to do A. (We're concerned with the case where he does A [partly) because of the offer.) (d) The Rational Man, when placed in the offer situation, does not normally prefer being back in the preoffer situation. Turning to threats: (a) The Rational Man is normally unwilling to go and unwilling to choose to go from the prethreat situation to the threat situadon. (b) In the prethreat situation, the Rational Man is normally willing to do A if placed in the threat situa6on. (c) The Rational Man, in the prethreat situation, is unwilling to do A, and would not choose to do it. (We're concerned with the case where he does the act [partly) because of the threat.) (d) The Rational Man, when placed in the threat situation, would normally prefer being back in the prethre.at situation, and would choose to move back The rwo significant differences between these rwo lists are: (I) The Rational Man would be willing to move and to choose to move from the preoffer to the offer situation, whereas he would normally not be willing to move or to choose to move from the prethre.at situation to the threat situation. (2) The Rational Man, once in the offer situation, would not prefer being back in the preoffer situation, whereas the Rational Man in the threat situation would normally prefer being back in the
prethreat situation.
If we concentrate solely on the choices made in the threat and offer

situations, we shall be hard put to find a difference berween these situations, which seems to make a difference as to whose will is operating, whose choice it is, whether the act is fully voluntary, done willingly or unwillingly, and so forth. If, however, we widen our focus and look not only at the choices made in the postsituations, but look also at the
choice that would be made about moving from the presituation to the
postsituarion, then things look more promising. For now we face not just two choices but two pairs of choices: (I) To move from the preoffer to the offer situation, and ro do A in rhe offer siruation (2) To move from the pretbreat to the threat siruation, and to do A in the threat situation. And the Rational Man would be (willing to) make both choices in (1 ), whereas he would not make both choices in (2). This difference in what choices are or would be made (when other factors are appropriately the same) seems ro me to make the difference, when someone else intentionally moves you from the presiruation to the postsiruation, to whose choice it is, whose will operates, whether the act is willingly or unwillingly done, and to whether or not the act is fully voluntary. One would like to formulate a principle that is built upon the preceding considerations, bur I find it difficult to formulate one that I am con.fident is nor open to very simple counter-example.s. Very hesitantly and tentatively, I suggest the foUowing plausible-looking principle:
If the ah:ernatives among which Q must choose are intentionally
changed by P, and P made this change in order to get Q to do A, and before the change Q would nor have chosen (and would have been unwilling to choose) to have the change made (and after it's made, Q would prefer that it hadn't been made), and before the change was made Q wouldn't have chosen to do A, and after the change is made Q does A, then Q's choice to do A is not fully his
own.
Notice that I have not said that the feature I am emphasizing which is mentioned in the principle, namely, being willing to choose to move from one situation to another, is by itself sufficient for a choice in the latter situation to be not fully one's own, but instead I have said that this feature, in conjunction with the other features listed in the antecedent of the principle, is sufficient.
Since this principle presenrs a sufficiem condition for Q's choice not being fully his own, it does not yield the consequence that in the offer situation., normally Q's choice is fully his own. A detailed discussion of when c.hoices are fully one's own, or fully voluntary, yielding rbis consequence, would take us far afield. Here I just wish to suggest that the crucial difference between acting because of an offer and acting bec.ause Copyrighted material Coercion -& 43 of a threat vis-a-vis whose choice it is, etc., is that in one case (the offer case) the Rational Man is normally willing to move or he moved from the presituation to the situation itself, whereas in the other case (the threat case) he is not. Put baldly and too simp!)', the Rational Man would normally (be willing to) choose to make the choice among the alreroatives facing him in the offer situation, whereas normally he would not (he willing to) choose to make the choice among the alternatives facing him ill the threat situation. The principle seems to me to be on the right track in concentrating not just on the choice of whether or not to do A, but also on the choice to move into the threat or offer situation. But it is difficult to state a principle which gets all the details right and which is not trivial and unilluminating (as one would he which said: if P moves Q from St to S2 via threats then ... ). It seems that rather than speaking (just) of act A being fully one's own choice, one should speak of its being fully one's own choice to do A rather than B. I have in mind the following sort of case. P intentionally breaks Q's leg (intentionally moving him from S1 [no broken leg] to S2 [broken leg]). Q would prefer not making this move, and afterwards would prefer not having made it. But once Q has a broken leg, he chooses ro have a decorated cast put on it, rather than a plain white one. If we just look at the act of wearing a decorated cast, we will have difficulties, for surely it is not Q's own choice (he was forced into a position where he had to wear a cast,
etc.), yet in some sense it is. It seems to me more illuminating to say that wearing a cast rather than none was not fully Q's own choice, wearing a decorated cast rather than a plain one was fully Q's own choice, and wearing a decorated cast rather than none was not fully Q's own choice. lt is not clear how to state a principle which takes this and similar complications into account, and is not open to obvious difficulties. I do, however, want to suggest that we shall not be able to understand why acts done because of threats are not normally fully voluntary, fully one's own choice, etc., whereas this is not normaUy the case with acts done in response to offers, if we attend only to the choice confronting the person in the threat and offer situations. We must look also at the (hypothetical) choice of getting (and willingness to get) into the threat and offer situations themselves. We have said that if P coerces Q into not doing A then (part of) Q's reason for not doing A is to avoid or lessen the likelihood of P's threatened consequence. Assuming that all of the conditions in the first section of this paper are satisfied, then
Copyrighted material 44 -<> CHOICB AND UTILITY
(a) In the case where Q's whole reason for not doing A is to avoid or lessen the likelihood of P's threatened consequence (ignoring his reasons for wanting to avoid this consequence), P coerces Q into nor doing AY (b) In the case where P's threatened consequence is not part of Q's reason for not doing A (even if it is a reason Q has for not doing A) then P docs not coerce Q into not doing A. But the case is more difficult when P's threatened consequence is part of Q's reason for not doing A, and other reasons which Q has for not doing A (which do not involve threats) are also part of his reason for not doing A. for in this case, Q conrributes reasons of his own; it is not solely because of the threat that he refrains from doing A. If we had to say either that this situation was one of coercion or was not one of
coercion we would, I think, term it coercion.48 But, I think, for such cases one is inclined to want to switch from a classificatory notion of coercion to a quantitative one.49 Let me indulge in a bit of science fiction. Suppose that one were able to assign weightS to the parts of Q's reasons for not doing A, which indicated what fraction of Q's toral reason for not doing A any given part was.50 One might then say, if P's threat was nlmth of Q's total reason for not doing A, that Q was n/m-eoerced into not doing A. If P's threat is Q's whole reason for not doing A (no part of Q's reason for not doing A) tben Q is 1-coerced (0-coerced) or, for short, coerced (not coerced). And, in the absence of precise weightS, one might begin to speak of someone's being partially coerced, slighdy coerced, almost fully coerced into doing somerhing, and so forth.51 Furthermore, without claiming that a person is 11ever to be held responsible for an act he was coerced into doing, we might, for some cases in which his reasons (other than the threat.) for doing an act aren't sufficient to get him to decide to do the act, be led to speak of a person's being (held) partially responsible for his act; not completely responsible because he did it partly because of the threat, and nor complete absence of responsibility because he didn't do it solely because of the threat, but contributed some reasons of his own. I would end by saying that the consideration of such a view of responsibility, and the tracing of the modifications in what bas been said thus far introduced by a thoroughgoing use of the notion of n/mcoerced, would require another paper- were it not for the thought that some readers might take this as a threat.
2 NEWCOMB'S PROBLEM AND TWO
PRINCIPLES OF CHOICE
Both it 011d its opposite must involve no mere artificial illusion such a.s at onGe vaniJhts upon detection, btd a natural and unavoidable illusion, which even after it l1as ceased to beguile still contimtes to de/,.de tho,.gh 1101 to dtuive u<, a11d ruhich tho,.gh thus capable of being rendered harmless can never [Je eradicated.
1M MANu L KANT, Critique
of Pure Reasm-t, .\41.1., B4 50
Suppose a being in whose power to predict your choices you have enor mous confidence. (One might rell a science-fiction story abour a being from another planet, with an advanced technology and science, whom you know tO be friendly, etc.) You know that this being has often cor reedy predicted your choices in the past (and has never, so far as you know, made an incorrect prediction about your choices), and furthermore you know that this being has often correctly predicted the choices of other people, many of whom are similar to you, in the particular sit uation to be described below. One might tell a longer story, but all rhis leads you to believe rbat almost certainly this being's prediction abour your choice in the situation to be discussed will be correct. There are rwo boxes, (Bl) and (82). (B I) contains $1,000. (82) con rains either $1,000,000 ($M), or nothing. What the content of (82) depends upon will be described in a moment. {Bl) {$1,0001 {82) {
~~ }
You have a choice between two ac.tions:

(1) taking what is in both boxes (2) taking only what is in the second box. Furthermore, and you know this, the being knows that you know this, and so on: (I) IJ the being predicts you will rake what is in both boxes, he does not put the $M in the second box. (II) Ii the being predicts you will take only what is in the second box,
he does put the $M in the second box.1 The situation is as follows. First the being makes its prediction. Then it puts the $M in the second box, or does not, depending upon what it has predicted. Then you make your choice. What do you do? There are two plausible-looking and highly intuitive arguments which require different decisions. The problem is to explain why one of them is not legitimately applied to this choice situation. You might reason as follows: First Argument: If l rake what is in both boxes, the being, almost certainly, will have predicted this and will not have put the $M in the second box, and so I will, almost certa inly, get only $1,000. If I take only what is in rhe second box, the being, almost certainly, will have predicted this and will have put the $M in the second box, and so I will, almost certainly, get $M. Thus, if I take what is in both boxes, l, almost certainly, will get $1,000. If I rake only what is in the second box, I, almost certainly, will get $M. Therefore I should take only what is in the second box. Second Argumellt: The being has already made his prediction, and has already either put the $M in the second box, or has not. The $M is either already sitting in the second box, or it is not, and which situation obtains is already fixed and determined. If the being has already put the $M in the second box, and I take what is in both boxes I get $M + $1,000, whereas if I take only what is in the second box, I get only $M. If the being has not put the SM in the second box, and I rake what is in both boxes I get $1,000, whereas if I rake only what is in the second box, I get no money. Therefore, whether the money is there or not, and which it is already fixed and determined, I get $1,000 more by taking what is in both boxes rather than taking only what is in the second box. So I should rake what is in both boxes. Let me say a bit more to emphasize the pull of each of these arguments:
Copyrighted material Newcomb's Problem and Two Principles of Choice ~ 47
The First: You know thar many persons like yourself, philosophy teachers and students, etc., have gone through this experiment. All those who rook only what was in the second box, including those who knew of the second argument but did nor follow ir, ended up with $M. And you know rhat all the shrewdies, all those who followed rhe second ar gument and took what was in borh boxes, ended up with only $1,000. You have no reason to believe rhat you are any different, vis-a-vis predictability, than they are. Furthermore, since you know thar I have all of rhe preceding information, )'Ou know thar I would bet, giving high odds, and be rational in doing so, thar if you were tO take both boxes you would get only $1,000. And if you were ro irrevocably rake both boxes, and there were some delay in the results being announced, would nor ir be rational for you to then bet with some third party, giving high odds, that you will ger only $1,000 from the previous transaction? Whereas if you were to rake only what is in the second box, would nor ir be rational for you to make a side bet with some third parry that you will get $M from the previous transaction? Knowing all rhis (though no one is actually available ro ber wirh), do you really want ro rake what is in both boxes, acting against what you would rationally want to bet on? The Second: The being has already made his prediction, placed the $M in the second box or not, and then left. This happened one week ago; rhis happened one year ago. Box (Bl) is transparent. You can see rhe $1,000 sitting there. The $M is already either in rhe box (B2) or not (though you cannot see which). Are you going ro take only what is in (B2)? To emphasize further, from your side, you cannot see through (B2), but from the other side it is transparent. I have been sitting on the other side of (B2), looking in and seeing what is there. Either I have already been looking at the $M for a week or I have already been looking at an empry box for a week. If the money is already there, it
wiU stay there whatever you choose. It is not going ro disappear. If it is not already there, if I am looking at an empty box, it is nor going to suddenly appear if you choose only what is in the. second box. Are. you going to rake only what is in the second box, passing up the additional $1,000 which you can plainly see? Furthermore, I have been sitting there looking at the boxes, hoping that you will perform a particular action. Internally, I am giving you advice. And, of course, you already know which advice I am silently giving to you. In either case (whether or nor I see rhe $M in rhe second box) I am hoping thar you will take what is in both boxes. You know that the person sitting and watching it all hopes that you will rake the contents of both boxes. Are you
going to take only what is in the second box, passing up the additional $1,000 which you can plainly see, and ignoring my internally given hope that you rake both? Of course, my presence makes no difference. You are sitting there alone, but you know that if some friend having your interests at heart were observing from the other side, looking into both boxes, he would be hoping that you would take both. So will you take only what is in the second box, passing up the additional $1,000 which you can plainly see? I should add that I have pur this problem to a large number of people, both friends and students in class. To almost everyone it is perfectly clear and obvious what should be done. The difficulty is that these people seem to divide almost evenly on the problem, with large numbers thinking that the opposing half is just being silly.2 Given two such compelling opposing argu.ments, it wi ll not do to rest content with one's belief that one knows what to do. Nor will it do to just repeat one of the arguments, loudly and slowly. One must also disarm the opposing argument; explain away its force while showing it due respect. Now for an unusual suggestion. It might be a good idea for the reader to stop reading this paper at the end of this section (but do, please, return and finish it), mull over the problem for a while (several hours,
days) and then return. It is not that I claim tO solve the problem, and do not want you to miss the joy of puzzling over an unsolved problem. It is that I want you to understand my thrashing about. II My strategy in attacking this problem is ostrich-like; that is, I shall begin by ignoring it completely (except in occasional notes) and proceed to discuss contemporary decision theory. Though the problem is not, at first., explicitly discussed, the course my discussion takes is influenced by my knowledge of the problem. Later in the paper, I shall remove my head from the sand and face our problem directly, having advanced toward a solution, I hope, or at least having sharpened and isolated the problem. Writers on decision theory state two principles to govern choices among alternative actions. Expected-Utility Principle: Among those actions available to a person, be should perform an action with maximal expected utility.
Newcomb's Problem and Two Principles of Choice -1> 49 The expected utility of an action yielding the exclusive outcomes Ot, . .., 0. with probabilities Ph ... , Pn respectively,
i.e.,
L p; x u(O;).
,=I
Dominance Principle: lf there is a partition of states of the world such that relative to it, action A weakly dominates action 8, then A should be performed rather than 8. Action A weakly dominates action B for person P iff, for each state of the world, P either prefers the consequence of A to the consequence of 8, or is indifferent berween the two consequences, and for some
state of the world, P prefers the consequence of A to the consequence of B. There are many interesting questions and problems about the framework used or assumed by these principles and the conditions governing preference, indifference, and probability which suf6ce to yield the utility measure, and the exact way the principles should be formulated. 3 The problem I want to begin with is raised by the fact that for some situations, one of the principles listed above require.s that one choose one action whereas the other principle requires that one choose another action. Which should one follow? Consider the following situation, where A and B are actions, 51 and S2 are states of the world, and the numerical entries give the utiliry of the consequences, results, effects, outcomes, upshots, events, states of affairs, etc., that obtain, happen, hold, etc., if the action is done and the state of the world obtains.
s1 s2
A: 10 4
8: 8
According to the dominance principle, the person should do A rather than B. (ln this situation A strongly dominates 8 , that is, for each state of nature the person prefers the consequence of A to the consequence of B.) But suppose the person believes it very likely tbat if be does A, S2 will obtain, and if he does B, 51 will obtain. Then he believes it very likely that if he does A he will get 4, and if he does 8 he will get 8.
Copyrighted material SO ~ CHOICE AND UTILITY
s1 s2
A: 10 4 8: 8 3 The expected utility of A = prob(SJIA) 10 + prob(S2JA)4. The expected utility of 8 = prob(S1j8)8 + prob(S218)3. U, for example,
prob(SdA) = .2 prob(S2IA) = .8 prob(Stl8) = .9 prob(Sz!B) = .1, then the expected utility of A= 5.2, and the expected utility of 8 = 7.5. Thus the expected-utility principle requires the person to do 8 rather than A.4 The dominance principle as presented here speaks of dominance relative to a partition of the stares of the world. This relativization is normally nor made explicit, which perhaps accountS for the fact that writers did not mention that it may be that relative to one partition of the states of the world, one action A dominates another, whereas rcla tive to another partition of the states of the world, it does not. It will be helpful to have before us rwo facts: First: Suppose a matrix is given, with states Sh ... , S.,, in which action A does not dominate action 8. U there is some rearrangement of the utility entries in the row for action A which gives a new row which dominates the row for acrion 8, then there are stares Th ... , T., such that in the matrix with these stares, action A dominates action 8. Proof: I shall describe how one can get the appropriate states T1> . .. , T., in one case. It is obvious how this procedure can be used generally. Suppose that ah .. . , a., and bt, ... , b., are utility numbers such that, for all i, a1 > b1, and for some i, a1 > b1 We may suppose rhar a1 is the en tty in the A row for the i rh column, that is, for state S;. We might, for example, have the following matrix:
s.,
A: a1 az a3 a.,
8: b12 b3 b19 bG
Let
Newcomb's Problem and Two Principles of Choice -& S I Tt = A & S12 or B&St5 T2 = A & S3 or 8&S2 T3 = A & S19 or 8 & S3
T, = A & S6 or 8 & Sn .
Thus we get the matrix, T. A: 0)2 OJ Ot9 06

B: bn b3 b19 b6
In this matrix, action A dominates action 8. Since the kind of procedure followed does not depend on any specific fearures of the example, the point is made.
Seco11d: Suppose there is a matrix with states s~, ... , s. such that action A dominates action B. U there is some rearrangement of the utility entries in the B row so that the rearranged row is not dominated by A, then there are states T, ... , Tn such that if the matrix is set up with these states, B is not dominated by A. Proof: Suppose that a1 > b1, for aU i; a1 > b1 for some i; and that some B-row value is greater than some Arow value. (Given that there is some arrangement in which A dominates 8, this last supposition follows from its being possible to rearrange the 8 row so that it is not dominated by the A row.) Suppose, without loss of generality that b12 > 02. Thus we have the following matrix: Let
Tt = St
T2 = S & S2 orB & Sn T3 =SJ Copyrighted material S2 -& CHOICE AND UTILITY Ttt = Sn Ttz = A & S12 or B & Sz
Tu = St3 T. = S .
Thus we get the following matrix: Since b,z > a2, A does not dominate B. It may seem that the states Tt. . . . , T. defined in terms of the actions
A and B, and the states Sh ... ,
s. are contrived states, which some
general condition could exclude. It should be noted that-since the states Sh ... , can be defined in terms of the actions A and B and the states Th . .. , T. (I will give some examples below)-anempts to show that T, . . . , r. are contrived will face many of the problems encountered in ruling out Goodman-style predicates. Furthermore, as we shall see soon, there are cases where tbe S states and the T states, which are inrerdefinable in this way, both seem perfectly natural and uncontrived. The fact that whether one action dominates another or not may depend upon which particular partition of the states of the world is used would cause no difficulty if we were willing to apply the dominance principle to ar~y partition of the states of the world. Since we are not, this raisecS the question of when the dominance principle is to be used. l et us look at some examples. Suppose that I am about to bet on the outcome of a horse race in which only two horses, H1 and H2, are running. let: St =Horse Ht wins the race. S2 = Horse Hz wins the race. A 1 = I bet on horse Ht. A2 = I bet on horse Hz. Suppose that I will definitely bet on one of the two horses, and can only bet on one of the two horsec5, and that the following matrix describes the situation. (I might have been offered the opportunity to enter this situation by a friend. Certainly no race track would offer it to me.)
s.
Newcomb's Problem and Two Principles of Choice ~ 53 St l win $50 I lose $6
s2
I lose $5 I win $49 Suppose further that the personal probability for me that H1 wins is .2, and the personal probability for me that H2 wins is .8. Thus the expected utility of A1 is .2 x u(l win $50) + .8 x u(l lose $5). The expected utility of A 2 is .2 x 11(I lose $6) + .8 x u(l win $49). Given my utility assignment ro these outcomes, the expected utility of Az is greater than that of A1. Hence the expected-utility principle would have me do Az rather than At However, we may set the matrix up differently. Let: S3 =I am lucky in my bet. S4 = I am unlucky in my bet. (Given that I am betting on only one horse today, we could let S3 =The only horse I bet oo today wins. Similarly for s., with " loses" substituted for "wins.") Thus we have the following matrix: S3 I win $50 I win $49
s4
I lose $5 I lose $6 But when set up in this way, A 1 dominates Az. Therefore the dominance principle would have me do A 1 rather than Az.6 In this example, the states are logically independent of which action I perform; from the fact that I perform A1(Az) one cannot deduce which state obtains, and from the fact that S1 (Sz, S3, S4) obtains one cannot deduce which action I perform. However one pair of states was nor probabilisricaUy independent of my doing the actions? Assuming that S1 and Sz are each probabilistically independent of both A 1 and
Az, prob(S31l do At)= .2; prob(SJII do Az) = .8; prob(S411 do At) = . 8; prob(S4(l do A2) = .2. Thus neither of the states S3 or S4 is probabilistically independent of each of the actions A1 and Az.8 In this example, it is clear that one doe.s not wish to follow the recommendation of the dominance principle. And the explanation seems to hinge on the fact that the states are not probabilisticaUy independent of the actions. Even though one can set up the situation so that one action dominates another, I believe that if I do A1, the consequence will probably be the italicized consequence in its row, and I believe that if I do A2, the consequence will probably be tbe italicized consequence in Az's row.
And given my assignment of utilities in this case, and the probabilities I assign (the conditional probabilities of rhe states given the actions) it is clear why I prefer to do Az, despite rhe fact that A1 dominates A2.
SJ
I win $50
I win $49
s4
/lose $5 I lose $6
Let us consider another example: Suppose that I am playing roulette on a rigged wheel, and that rhe owner of tbe casino offers me a chance ro choose between actions A 1 and Az so that the following matrix describes the situation (where S1 black comes up on the next spin; S2 red comes up on the next spin):
sl
I win $10
I win $5
s2
I win $100
1 win $90
Fina lly suppose that I know that the owner's employee, who is oversee ing the wheel and who I am confident is completely loyal to the owner, has been instructed to make black come up on the next spin iJ I choose A 1 and ro make red come up on the next spin if I choose A2. Clearly even though A 1 dominates Az, given my knowledge of the situation I should choose A2. I take it that this needs no argument. It seems that the reason that I should nor be guided by dominance considerations is that the states S1 and Sz are not probabilistically independent of my actions A 1 and A2. We can set up the situation so that the states are probabilisticaUy independent of the actions. But when set up in this way, I am led, given my utility assignment to the outcomes, to do A2. Let S3 = rhe fellow running the roulette wheel follows his boss's instructions; S4 = the fellow running the roulette wheel disobeys his boss's instructions. (Nore that S3 = A1, &S1 or A1 & S2; s. =A, & Sz or A2 & S1.) We then have the following matrix:
SJ
lwinSlO I win S90
s.
I win SlOO
I win $5 Even if I am not sure that S3 is true, so long as the personal probability of S3 for me is sufficiently high, I will be led to do A2, given my utility assignment to the outcomes.
Newcomb's Problem and Two Principles of Choice ~ 55 These examples suggest that one should not apply the dominance principle to a situation where the states are not probabilistically independent
of the actions. One wishes instead to maximize the expected utility. However, the probabilities that are to be used in determining the expected utility of an action must now be the conditional probabilities of the states given that the action is done. (This is true generally. How ever, when the stares are probabilistically independent of the actions, the conditional probability of each state given that one of the actions is done will be equal to the probability of the state, so the latter may be used.) Thus in the roulette wheel example, we may still look at the first matrix given. However, one does not wish to apply the dominance principle but to lind the expected utility of the actions, which in our example are: E.U.(At) = prob(Sti At) x u(I win $10) + prob(SliAt) x u(l win $100) E.U.(A2) = prob(SdA2) x u(l win $5) + prob(S2IA2) x u(l win $90).9 The following position appropriately handles the examples given thus far (ignoring Newcomb's example with which the paper opens) and has intuitive appeal.10 (1) It is legitimate to apply dominance principles if and only if the states are probabilistically independent of the acrions. (2) If the states are not probabilistically independent of the actions, then apply the expected-utility principle, using as the probabilityweights the conditional probabilities of tbe states given the ac tions. Tbus in the following matrix, where the entries in the matrix are utility numbers,
s. s2 s.
A: 0'1 02 On
8: u.
u2 u.
the expected utility of A is 2::7=1 prob(S;IA)0 1, and the expected utility of 8 is 2::7=1 prob(S!IB)U;.
Copyrighted material 56 ..,. CHOIC AND UTILITY Ill
Is this position satisfactory? Consider the following example: P knows that S or T is his father, but be does not know which one is. S died of some terrible inherited disease, and T did not. It is known that this disease is genetically dominant, rutd that P's mother did not have it, and that S did not have the recessive gene. US is his father, P will die of this disease; if T is his father, P will not die of this disease. Furthermore, there is a well<anfitmed theory available, let us imagine, about the genetic transmission of the tendency to decide to do acts which form parr of an intellectual life. This tendency is genetically dominant. S had this tendency (and did not have the recessive gene), T did not, and P's mother did not. P is now deciding whether (a) to go to graduate school and then teach, or (b) to become a professional baseball player. He prefers (though not enormously) the life of an academic to that of a professional athlete. S is P's father T is P 's father
A: X
8: z y
w
x = P is an academic for a while, and then dies of the terrible disease; z = P is a professional athlete for a while, and then dies of the terrible disease; y = P is an academic and leads a normal academic life; w = P is a professional athlete and leads the normal life of a professional athlete, though doing a bit more reading; and P prefers x to z, and y to w.
However, the disease is so terrible that P greatly prefers w to x. The matrix might be as follows: A:
8:
S is P's father
- 20
-25 T is P's father
100 95
Suppose that our well-confirmed theory tells us, and P, that if P chooses the academic life, then it is likely that he has the tendency to choose it; if he does not choose the academic life, then it is likely that he does not have the tendency. Specifically,
Newcomb's Problem and Two Principles of Choice -& 57 prob(P has the tendency( P decides to do A) = .9 prob(P does not have che tendency(? decides co do A)= .1 prob( P has the tendency( P decides ro do B) = .1 prob(P does not have the tendency(? decides to do B)= .9. Since P has the tendency iff S is P's father, we have prob(S is P's father( P decides to do A) = .9 prob(T is P's father( P decides to do A) = .1 prob(S is P's father(P decides to do B)= .1 prob(T is P's father(P decides to do B) = .9. The dominance principle tells P to do A rather than B. But according to the position we are now considering, in situations in which the states are not probabilistically independent of the actions, the dominance principle is not to be used, but rather one is to use the expected-utility principle with the conditional probabilities as the weights. Using the above conditional probabilities and the above numerical assumptions about the utility values, we get:
The expected utility of A = (.9 x -20) + (.1 x 100) = -8 The expected utility of 8 = (.1 x -25) + (.9 x 95) = 83. Since the expected utility of 8 is greater tban that of A, the position we are considering would have P do B rather than A. But th.is recom mendation is perfectly wild. Imagine P saying, "I am doing B because if I do it it is less likely that l will die of the dread disease." One wants to reply, "It is true that you have got the conditional probabilities correct. If you do A it is likely that S is your father, and hence likely that you will die of the disease, and if you do B it is likely that T is your farber and hence unlikely that you will die of the disease. Bur which one of them is your father is already fixed and determined, and has been for a long time. The action you perform legitimately affectS our estimate of the probabilities of the two states, but wh.ich state obtains does not depend on your action at aU. By doing B you are not making it less likely that S is your father, an.d by doing B you are nor making it less likely that you will die of the disease." I do not claim that chis reply is without irs problems.11 Before considering another example, let us first state a principle not under attack:
The dominance principle is legitimately applicable to situations in which the states are probabilistically independent of the actions. 12 If the states are not probabilistically independent of the actions, it seems intuitive that the expected-utility principle is appropriate, and that it is not legitimate to use the dominance principle if it yields a different result from the expected-utility principle. However, in situations in which the states, though not probabilistically independent of the actions,
are already fixed and determined, where the actions do not affect whet.her or not the states obtain, then it seems that it is legitimate to use the dominance princ.iple, and illegitimate to follow the recommendation of the expected-utility principle if it differs from that of the dominance principle. For such situations--where the stares are nor probabilisrically independent of the actions, though which one obtains is already fixed and determined- persons may differ over what principle to use. Of the rwelve sorts of situation in which it is not the case both that none of the states are already fixed and determined and none of the stares are probabilistically independent of the actions, I shall discuss only one; namely, where each of the states is already fixed and determined, and none of the states is probabilistically independent of the alternative acrions.13 The question before us is: In this sort of situation, in which all of the states are already fixed and determined, and none of the states are probabilistically independent of the acts, and the dominance principle requires that one do one action, whereas the expected-utility principle requires that one do another, should one follow the recommendation of the dominance principle or of the expected-utility principle? The question is difficult. Some may think one should follow the recommendation of the dominance principle; others may think one should follow the recommendation of the expected-utility principle in such situations. Now for the example which introduces a bit of reflexivity, which I hope will soon serve us in good stead. Suppose that there are rwo inherited tendencies ("tendencies" because there is some small probability that it would not be followed in a specific situation): (1) an inherited tendency to think that the expected-uti lity principle should be used in such situations. (If P bas this tendency, he is in
state SJ.)
Newcomb's Problem and Two Principles of Choice ~ 59 (2) an inherited tendency to think that the dominance principle should be used i.n such situations. (lf P has this tendency, he is in state s2.) It is known on the basis of postmortem genetic examinations that (a) P's mother had two neutral genes. (A gene for either tendency genetically dominates a neutral gene. We need not here worry about the progeny who has a gene for each tendency.) (b) One of the men who may be P's father had rwo genes for the first tendency. (c) The other man who may be P's father bad two genes for the second tendency. So it is known that P has one of the tendencies, but it is not known which one he has. P is faced with the following choice:
s, Sz
A: 10 4
8: 8 3 The choice matrix might have arisen as follows. A deadly disease is going around, and there are two effective vaccines against it. (li both are given, the person dies.) For each person, the side effects of vaccine 8 are worse than that of vaccine A, and each vaccine has worse side effects on persons in Sz than either does on persons in S1 Now suppose that the theory about the inherited tendencies to choice reUs us, and P knows this, that from a person's choice in this situation the probabilitie.s of his having the two tendencies, given that he has one of the rwo, can be estimated, and in particular prob(S11A) = .1 prob(S2IA) = .9 prob(S118) = .9 prob(S2!B) = .1. What should P do? What would you do in this situation? P may reason as follows: if I do A, then very probably S2 obtains,
and I will get 4. I.f I do 8, then very probably S1 holds, and I will get 8. So I will do 8 rather than A. One wants to reply: whether S1 or S2 obtains is already fixed and determined. What you decide to do would not bring about one or the other of them. To emphasize this, let us use the past tense. For you are in
S1 iff you were in S1 yesterday; you arc in Sz iff you were in Sz yesterday. But to reason "If I do A then very probably I was in Si yesterday, and I will get 4. lf I do 8, then very probably I was in S1 yesterday, and I will get 8. So I will now do B rather than A" is absurd. What you decide to do does not affect which state you were in yesterday. For either state, over which you have no control, you are better off doing A rather than B. To do B for reasons such as the above is no less absurd than someone who has already taken vaccine B yesterday doing some other act C today because the prob (He was in S1 yesterday! He does C today) is very high, and he wants the (delayed) side effects of the vaccine he has already taken to be less severe. If an explanation runs from x to y, a correct explanatory theory will speak of the conditional probability prob(ylx). Thus the corre<.< explanatory theory of P's choice in this situation will speak of prob(P does AlP is in S,) prob( P does A I P is in Sz) prob(P does B IP is inS,) prob(P does BiP is in Sz). From these, the theory may enable us to determine prob(P is in S1iP does A) prob(P is in SziP does A) prob(P is in SliP does B) prob(P is in SziP does 8)
but these would not be the basic explanatory probabilities. Supposing that probabilistic explanation is legitimate, we could explain why P does A by having among our antecedent conditions the statement that P is in Sz, but we cannot explain why P is in Sz by having an1ong our antecedent conditions the statement that P does A (though P 's doing A may be our reason for believing he is in S2). Given that when the explanatory line runs from x toy (xis part of the explanation of y) and not from y to x, the theory will speak of and somehow distinguish the conditional probabilities prob(y(x), then the probability prob(x(y) will be a likelihood (as, I think, this term is used in the statistical literature). Looking at the likelihoods of the states given the actions may perhaps give one the illusion of control over the stares. Bur I suggest that when the stares are already fixed and determined, and the explanatory theory Copyrighted material Newcomb's Problem and Two Principles of Choice ~ 61 has the influence running from the states to the actions, so that the conditional probabilities of the states on the actions are likelihoods, then if the dominance principle applies, it should be applied. If a state is part of the explanation of deciding to do an action (if the decision is made) and this state is already fixed and determined, then the decision, which has not yet been made, cannot be part of the explanation of the state's obtaining. So we need not consider the case where prob (statelaction) is in the basic explanatory theory, for an already fixed state.14 What other possibilities are there for already fixed and determinded states? One possibility would be a situation in which the states are not part of the explanation of the decision, and the decision is not part of the explanation of which state obtains, but some third thing is part of the explanation of the states obtaining and the decision's being made. Hence neither prob(state of the matrix obtaining I P does a specific action) nor prob(P does a specific actionlstate of the matrix obtains) would be part of the basic explanatory theory (which has
conditional probabilities from antecedent to consequent going in the direction of explanation). Let us consider a case like this, whose matrix exemplifies the srructure of the prisoners' dilemma situation, much discussed by game theorists.15 There are two people, ([} and (II), and the following matrix describes their situation (where the first entry in each box represents the payoff
to
person (f) and the second entry represents the payoff to person (H)). The situation arises just once, and the persons cannot get together to agree upon a joint plan of action.
(I) A: 8:
c
10,3 8, 8 (II) D
4, 4.
3, 10 Notice that for person (I), action A dominates action 8, and for person (ll), action D dominates action C. Hence if each performs his dominant action, each ends up with 4. But if each performs the oondominant action, each ends up with 8. So, in this siruation, both persons' following the dominance principle lea,es each worse off than if both did not fo llow the dominance principle. People may differ over what should be done in this situation. Let us, once again, suppose that there are two inherited tendencies, one to perform the dominant action in this situation, and one to perform Copyrighted material
62 -t> CHOICE AND UTILITY
the other action. Either tendency is genetically dominant over a
possible third inherited trait. Persons (I) and (II) are identical twins, who care only about their own payoffs as represented in this matrix, and know that their mother had the neutral gene, one of their two possible fathers had only the gene to perform the dominant action, and the other had only the gene not to perform the dominant action. Neither knows which man was their father, nor which of the genes they have. Each knows, given the genetic theory, that it is almost certain that if be performs the dominant (dominated) action his brother will also. We must also suppose that the theory tells us and them that given all this information upon which they base their choice, the correlation between their actions holds as almost certain, a.nd also given this additional information, it holds as almost certain, etc. I do not wish here to discuss whether one should or should not perform the dominant action in prisoners' dilemma situations. I wish merely to consider the following argument for not performing the dominant action in the situation I have just described. Suppose brother I argues: "If I perform the dominant action then it is almost certain 1 that I have that gene, and therefore that my brother docs also, and so it is almost certainz 16 that he will also perform the dominant action and so it is almost certain2 that I will get 4. Whereas if I perform the dominated action, for similar reasons, it is al.most certain that my brother will also, and hence it is almost certain that I will get 8. So I should perform the dominated action." Here one surely wishes to reply that this argument is not a good argument for performing the dontinared action. For what this brother does will not affect what rhe other brother does. (To emphasize this, suppose that brother D has already acted, though brother I does not yet know what he has done.) Perhaps in prisoners' dilemma situations one should perform the dominated action, but this argument does not show that one should in this situation.
The examples thus far considered lead me to believe that if the actions or decisions to do the actions do nor affect, help bring about, influence, etc., which state obtains, then whatever the conditional probabilities (so long as they do not indicate an influence), one should perform the dominant action. If the considerations thus far adduced are convincing, then it is clear that one should also choose the dominant action in the following situations, having the same structure (matrix) as Newcomb's, and differing only in that:
Newcomb's Problem and Two Principles ol Choice ~ 63 (1) The being make.s his prediction and sets the process going whereby the $M gets placed in the second box, or not. You then make your choice, and after you do, the (long) process terminates and the SM gets in the box, or not. So while you are deciding, the SM is not already there, though at this rime he has already decided whether it will be or not. (2) The being gathers his data on the basis of which be makes his prediction. You make your choice (e.g., press one of two burtons which will open one or both boxes later by delayed action), and he then makes h.is prediction, on the basis of the data previously gathered, and puts the SM in, or not. This suggests that the crucial fact is not whether the states are already fixed and derermined but whether the actions influence or affect which state obtains. Serting up a simple matrix, 17 we have the following possibilities (with the matrix entries being recommended decision policies for the situation). The actions influence which state obtains. The conditional probabilities differ. No influence of actions on states. However, conditional
probabilities differ. No influence of actions on states. The conditional probabilities are all the same. A dominant action is available (I) Maximize (Ill) expected utility (V) Do dominant action (or, equivalently,
maXJmtze
expected utility) No dominant action is available (II) Maximize

(IV)
expected utility (Vl) Maximize expected utility The standard theories make the recommendations in (V) and (Vl). They do not consider (I) and (II), but (ignoring other difficulties there might be with the policy) maximizing expected utility seems reasonable here. The di6culries come in the middle row. (Ill) is the situarion exemplified b)' Newcomb's situation and the other examples we have listed
from the person choosing whether to lead the academic life, onwards. I have argued that in these situations one should choose the dominant action and ignore the conditional probabilities which do nor ind.icate an influence. What then should one do in situation (fV), where which action is done does not influence which state obtains, where the conditional probabilities of the states given the actions differ, and where no dominanr action is available. If the lesson of case (ill) is that one should ignore conditional probabilities which do not indicate an inHuence, must not one ignore them completely in case (IV) as well? Not exactly. What one should do, in a choice between two actions A and B, is the following. 18 Let Pt. ... , Pn be the conditional probability distribution of action A over the n states; let q~, ... , q. be the conditional probability distribution of action B over the n stares. A probability distribution '" . .. , r., summing to 1, is between Pt. ... , Pn and qh ... , q. iff for each i, r1 is in the closed interval (Ph q1] or (q" p;j. (Note that according to this account, Ph .. . , Pn and q~, . . . , q. are each between p., ... , Pn and q~, . .. , q . ) Now for a recommendation: If relative to each probability distribution between Ph . .. , Pn and qh . .. , q., action A has a higher expected utility than action B, then do action A. The expected utility of A and B is computed with respect to the same probability distribution. It will not, of course, be rhe case that relative to every possible probability distribution A has a higher expected utility than B. For, by hypothesis, A does not dominate B. However, it may be that relative tO each probability distribution between Ph . .. , p. and q., . .. , q,, A has a higher expected utility than 8. If, on the other hand, ir is not the case that relative ro each probability distribution between Ph .. . , Pn and qh ... , q., A has a higher expected utiliry than 8 (and it is not the case that relative to each, B has a higher expected utility than A), then we are faced with a problem of decision under constrained uncertainty (the constraints being the end probability distributions), on
which kind of problem there is nor, so far as I know, agreement in the literarure_I9 Since consideration of the issues raised by such problems would take us far afield, we thankfully leave them. To talk more objectively than some would like, though more intuitively than we otherwise could, since the actions do not affect or influence which state obtains, there is some one probability distribution, which we do not know, relative ro which we would like to compare the action A and B. Since we do not know the distribution, we cannot proceed as in cases (V) and (VI). Bur since there is one unknown correct distribution "out there," unaffected by what we do, we must, in the
procedure we use, compare each action with respect to the same dis tribution. Thus it is, at this point, an irrelevant fact that one action's expected utility C{)mputed with respect to one probability distribution is higher than another action's expected utility computed wirh respect ro another probability disrriburion. It may seem strange that for case (IV) we bring in the probabilities in some way (even though they do not in dicare an influence) whereas in case (Ill) we do not. This difference is only apparent, since we could bring in the probabilities in case (ill) in exactly the same way. The reason why we need not do this, and need only note that A dominates B, is that if A dominates 8, then relative to each probability distribution (and therefore for each one between the conditional ones established by the two actions) A has a higher expected utility than 8. 20 Now, at last, to rerum to Newcomb's example of the predictor. Ii one believes, for this case, that there is backwards causality, that your choice causes the money to be there or not, that it causes him to have made rhe prediction that he made, then there is no problem. One rakes only whar is in rhe second box. Or if one believes thar the way the predictor works is by looking into the furure; he, in some sense, sees what you
are doing, and hence is no more likely to be wrong about what you do than someone else who is standing there at the time and watching you, and would normally see you, say, open only one box, then there is no problem. You take only what is in the second box. But suppose we establish or take as given that there is no backwards causality, that what you actually decide to do does not affect what he did in the past, that what you actually decide to do is not part of the explanation of why be made the prediction he made. So let us agree that the predictor works as follows: He observes you sometime before you are faced with the choice, examines you with complicated apparatus, etc., and then uses his theory to predict on the basis of this state you were in, what choice you would make later when faced with the choice. Your deciding to do as you do is not parr of the explanation of why he makes the prediction he d oes, though your being in a certain state earlier is parr of the explanation of why he makes the prediction he docs, and why you decide as you do. I believe that one should take what is in both boxes. I fear that the considera6ons I have adduced thus far will not convince those propo nents of taking only what is in tbe second box. Furthermore, I suspect that an adequate solution to this problem wiU go much deeper than I have yet gone or shall go in this paper. So I want to pose one question.
Copyrighted material 66 -t> CHOICE AND UTILITY
I assume that it is clear that in the vaccine example, the person should not be convinced by the probability argument, and should choose the dominant action. I assume also that it is clear that in the case of the two brothers, rhe brother should not be convinced by the probability argument offered. The question I should like to put to proponents of taking only what is in the second box in Newcomb's example (and hence nor performing the dominant action) is: what is the difference between Newcomb's example and the ocher two examples which make the difference between not following the dominance principle and following it? U no such difference is produced, one should nor rush ro conclude
that one should perform the dominant action in Newcomb's example. For it must be granted that, at the very least, it is nor as dear that one should perform the dominant action in Newcomb's example, as in the other two examples. And one should be wary of anempting w force a decision in an unclear case by producing a similar case where the decision is clear, and challenging one to find a difference between the cases which makes a difference to the decision. For suppose the undecided person, or the proponent of another decision, cannot find such a difference. Does nor the forcer, now, have to find a difference betwe.en the cases which explains why one is clear and the other is nor? And might not this difference then be produced by the other person as that which perhaps should yield different decisions in the two cases? Sometimes chis will be implausible; e.g., if the difference is that one case is relatively simple, and the other has much additional detail, individually irrelevant, which prevents the other case from being taken in as a whole. But it does seem that someone arguing as I do about a case must not only: (a) describe a similar case which is clear, and challenge the other to state a difference between them which should make a difference to how they are handled, but must also (b) describe a difference between the cases which explains why though one case is clear, the other is not, or one is tempted to handle the other case differently. And, assuming that all accept the difference stared in (b) as e>.:pla in.ing what it is supposed to explain, (I) The simplest situation is that in which all agree that the difference mentioned in (b) is not a reason for different decisions in the two cases. (n) However, if the forcer says it is not a reason for different decisions in the rwo cases, and the other person says it is or may be,
difficult questions arise about upon whom, if anyone, the burden of further argument falls. What then is the difference that makes some cases clear and Newcomb's example u.nclear, yet does not make a difference to how the cases should be decided? Given my account of what the crucial factors are (influence, etc.), my answer to this question will have 10 claim that the clear cases are clear cases of no influence {or, tO recall the cases which we considered at the beginning, of influence), and that in Newcomb's example there is the illusio11 of influence. The task is to explain in a sufficiently forceful way what gives rise to this illusion so that, even as we experience it, we will not be deceived by it. I have said that if the action is referred to in an explanation of the state's obtaining, so that the doing of the action affects or influences which state obtains, then the dominance principle should not be applied. And if the explanation of the states' obtaining does not make reference to the action, the action does not influence which state obtains, does not {partly) bring it about that a state obtains, then the dominance principle should be applied to such situations where a dominant action is available. But if this is so, where is there room for unclarity about a case? What other possibility is there? Either the action is referred 10 in the explanation of the state's obta_ining, or it is not. How does the temptation to take only what is in the second box arise in the Newcomb example, and why does it linger? The possibility to which I wish co call anention can be described differently, depending upon other views which one holds. (I describe the possibility specifically with Newcomb's example in mind.) {1) The action is referred to in the explanation of the state's obtaining, but the term which refers to the action occurs in the explanation, in a nonextensional belief context. Thus it does not foUow from the fact that the
action is referred to, in this way, in the explanation of the state's obtaining, that the doing of the action affect.s which state obtains. {2) The action is not referred to in the explanation of the state's obtaining. What is brought in by the explanation of the state's obtaining is some being's weU-founded beliefs about the person's doing the action. Since the person's doing the action is not part of the explanation of the state's obtaining, it does not affect or influence which state obtains. In Newcomb's example, the predictor makes his prediction on the basis of determining what state the person is in, and what his theory teUs him about what such a person will do in the choice situation.
Copyrighted material 68 -i> CHOICE AND UTILITY
Bdieving his theory accurate, he putS the money in or not, according to his belief about the person's future actions, where th.is belief depends upon his finding out what initial state the person is in, and what his theory tells him about a person in such an initial state. Thus, if the predic.tor puts the $M in the second box, part of the explanation of this is his belief that the person will take only what is in the second box. If he does not put the SM in the second box, part of the explanation of this is his belief that the person will take what is in both boxes. Thus the explanation of the money's being in the second box (or not) refers to the person's action only in a nonextensional belief context (or does not refer ro it at all but only to the predictor's beliefs about it). It is apparendy a persistent temptation for people to believe, when an explanation of something x brings in terms referring to y in a nonextensional belief context (or brings in beliefs about y), that y, in some way, influences or affects x. Thus one finds writers on teleological explanation having ro stare that in the simple case where someone goes to the refrigeraror 10 get an apple, it is not the apple's being there when he getS there which caused him to go, or which (pardy) explains his actions, but rather his beliefs about an apple's being there. But this cannot be
the whole story about Newcomb's example. For there are many persons not at all tempted to say that the apple's being there when he gets there in11uenced his action of going there, who do want to or are tempted to take only what is in the second box. Let us return to the writers on teleology. To show that the apple's being there does not influence the person's actions, but rather it is his beliefs about the apple's being there that do, they usually argue t.hat even if the apple were not there, so long as the person had the beliefs, he would act in the same way. The relevant feature of nonextensional belief contexts here is that from P believes that . . . x . .. , it does not follow that x exists, from P believes that p, it does not follow that p is true. So, the argument runs, he could have his beliefs without there being an apple there, and this shows that the apple does not in1luence his actions in this case. And surely the explanation of his action should be the same, in the case where the apple is in the refrigerator, as in the case where it is not though he believes it is. The parallel argument for Newcomb's example would run: The predictor could believe that you will rake only the second even if you do not. This shows that your action does not influence whether or not the money is there, but rather the predictor's beliefs about your action has this influence. But by the conditions of the problem, the predictor is almost certain to predict
Newcomb's Problem and Two Principles of Choic-e ~ 69 correctly, so that it is not clear that the predictor could believe that you will take only the second even if you do not. Thus, the condition of the problem which has the predictor's predic.tions almost certainly being correct tends to get us to treat the predictor's beliefs as though they do not have these nonextensional features. For if his predictions are almost certainly correct, then almost certainly: if he believes you will do A then you will do A. One further thing should be mentioned. lt is a reasonably intuitive principle that if R brings it about that p, and if p if and only if q (for
some "iff" stronger than the material biconditional), then R brings it about that q. Or, if it is up toR whether p, and p iff q (for some strong "iff"), then it is up to R whether q. Thus one finds writers arguing that if there are necessary and sufficient causal conditions for our actions, which conditions go back to a time before we were born, then what we do is not up to us. For, so the argument runs, those conditions obtaining before we were born clearly were not up to us, and so what they are necessary and sufficient for is not up to us either. I do not wish here to discuss whether this principle is correct. AU that is required for my purposes is that the principle have intuitive appe.al, and be a hard one to escape. This would also reinforce the feeling that as choosers in Newcomb's example, we can, somehow, influence what the predictor did. For, one might argue., Newcomb's problem is a problem for the chooser only if what he does is up to him. And if one assumes this, and the principle is operating, then it will be difficult to escape the feeling that what the predictor did is up to you, the chooser. I do not claim that this last principle alone creates the problem. For the problem does not arise in e.g., the vaccine case.21 But it does, I believe, contribute to it. Thus I wish to claim that Newcomb's example is le.ss clear than the others because (a) in it the explanation of the state's obtaining refers to the action (though this reference occurs in a nonexrensional belief context) and that (b) the conditions of the problem prevent one obvious way of refut ing the teleologist's view in this case (which view depends upon the truth that generally if y is part of the explanation of x, then y influences x ).
Copyrighted material 70 ~ CHOICE AND UTILITY This leads ro rhe feeling rhar, somehow, you as chooser can influence what rhe predictor did, and this feeling is perhaps reinforced by
the operation of the intuitive principle. All this leads to the lurking feeling rhat one can now choose ro rake only what is in the second box, and so make oneself rhe sort of person who does so, and so, somehow, influence what the predictor did. I hope you find this explanation of why some cases are clear and Newcomb's is not clear acceptable, and rhat it is clear rhar rhis difference berween the cases should nor make a difference to how they are decided.22 At this point one perhaps wants to say, "lf you produce a case having the features you say disringuish Newcomb's example from the orhers, where it is clear that the dominant action should be performed, rhen I will be convinced rhat the dominant action should be performed in Newcomb's example. But not unti l." I.f I am right about the role of similar examples, then this cannot be done; an answer to Newcomb's example cannot be forced in this way. Or rather, if it can be done, then it will show that I have nor picked out the right difference. For i_f one case that 6rs my description is clear, and another which 6rs it is not dear, then we still have to produce features to explain why one
is dear and the other is not. And perhaps those features should make a difference between the decisions in the two cases. At some point, given an acceptable explanation of why one case is clear and another is not, one just has to see that the explanatory features do not make a difference ro what should be decided in rhe two cases. Or, at any rare, the point that the explanatory features do nor make a difference to what should be decided can itself be forced by a dear case only at the cost of the claim that those very features explain why some cases are clear and others are not. In closing this paper, I must muddy up the waters a bit (more?). (l ) Though Newcomb's example suggests much about when to apply the dominance principle, and when to apply the expectedutiliry principle (and hence is relevant to formal decision theory), it is not the elo:pected-utiliry principle which leads some people to choose only what is in the second box. For suppose the probabiliry of the being's predicting correctly was just .6. Then the expected utility of taking what is in both boxes = prob(he predicts correctlyll take both) xu(( receive $1,000) Copyrighted material Newcomb's Problem and Two Principles of Choice ~ 71 + prob(he predicts correctly! I take only second)xuO receive $1 ,001,000) = .6 X u($1,000) + .4 >< u(S!,001,000).
The expected utility of taking only what is in the second box = .6 x u($1,000,000) + .4 x u($0). And given the utility I assume each of my readers assigns to obtaining these various monetary amounts, the expected utility of taking only what is in the second box is greater than the expected utility of taking what is in both boxes. Yet, I presume, if the probability of the being's predicting correctly were only .6, each of us would choose to take what is in both boxes. So it is not (just) the expected-utility argument that operates here to create the problem in Newcomb's example. It is crucial that tbe predictor is almost certa in to be correct. I refrain from asking a proponent of taking only what is in the second box in Newcomb's example: if .6 is not a high enough probability to lead you to take only what is in the second box, and almost certainty of correct predictions leads you to take only the second, what is the minimum probabi.lity of correct prediction which leads you to take only what is in the second box? I refrain from asking this question because I am very unsure about the force of drawing-the-line arguments, and also because the person who wishes to take what is in both boxes may also face a problem of drawing the line, as we shall see in a moment. (2) IJ the fact that it is almost certain that the predictor will be
correct is crucial ro Newcomb's example, this suggests that we consider the case where it is certain, where you know the prediction is correct (though you do not know what rhe prediction is). Here one naturally argues: I know that if I take both, I will get $1,000. I know that if I take only what is in the second, I get $M. So, of course, I will take only what is i.n tbe second. And does a proponent of raking what is in both boxes in Newcomb's example (e.g., me) really wish ro argue that it is the probability, however minute, of the predicror's being mistaken which makes the difference? Does he really wish ro argue that if he knows the prediction will be correct, he will take only rhe second, but rhar if he knows someone using the predictor's theory will be wrong once in every 20 billion cases, he will take what is in both boxes? Could the difference between one in 11, and none in 11, for arbitrarily large finite 11, make this difference? And how exactly does the fact that Copyrighted material 72 -1> CHOICE AND UTILITY the predictor is certain to have been correct dissolve the force of the dominance argument? To get the mind to really boggle, consider the following:
s, Sz A: 10 4 8: 8 3 Suppose that you know that either S1 or Sz already obtains, but you do not know which, and you know that S1 will cause you to do 8, and Sz wi ll cause you ro do A. Now choose! ("Choose?") To connect up again with a causalized version of Newcomb's example, suppose you know that there are two boxes, (81) and (82). (81) contains $1,000. (82) contains either a valuable diamond or nothing. You have to choose between taking what is in both boxes and taking only what is in the second. You know that there are two states: S1 and S2 You do not know which obtains, but you know that whichever does, it has obtained for the past week. If S2 obtains, it causes you to rake only what is in the second, and it has already caused a diamond to be produced in box (82). If S1 obtains, it causes you to take what is in both boxes, and does nor cause a diamond robe produced in the second box. You know all this. What do you choose to do? While we are at it, consider the following case where what you decide (and why) either (1) does affect which future state will obtain, upon which consequences depend, or (though this would not be the same problem for the view I have proposed, it might be for yours)
even if it does nor affect which state obtains, the conditional probabilities of the states, given what you do and why, differ. S1 Sz A: live die 8: die live (1) Apart from your decisions {if you do not know of this matrix, or know of it and cannot reach a decision), prob S1 > prob Sz (2) prob(S1ido A with (1) as reason) < prob(Szido A with (1) as reason) (3) prob(S1ido 8 with (2) as reason)> prob(S2 jdo 8 with (2) as reason) Copyrighted material and Newcomb's Problem and Two Principles of Choice ~ 73 even (n) prob(S11do A with n - 1 as reason) < prob(S2Ido A with 11 - I as reason) odd (n) prob(S1ido B with n - 1 as reason)> prob(Szldo B with n - 1 as reason) Also: prob(S11you do what you do because indifferent between A and B) > prob(S2Iyou do what you do because indifferent between A and B) prob(Sddoing A with all of the above as reason) < prob(S2Idoing A with all of the above as reason) prob(Sddoing B with all of the above as reason) > prob(S2Idoing B with all of the above as reason). Finally, where "all this" refers to all of what is above this place and, reflexively, to the next rwo, in which it appears: and prob(Sddoing A with all this as reason) < prob(S2Idoing A with all this as reason)
prob(S11doing B with all this as reason) > prob(S2Idoing B with all this as reason). What do you do? Copyrighted material 3 REFLECTIONS ON NEWCOMB'S PROBLEM This essay original!) appeartd as a gueSt eolumn in Manin Gardner's .series, .. Mathemar.ical Games; in Seie.nti/ic Ameriean. Gardner's column on Newcomb's Problem, published in july 1973, .. produd an enormous outpouring of lent:rs ... This correspoodem:c was forwarded to Noz.jck, who responded with tbc following comments. (Copyright @1974 by Scieoti6c A.merica11, Inc. All right:s reserved.) Newcomb's problem involves a being who has the ability to predict the choices you will make. You have enormous confidence in the be ing's predictive abi lity. He has already correctly predicted your choices in many other situations and the choices of many other people in the situation to be described. We may imagine that the being is a graduate student from another planet, checking a theory of terrestrial psychology, who first takes measurements of the state of our brains before making his predictions. (Or we may imagine that the being is God.) There are
two boxes. Box 1 contains Sl,OOO. Box 2 contains either $1 million or no money. You have a choice between two actions: taking what is in both boxes or taking only what is in the second box. If the being predicts you will take what is in both boxes, he does not put the $1 million in the second box. If he predicts you will take only what is in the second box, he puts the million in the second box. (If he predicts you will base your choice on some random event, he does not put the monel' in the second box.) You know these facts, he knows you know them, and so on. The being makes his prediction of your choice, puts the $1 million in the second box or not, and then you choose. What do you do? There are plausible arguments for reaching two different decisions: l. The expected-utility argument. 11 you take what is in both boxes, tbe being almost certainly will have predicted this and will not have put Copyrighted material Reflections on Newcomb's Problem ~ 7S the $1 million in the second box. Almost certainly you will get only $1,000. If you rake only what is in the second box, the being almost certainly will have predicted this and put the money there. Almost certainly
you will get $1 million. Therefore (on plausible assumptions about the utility of the money for you) you should take only what is in the second box. Take both Take only second He predicts your choice correctly $1,000 $1,000,000 He predictS your choice incorrectly $1,001,000 $0 2. The dominance argumettt. The being has already made his prediction and has either put the S 1 million in the second box or has not. The money is either sitting in the second box or it is not. The situation, whichever it is, is fixed and determined. If the being put the miUion in the second box, you will get $1,001,000 if you take both boxes and only $1 million if you take only the second box. If the being did not put the money in the second box, you will get $1,000 if you take both boxes and no money if you take only the second box. In either case you will do better by $1,000 if you take what is in both boxes rather than only what is in the second box. Take both
Take only second He put $1,000,000 into box 2 $1,001,000 $1,000,000 He did not put $1,000,000 into box 2 $1,000 $0 Each argument is powerful. The problem is to explain why one is defective. Of the first 148 letters to Scientific American from readers who tried to resolve the paradox, a large majority accepted the problem as being meaningful and favored one of the rwo alternatives. Eighty-nine believed one should take only what is in the second box, 37 believed one should take what is in both boxes-a proportion of about 2.5 to Copyrighted material 76 -& CHOICE AND UTILITY one. Five people recommended cheating in one way or another, 13 be lieved the problem's conditions to be impossible or inconsistent and four maintained that the predictor carlJlot exist because tbe assumption that he does leads to a logical contradiction. Those who favored taking only the second box tried in various ways tO undercut the force of the dominance argument. Many pointed out that if you thought of that argument and were convinced by it,
the predictor would (almost certainly) have predicted it and you would end up with only $1,000. They interpreted the dominance argument as an attempt to outwit the predictor. This position makes things too simple. The proponent of the dominance argument does believe he will end up with only $1,000, yet nevertheless he thinks it is best to take both boxes. Several proponentS of the dominance principle bemoaned the fact that rational individua ls would do worse than irrational ones, but that did not sway them. Stephen E. Weiss of Morgantown, West Virginia, rried to reconcile the two views. He suggested that following the expected-utility argu ment maximizes expectation, whereas following the dominance argu mcnt maximizes correct decision. Unfortunately that leaves unexplained why the correct decision is not the one that maximizes expectation. The assumptions underlying the dominance argument, that the S1 million is already in the second box or it is not and that the situa tion is fixed and determined, were questioned by Mohan S. Kalelkar, a physicist at. the Nevis Laboratories of Columbia University, who wrote: "Perhaps it is fa lse to say thar the being has definitely made one
choice or the other, just as it is false to say that the electron [in the twoslit ex pcriment.J went through one slit or the other. Perhaps we can only say that there is some amplitude that 82 [second box] has $1 million and some other amplitude that it is empty. These amplitudes interfere unless and until we make our move and open up the box ... To assert that 'either B2 contains $1 million or else it is empty' is an intuitive argument for which there is no evidence unless we open the box. Admittedly the intuitive evidence is strong, but as in the case of the double-slit electron diffraction our intuition c.an sometimes pwve to be wrong." Kalelkar's argument makes a version of the problem, in which the sec ond box is transparent on the other side and someone has been staring into it for a week before we make our cboice, a significantly different decision problem. It seems not to be. Erwin Schrodinger, in a famous thought experiment, imagined a cat left alone in a closed room with a vial of C)'anide that breaks if a radioactive atom in a detector decays. Copyrighted material R<flrctions on Newcomb's Problem -& 77 Must a disciple of Niels Bohr's assert that the cat is neither alive nor
dead, SchrOdinger asked, wuil measurements have been made to decide the case? Even if one accepts the Bohr interpretation of quantum mechanics, however, what choice does one make, in Newcomb's problem, when one knows that others can see ioro the b<>x from the other side and observe whether it is filled or empty? Many who wrote asserted that the dominance argument assumes the states to be probabilistically independent of the actions and pointed out that this is not true for the two states "Tbe $1 million is io Box 2" and "The S1 million is not in Box 2." The states would be probabilistically independent of the actions (let us assume) in the matrix for the utility argument, which has the states "He predicts correctly" and "He predicts incorrectly." Here, however, there is no longer dominance. Therefore it appears that the force of dominance principles is undercut. "It is legitimate to apply dominance principles if and only if the states are probabilistically independent of the actions. If the states are not probabilistically independent of the actions, then apply the expected-uti lity principle, using as the probability-weights the conditional probabilities of the states given in the actions." The quotation is from my original 1970 essay, which formulated this position, then went on to reject it as
unsatisfactory for the following reasons. Suppose a person knows that either roan S or man T is his father but he does not know which. S died of some very painful inherited disease that strikes in one's middle thirties and T did not. The disease is genetically dominant. S carried only the dominant gene. T did not have the gene. If S is his father, the person will die of the dread disease. If T is h.is father, he will not. Funhermore, suppose there is a wellconfirmed theory that stares a person who inherits this gene will also inherir a tendency toward behavior that is characteristic of intellectuals and scholars. S had this tendency. Neither T nor the person's mother had such a tendency. The person is now deciding whether to go to graduate school or to become a professional baseball player. He prefers (although not enormously) the life of an academic to that of a professional athlete. Regardless of whether or not he will die in his middle thirties, he would be happier as an academic. The choice of the academic li fe would thus appear to be his best choice. Now suppose he reasons that if he decides to be an academic, the decision will show that he has such a tendency and therefore it will be likely that he carries the gene for the disease and so will die in his middle
thirties, whereas if he chooses to become a baseball player, it will be Copyrighted material 78 _,. C HOICE AND UTILITY likely that T is his father, therefore he is not likely to die of the disease. Since he very much prefers not dying of the disease (as a baseball player) ro dying early from the disease (as an academic), he decides to pursue the career of an athlete. Surely everyone would agree that this reasoning is perfectly wild. It is true that the conditional probabilities of the states "Sis his father" and "Tis his father" are not independent of the actions "becoming an academic" and "becoming a professional athlete." l.f he does the first, it is very likely that S is his father and that he will die of the disease; if he does the second, it is very Likely that T is his father and therefore unlikely that he will die of the disease. Bur who his father is cannot be changed. It is fixed and determined and has been for a long rime. His choice of how to act legitimately affects our (and his) estimate of the probabilities of the two stares, bur which stare obtains (which person is his father) does nor depend on his action at all. By becoming a professional baseball player he is nor making it less likely that S is
his father, therefore he is not making it less likely that he will die of the disease. This case, and others more clearly including a self-reference that this case may seem to lack, led me to think probabilistic nonindependence was not sufficient to reject the dominance principle. It depends on whether the actions inOuence or affect the stares; ir is nor enough merely that they affect our judgments about whether the states obtain. How do those who reject the dominance principle for Newcomb's problem distinguish it from those other cases where dominance principles obviously apply even though there is probabilistic nonindependence? But one must move carefully here. One cannot force a decision in a difficult case merely by finding another similar case where the decision is clear, then challenging someone to show why the decision should be different in the two cases. There is always the possibility that whatever makes one case difficult and the other clear will also make a difference as to bow they should be decided. The person who produces the parallel example must not only issue his challenge; he must also offer an explanation of why the difficult case is less clear, an explanation that does not involve any reason why the cases might diverge in how they should be decided. Interested readers can find my additional parallel ex
amples where dominance is appropriate, plus an attempt to explain why Newcomb's case, although less clear, is still subject ro dominance principles, in my original essay, "Newcomb's Problem and Two Principles of Choice" (Essay 2). Copyrighted material Reflections on Newcomb's Problem ~ 79 This obligation to explain differences in the clarity of parallel examples in order to show that no different decision should be made also rests on those who argued in their letters for taking only what is in the second box. For example, it rests on Robert Heppe of Fairfax, Virginia, who said that the situation "is isomorphic with one in which the human moves first and openly," and on A. S. Gilbert of the National Research Council of Canada, who called the Newcomb case "effectually the same as" one where you act first and an observer attempts to communicate with a " mindreader" in the next room who then gue.sses your choice, using a payoff matrix identical with Newcomb's. A large number of those who recommended raking only the second box performed the expected-value calculation and concluded that, provided the probability that the being was correct was at least .5005, they would rake only the second box. Nor only did they see no problem at all; they either maximized expected monetary value or made
utility linear with money in the range of the problem. Otherwise the cutoff probability would be different. Willjam H. Riker of the department of political science at the University of Rochester suggested that people making different decisions merely differed in their utility curves for money. Such persons, however, need nor differ in their choices among probability mixtures of monetary amounts in the standard situations in order to calibrate their utilities. Those who favored taking both boxes made almost no attempt tO diagnose the mistakes of the others. An exception is William Bamberger, an economist at Wayne State University. He wrote that the proponent of choosing only the second box "computes not the alternative payoffs of choosing one or two boxes for a given individual, bur the average payoff of those who choose two as opposed to the average observed payoff of those who choose one." The problem, of course, is how to compute the probability for a given individual of his payoff for each choice. Should one use the differing conditional probabilities, or ignore them because dominance applies only when the states are probabilistically independent of the action (and so when for each stare its conditional
probabilities on each act is the same), or ignore them since the conditional probabilities of the state on the acts are to be used only when they represent some process of the act's in fluencing or affecting which state obtains? A number of respondentS said their choice would depend on whether the predictor made his prediction after they had at least started to consider the problem. If so, they would do their best to decide to take only Copyrighted material 80 -<> CHOICE AND UTILITY the second box (so that this data would be available to the predictor), and some added that they hoped they would change their mind at the last minute and take both boxes. (They gave rhe predictor too little credit.) On the other hand, if the predictor made his prediction before they even considered the problem, these writers believed they would take both boxes, since there was no possibility of their deliberations affecting the prediction that had been made. Several respondents maintained that if the conditions of the problem could be realized, we might be forced to revise our views about the impossibility of backward causality. Newcomb himself seems to think that special difficulties arise for proponents of backward causality if the predictor writes some term designating an integer on a slip of paper i.n the
second box, with the understanding that you get $1 million only if that integer is a prime. Of course, the predictor writes a prime if, and only if, he predicts that you will take the second box. How can your choice determine whether a number is prime or composite? The advocate of backward causality need nor think it does. What your choice affecrs, i.n his view, is what term the predictor writes down (or wrote down earlier), not whether the integer it designates is prime or composite. The reasoning of some of the letters indicates it would be useful to specify precisely the conditions whereby we could discover in which time-direction causality operates. Might one even say that some conditions universally preceding certain decisions are part of the effects of the decision (by backward causality) rather than part of the cause? Not everyone was willing to choose one or the other action. Among the five respondentS who suggested some form of cheating, Robert B. Pitkin, editor of Amt!rican Legion Magazine, speculated that Dr. Matrix, the numerologist, would walk in with a device to scan the contents of the boxes, take the boxes with the money in them and never open an empty box. "He quite naturally succeeded in getting aU the money, for the rule of bridge that one peek is worth two finesses
applies here too ... By introducing a choice which the being has not anticipated, and is not permitted to take into account, he achieves a stunning victory for free will." (What prevents the being from taking this into account?) Other lener writers also struck blows for free will. Nathan Whiting of New York would take both boxes but would open only the first one, leaving the second box unopened. Ralph D. Goodrich, Jr., of Castle Rock, Colorado, would take only the first box. Richard B. Miles of Copyrighted material Reflections on Newcomb's Problem ~ 81 Los Altos, California, also recommended a "creative" solution: Turn to another person before you make your choice and offer to sell him for $10,000 the contents of whatever box or boxes you choose. Isaac Asimov wrote: "I would, without hesitation, take both boxes ... I am myself a determinist but it is pedecdy clear to me that any human being worthy of being considered a human being (including most certainly myself) would prefer free wil~ if such a thing could ex ist ... Now, then, suppose you take both boxes and it turns out (as it almost certainly will) that God has foreseen this and placed nothing
in the second box. You will then, at least, have expressed your willingness to gamble on his nonomniscience and on your own free will and will have willingly given up a million dollars for the sake of that willingness-itself a snap of the finger in the face of the Almighty and a vote, however futile, for free will . . . And, of course, if God has muffed and left a million dollars in the box, then not only will you have gained that million but far more important you will have demonstrated God's nonomniscience. If you take only the second box, however, you get your damned million and not only are you a slave but also you have demonstrated your willingness to be a slave for that million and you are not someone I recognize as human." (No one wrote to argue for tak ing only the second box on the grounds that either it results in gerting $1 million or it demonstrates the being's fallibility, either of which is desirable.) Those who held that the conditions of the problem could not be rc alized were of two types. There were those who believed the situation to be physically impossible because the being could not predict all the information input of every light signal that would arrive at your eyes in
the appropriate time interval. ("To gain such knowledge the being must have a physical agency for collecting information that travels faster than the speed of light." wrote George Fredericks, a physicist at the Univer sity of Texas.) And there were those who argued that if the room is closed, the problem reduces to that of Maxwell's demon-a suggestion made by Fredericks and by John A. Ball of the Harvard College ObservatOry. Those who believed the conditions of the problem to be inconsistent as well as physically impossible said that the almost certain predictabil ity of decisions was inconsistent with free will, and therefore with mak ing choices, yet the problem assumed that genuine choices could be made. This is a hard argument to drive through because it appears to Copyrighted material 82 ~ CHOICE AND UTILITY be the choices that are predicted. The relevant connections are difficult ro get straight. Predictability of decisions does nor logically imply determinism under which the decisions are caused (for example, the possibility of backward causaliry, where an uncaused decision causes an earlier prediction, or "seeing ahead" in time in a block universe). Nor, we should note in passing, does determinism email
predictability, even in principle. Events could be fixed in accordance with scientific laws that are not recursive. Is determinism incompatible with free will? It seems to many to be so, yet the argument that determinism is incompatible with responsibility for action, which free will implies, depends on a notion of responsibility insufficiently worked out to show precisely how the connections go. Some say merely that a free act is an uncaused one. Yet being uncaused obviously is not sufficient for an act ro be free; one surely would nor be responsible for such an action. What other conditions, then, must be satisfied by an uncaused act if it is to be a free one? The literature on free will lacks a sa6sactory specification of what a free action would be like (given that " uncaused" is not enough). Perhaps if we were given this specification of additional conditions, they would turn out to be sufficient apart from the ac6on's being uncaused. Another problem will help to exhibit some complicated relations between free will and determinism. It has been asserted (by C. S. Lewis, for instance) that no determinist rationally can believe in determinism, for if determinism is true, his beliefs were ca used, including his belief in determinism. The idea seems to be that the causes of belief, perhaps chemical happenings in the brain, might be unconnected with
any reasons for thinking determinism true. They might be, but they need not be. The causes might "go through" reasons and be effective only to the extent that they are good reasons. In the same way it migbt be a causal truth about someone that he is convinced only by arguments that constitute specified rypes of good reasons (deduc6ve, induc6ve, explanatory, and so on). Some philosophers have argued recently that we know some statement p only if part of the cause (or more broadly the explanation) of our believing p is, if we pursue the story far enough, the fact that p is true. You know this magazine is before you now only if its being there is part of the explanation of why you believe it is there. II psychologists are stimulating your brain to create the illusion that you are seeing a magazine, you would not really know there is a magazine before you Copyrighted material Reflections on Newcomb's Pr.oblem ~ 83 even if a psychologist happened ro have left one on the rable in front of you. The magazine's being there would nor play the proper causal role in the story of your belief. If we do nor mind our beliefs being caused by the facts, and indeed find it somewhat plausible to think we have
knowledge only to the extent that they are, then we may also find it less disturbing that our actions are caused by certain rypes of facts holding in the world, for example, the fact that ir would be better to do one thing rather than another. To say this, of course, is not to present a theory of free action; it is merely to hint that it may be possible to remove the sting of determinism. This approach is a comfortable one when we act correctly, but it is difficult ro see how it can be extended plausibly to wrong acts where questions of responsibility arc paticularly pressmg. Proponents of the C. S. Lewis position might reply that rhe determinist should not feel so comfortable. Even though he says he is caused ro believe in determinism (and anything else) by what are good reasons, he must also maintain that he is caused to believe that such reasons are good reasons. He may have a second set of reasons for believing the first set of reasons are good. Now, however, his opponent can raise the same question as before. Why does he believe the second ser of reasons? The determinist must end either b) finding self-supporting reasons (which say of themselves that they are good reasons) or by admitting that the
best explanation of why he believes they are good reasons is that they are. This surely leaves his opponent unsatisfied, and the march seems to be a draw. Those who believe in free will find rhemselves in similar dilemmas. Kun Rosenwald of Washington wrote: "When I was 19 or 20, I thought about the free-will problem ... and I came ro this conclusion: If we make an exhaustive srudy of that problem, and finally arrive at the result that our will is free, we still will nor know whether our will is indeed free or our mind is of such a nature that we have to find our will to be free, although it is not, in facr, free. This became one of my reasons for studying not philosophy but the natu.ral sciences. Thinki_ng about it now, 50+ years later, it still seems to me rhar I was right." Bur does nor the possibility that we are caused ro believe in false conclusions apply also to conclusions in the natural sciences? And ro the verdict of 50+ years later? I published my original essay after thinking about Newcomb's prob lem inrerminendy for five years. ln that essay l expressed the hope rhar Copyrighted material 84 -t> CHOICE AND UTILITY
someone would come forth with a solution to the problem that would enable me to stop returning to it. It is not surprising that no one did, yet it is surprising (to me) that the mere act of publishing Newcomb's problem, and sending my thoughtS on it into the world, rid me of it. That is, I was rid of it until the problem was presented in Scientific American and l was invited to read more than 650 pages of letters about it. Unfortunately the letters do not, in my opinion, lay the problem to rest. And they have started me thinking about it again! You can't win. Copyrighted material 4 INTERPERSONAL UTILITY THEORY How might the topic of interpersonal comparisons of utility be brought within the domain of positive economic science as part of a testable and disconfirmable empirical theory? This essay sketches how we could develop such a testable theory of interpersonal comparisons if we already possessed other theories. We do not currently possess these other theories but we do have their beginnings and it is reasonable to think they will be developed adequately in the future. This essay, then, describes only the last part of this route to interpersonal utility theory, the part that builds upon these other rheories yet to be developed. To show bow it is possible for there to be an empirical theory of interpersonal comparisons
leaves the nontrivial task of actually carrying through this project but, believing in the division of intellectual labor, I am content to leave that to economists. No satisfactory general procedure has thus far been devised for making interpersonal comparisons of utility, and many doubt its feasibility despite the fact that we make such comparisons every day on an ad hoc and intuitive basis. It would be surprising if our ordinary (and often apparently obvious) judgments were completely built on sand. Yet even the general accuracy of most of our intuitive judgments of interpersonal comparisons does not guarantee the existence of a simple systematic Copyrighted material 86 -& CH OICE AND UTILITY procedure that would explicitly yield these judgments. We seek an explicit systematic procedure tb.at would match those intuitive judgments (or at least delineate those distorting factors which make our judgments differ from the more accurate ones r ielded by the procedure) and so sometimes serve as a substitute for intuition, a surrogate. However, this systematic procedure need not replicate the intuitive process we follow; our aim is not an accurate psychological theory of how the judgments get made but a procedure that, by whatever process, yields those
judgments. The central difliculry in devising a systematic procedure is rhar interpersonal differences in uriliry (such as the fact that you want something, or prefer one thing to another, more than I do) do not have direct empirical effects, unmcdiatcd by people's judgments of them. Such interpersonal differences do not by themselves cause any phenomena, or explain any phenomena, e.xcept insofar as these differences are recognized by people who act on their discernments. (No single interpersonal difference in uti lity has a direct effect in the world, but combinations of them might. If I want something more than you do and hours of labor have the same disuriljry for each of us, then I will (be willing to) work for it longer than you do. Hence, one interpersonal difference, that I want somerhing more than you do, has an effect relative to another interpersonal comparison, that labor has the same disutility for us. However, there ~'<'ems to be no way to separately fix one such interpersonal fact so as ro thereby isolate the effect of others, nor is there any obvious way ro amass many such joint effects of conjunctions of interpersonal comparisons so as to "triangulate" and 6x one single interpersonal comparison.) For example, an explanation of certain redistributive activities
might involve people acting on their judgments of interpersonal comparisons (in accordance with a norm which they accept mandating redistribution) while the explanation of their making these judgments might be that they are true; the facts are that way and are recognized to be so. In this way, an interpersonal difference in utility might have effects in the world. These effects, however, do not help us very much, since they are mediated by people's unsystematic and intuitive judgments. A procedure utilizing effects based upon judgments seems shaky. Yet ir might be that such intuitive judgments concatenate together in a manner from which system can be extracted, system which can be used ro regiment such judgments, correct them and even test them. The analogy is inexact, bur compare how preferences, including preferences among probabilCopyrighted material Interpersonal Utility Theory ~ 87 ity mixtures, hang together so that the conditions de.scribing their con catenation suffice to yield the existence of a real-valued utility function unique up to a positive linear transformation. From no one particular preference could such information be extracted, only from the pattern they form together. In unity there is strength.
However, even one such procedure, with interpersonal comparisons, will leave us doubtful of irs va lidity. What we need to do, instead, is to tria11gttlate interpersonal comparisons, to formulate independent procedures, each with itS own virtues as a candidate for grasping interpersonal comparisons. 1i these independently motivated procedures converge to the same results, to the same judgmentS of interpersonal comparison, then each procedure will receive additional support and confirmation from the other. Moreover, in concert the procedures may be testable, and falsifiable. One procedure can be used to spec ify aspects of the utility functions of two people, thereby constraining the range of data compatible with the other procedure. Together, then, the procedures can constitute a theory with empirical conse quences and predictive power, hence a testable and falsifiable theory. If the conjoint theory passes such a test, it will receive empirical support. To be sure, like any other theory it will not be thereby proven to be true, but we justifiably will gain increased confidence in itS accuracy. A procedure for making interpersonal comparisons of utility will bring together and calibrate the separate urilit) scales of different individuals. Supposing that we have individual measurements of utility for persons I and U on a Von Neumann-Morgenstern interval scale, a
procedure for interpersonal comparisons will fix a common zero point and unit or will fix two common points. It wiU equate the utility of x for person number I (u1(x)) with the utility of z for person number ll and it wiU equate the utility of y for person number I with the utility of w for person number II. Suppose that each of those individuals satisfies the Archimedean axiom, so that if a is preferred to b is preferred to c then there is a probability p(O < p < /) such that the person is indifferent between b for sure and the option of a with probability p and c with probability (1p). Then, given the fixing of two interpersonal points between them, we can bring together the rest of their utility scales. Suppose person I is indiffer ent between r and a probability Pl of x and (1 - p 1) of y. Then for any alternatives.< oo person U's utility scale, u 1(r) = 11 11 (s) if and only if person Il is indifferent between sand a probability p1 of 1: and (l - p1) Copyrighted material 88 -~> CHOICE AND UTILITY of w. (It is obvious how to proceed with alternatives n01 preferentially between the two original interpersonal corupadson points.) The procedure for making interpersonal comparisons can thus calibrate the two scales together, utilizing the Archimedean axiom.
However, its ability to do so does not show its inrerpersonal comparisons are correct. Provided each individual satisfies the Von NeumannMorgenstern or similar axioms for the existence of uti lity on an interval scale, the procedure of interpersonal comparisons will not notice anything awry, however it fixes its original points of comparison. For all other points on the scales are equated by their relationships tO tpe two original points and so these later equatings will systematically show the same mistakes, if any, which infected the original comparisons. If "really" u1(x) is double that of u11 (z) and u,(y) is double that of u11(w) then the u1(r) that the procedure equated to tt11 (s) will really be double it. I have here used an example of a ratio distortion but all we know really, solely from the procedure's going smoothly, is that its equatings either are accurate or are systematically distorted in accordance with some linear transformation. To be sure, whatever intuitive reason we had for thinking the procedure plausible will be reason for thinking there is no such systematic distortion. But nevertheless the worry haunts proposals about interpersonal comparisons that although it equates two scales, one is really double the other, or 5 times the other (or more generally, some positive linear transformation of the other). If
one was the other writ large, but maintaining the same internal proportions, how could we tell the difference? Previous proposals about interpersonal comparisons, in addition tO their own intrinsic defects, have all stumbled on this structural difficulty. This difficulty can be avoided when we have two distinct independently motivated and independently plausible procedures, P1 amd P2 , for making interpersonal comparisons. The conjunction of rhese two proposals constitutes a testable theory. When procedure P1 is used to equate the two individuals' scales, this establishes predictions about what equatings between the two scales will be made by the operation of the independent procedure P2. When these predictions are borne out, when the two procedures do mesb in this way, then (the rationale and intuitive power of) each supports the other. The predictions are empirical and need not be borne oul. If they are nor then the conjunction of the two proposals is disconfumed-they cannot both be right. But if the predictions are confirmed, this, while it does not prove the joint accuracy of the procedures, will legitimately give us in Copyrighted material Interpersonal Utility Theory -& 89 creased confidence that some real phenomenon is being described and
delineated. Bur don't we face the same structural difficulty as before? Mighm't the procedures still yield not accurate equatings but ratio (or positive linear) distortions? With procedure Pt there is the possibiliry of such a distorting, and there also is the possibility that procedure P2 distorts. However, when the two procedures are independent, it is unlikely that they wi ll distort by exactly the same ratio or positive linear transform. Perhaps person D's utility is rea lly double what procedure Pt says, and perhaps person IT's utility is really 9 times what procedure Pz says. But it would be too much of a coincidence if these two independent procedures, while failing to be accurate, each failed in the same way, each happened to e,quate their separate units for the two people in the same m,israken ratio (or positive linear) relationship. Wheo the two independent procedures mesh in their result (and so in their predictions) we need some explanation of this agreement. One explanation is that the two theories are giving accurate renditions of the phenomenon that is out there and so agree. Another explanation is that both are inaccurate and just happen, by a coincidence, to distort by exactly the same factor. However, this "coincidence" will be
implausible, especially if the two procedures differ greatly in their intuitive rationale and angle of approach. Third, it might be said that there is a "universal distorting force" which distorts aU (otherwise accurate) procedures in exactly the same way. Such issues are familiar in discussions of the special theory of relativiry and the extent to which conventional elementS enter there. Economists need not worry unduly, I think, if their theory is on the same footing as other well-established parts of empirical science, facing the same philosophical possibilities of universal distorting forces. Finally, failing alternative explanations of the meshing of the two procedures, their independence might be placed into question; whether this is plausible will depend on the details of those procedures, The structural problem of a procedure's being distorted is avoided by a conjunctive theory which proposes more than one procedure. (Ideally, it would be bener to have more than just two independent procedures.) But if this avoidance strategy is to work, ir must yield success early. If over the years researchers investigate many procedures and find tbat two mesh in their predictions, we cannot then conclude that they are accurate. For it is not unlikely, in that case of investigating many
procedures, that we will encounter two inaccurate ones which, coincidentally, happen to distort by the same ratio. When large numbers of procedures Copyrighted material 90 -& CHOICE AND UTILITY are investigated, coincidence becomes a perfectly plausible explanation of the agreement of some two of them. Therefore, if our strategy is to work, we need to investigate accurate procedures early. The strategy, l think, is a promising one but the two procedures I propose later may well be inadequate (either not plausible or not independently mo tivated). I cannot, though, offer a general genial invitation to others to join in the effort and make their own proposals about procedures. You will understand if I invite only those with good ideas and good i.oruitions about interpersonal comparison procedures to make proposals so that the number investigated will not become too large too soon. (Once early success is achieved other, sleeker procedures, agreeing with the first established ones, can enter the field. It would raise interesting questions, of course, i.f two groups of proposals were found, each group's members meshing but the two groups differing.) I have said the separate procedures, jointly testable, would be independently
motivated. What sort of motivation can such a procedure have? It can yield some of the particular judgments of interpersonal comparison we make with confidence. Also it can yield some of the summary generalizations and maxims we have confidence in as holding ceteris paribus. For example, just about everyone finds it intuitive and plausible that a given additional amount of money will have less utility for a rich person than for a poor one. (To repeat, wben all other things are equal; the amount might enable some particularly important threshold to be passed by the rich person, while not doing anything comparably important for the poor person.) U a given procedure for making interpersonal comparisons yields this as a consequence, without simply building it in as an ad hoc feature, then the procedure will gain credibility. Yielding that intuitive result, one we have much confidence in, is from the point of view of evidential support like yielding some observable consequences. A theory that yields what any correct theory in that area must yield, whether this be observable consequences or intuitively correct generalizations, thereby reveals its mettle. 111 addition, there are some metaconditions on proposals. For
example, a proposal about interpersonal comparisons should, wben applicable, be plausible as a proposal about iotrapersonal intertemporal comparisons of utility. It should be plausible that it yields correct utility comparisons for tbe same person at different times (when it applies to yield any comparisons of that sort). A proposal that obviously feU down for one person over time could not be correct for two people. This Copyrighted material Interpersonal Utility Theory -i> 91 metacondition is nor empty, since we may have firmer intuitions about intrapcrsonal intertempo ral comparisons than about interpersonal ones, and so we may filter out iJl this way some proposals about interpersonal compansons. My suggestion of how to proceed on the topic of interpersonal comparisons, then, is to formulate plausible and independently motivated procedures and check to see whether they converge to (approximately) the same results (and are plausible, when applicable, for intrapersonal intertemporal comparisons). When these independent proposals converge, bringing along their own supporting strutS, then it will be rational to believe there is a real phenomenon "out there" they are delineating and demarcating. (Compare the situation in logic
where three separate important notions which arose independently, general recursiveness, lambda-definability, and computability, turned out tO delintit exactly the same functions.) What otherwise explains the coovergence? This seems a sensible way to proceed on the topic, unbefuddled by the philosophical problem of "other minds." This "other minds" problem also would apply t.o specifying a utility function for another person, apart from any issues about interpersonal comparisons. Though there is behavior, how do we know a prc(erem;c is exhibited, or that a choice is made, or that the person is aware of the different alternatives available, or how the person cor~ceives of the alternative selected, or what probability judgmer~tcS the person makes, and so on. Sinc.e the procedure for interpersonal comparisons is to be a systematized su.rrogate for our ordinary inruirive judgments, not a solution to the philosophical problem of other roiuds, it is permissible for that procedure to utilize psychological predicates (not involving interpersonal comparisons) which apply to other people. It would be real progress, even, to somehow get from the existence claim that some procedure or other for interpersonal comparisons
exists to a panicular determinate procedure. Must our procedures avoid other "circles," though? Would it be aU right to use a procedure whose very operation and application involved utilizing panicular inru. irive judgments of interpersonal comparisons? Nor if the purpose is to justify making interpersonal comparisons, to show that there can be correct ones, but it would be permissible if the purpose is to find an efficient surrogate for the mess of our intuitive judgmentS. A procedure that utilized only a few such intuitive judgments might constitute a great advance in systematization. Intermediate purposes will raise more delicate questions about the legitimacy of circularities. Copyrighted material 92 & CHOICE AND UTILITY The triangulation strategy can yield genuine progress and illumination, even though it does not prove its hypotheses, by making "interper sonal comparisons" a topic within empirical social science. Within this framework, the difficulties which remain have a familiar character; they are the ones attendant upon any endeavor to fit a theory to data, thereby tentatively supporting the theory. After procedures are found which can be shown to work, there will be the task of discovering the deeper explanation
of why they, in particular, do so. Utility, as we are concerned with it here, is a measure of how much someone wants/desires a particular situation w obtain or continue. (More accurately, it is concerned with the magnitude of the desire/want as expressed in choice and other behavior, verbal reportS of preference, etc. These may reveal, as the psychoanalysts tell us, only a porrion of the magnitude of what's there in the unconscious yet reveals itself more deviously in compromise formations. It would be a very different sort of theory than utility theory as currently investigated that could get to desires partia lly masked in these and other ways.) The utility we discuss here is not a measure of the resulting level of want-satisfaction or felt well-being. If I want an additional $1,000 exactly as much as David Rockefeller wants an additional $55,000, this does not mean that when each of us gains our desire we end up at an equal level of overall satisfaction or felt well-being or wealth. We are equally satisfied by gaining our wish but we need not be equally satisfied in result of gain ing our wish. The increment we each gain is equal but the resulting level we attain will not be equal if we do not start from equal levels to begin
with. Interpersonal comparisons of incremems rather than levels may not be useful for purposes of redistribution according to some theories. It will serve, however, for retribution where, roughly, we compare bow much this victim diswants the change inAicted on him with how much this wrongdoer diswants the change involved in a candidate punishment. (For more derails, including discounting punishment by the wrongdoer's degree of responsibility, see Nozick, 1981, chap. 5, part Ill.) We don't try to pur the wrongdoer at the same absolute level as where his victim ends up. A person who does mild harm to an already miserable person does not deserve ro end up as totally miserable as his victim, nor does someone who mildly injures someone much happier than himself thereby deserve to have his own level raised significantly to become on a par with his victim's. It is sometimes said that in meting out retributive punishment we are "playing God"; however, our retCopyrighted material lnt<rpcrsonal Utility Theory ~ 93 ribution is designed only to achieve certain incremental changes while divine justice, if it existed, presumably would focus upon a person's re
suiting absolute level. (Differeot kinds of interpersonal comparisons are distinguished in Sen, 1979.) Zero Point Utility functions measured on an interval scale, such as those of Von Neumann and Morgenstern, are unique up to a positive tinear transformation. Like the familiar Fahrenheit and Centigrade scales of tern perature, they have an arbitrary zero point and an arbitrary unit. (In the familiar formula, Fahrenheit equals (9/5) Centigrade + 32; the 915 scales from one arbitrary unit to another, while the +32 moves from one arbiuary zero point to another.) The problem of interpersonal comparisons is to specify, for two persons' utility functions which are given, a com mon unit and a common zero point, to calibrate their utility functions so that the unit on each represents the same difference of utility for each, and the zero point on each represents the same degree of wanting. Most people, learning the utility lingo, fall quite naturally into speaking of positive and negative utility, although this has no basis within the Von Neumann- Morgenstern mode of measurement, where the zero point might fall anywhere. It seems plausible that things (situations, states of affairs, etc.) that are wanted or desired have positive utility for the person while things that are diswanted, that are desired-not,
have negative utility for the person. Zero utility, then, would apply to things, situations, changes that are neither wanted nor diswanted, to things the person is indifferent about. My preference would be to specify zero utility directly in terms of in difference. While for some purposes it may be more desirable to proceed with a zero defined in this way, I am not sure it is necessary or appropri ate here. Indeed, since we are not attempting interpersonal comparisons of resulting utility levels perhaps the issue of the zero can be avoided altogether and we need only pursue the task of finding a common unit. (Some economists, I know, will think me quixotic or confused in focussing any attention upon the zero point when level comparisons are not at stake. I prefer to tbink I persist beca use there is some presupposition of tbe economist's approach which I am denying, but my case would be stronger, I admit, if I could explicitly identify that presupposition which, once denied, necessitates my approach.) Therefore, readers Copyrighted material 94 -~> CHOICE AND UTILITY who object to the words of this section can minimize their irritation by proceeding directly to the next section on the w1it. Since we are not attempting to make interpersonal comparisons
of utility le,els but rather of utility differences or units, the fixing of the zero point may appear, if it is necessary at all, trivia l. Economists may simply want to say that something has zero utility if it has zero marginal utility. (Is there room, though, for the question of whether this provides an interperso11al zero? Must what has zero marginal uti lity for you be wanted the same by you as 1 want what has zero marginal uti lity for me?) My preference, I have said, is to specify zero utility direcdy in terms of the notion of being indifferent about something, neither wanting it nor diswaming it. To be indifferent about an altemati,rc x is to be indifferent between x and not-x. Something x bas zero utility for a person if and only if x lnor-x; if and only if be is u1different between having x and not, between x's occurring and not. It makes no difference to the person whether x or notx, no difference he cares anything about. Similarly, we can understand x's having positive utility for a person as the person's preferring x tO not-x; x has negative utility for the person if he prefers not-x to x. IJ1 one way this is more general than the economist's usual way
of proceeding, in another less general. We are not restricted, as with marginal utility, only to repeatable units; however, in the notation here, the variables x, y, . . . range not over total states but over partial ones, or aspects of total states. Thus, when we speak of preferences between (partial) alternatives, these are conditional on implicit things. It is a mistake, I think, to believe one bas actually specified everything the preference depends upon; for each purported full description one can always find further factors, left out, whose variation affects the. preference. This is no reason not to make these even more explicit bur it is a reason not to use a theoretical structure which supposes that everything relevant already bas been made explicit. It seems plausible to use indifference between x and not-x as a cri terion of x's having zero utility. But this use, even in the case of one person, depends upon a further empirical fact. It must also be the case tha t: If a person is indifferent between x and not-x, and that person also is indifferent between y and noty, then the person is indifferent between x and y. Copyrighted material Interpersonal Utility Theory ~ 95 (And also indifferent between x and not-y, and also between notx
andy. However, since indifference is an equivalence relation, we needn't state these as additions.) There has to be inclifference all around. For if the person preferred x to y, the utility function, assigning zero co everything, would nor mirror preference. The above indence.d condition is an empirical precondition for using indifference between something's obtaining and not in order to specify a zero point on the person's utiliry sca le. (Sometimes, though, it's not being satisfied will indicate that the alternatives were illegitimately specified, that they were contrary but not contradictory alternatives, relative but not absolute complements.) Proceeding in this way, there is room to raise the question of whether this procedure serves co fix: a common zero point for the two people. If one person is generally enthusiastic or even manic, mightn't his being indifferent between x and not-x represent a more positive state of wanting each than is represented by a depressed person's attitude toward y when he is indifferent between y and not-y? However, by uriliry we refer ro strength of want, not to hedonic tone. Even if everything about life feels better to a manic person, an alternative he's indifferent about when manic isn't wanted more by him than an alternative
someone else is indifferent about when depressed. What may be true, though, is that a depressed person has far more indifferences than average; getting depressed is a process in which many preferences are replaced by in differences. Unit: Bargaining Facts of interpersonal differences in utility, I have said, do not have direct empirical consequences unmediated by people's discemings of those differences. I mean interpersonal differences as embodied in different magnitudes on the same side of the zero point. Of course, if x has positive utiliry for me but negative utiliry for you, then this difference alone, unmediated by people's discerning of the differences, may give rise to the fact that I seek x while you shun it. To establish an interpersonal unit, we turn to consider consequences of interpersonal differences mediated by people's (apparent) discern ing of these. One way would be simply to survey people's opinions about particular comparisons but, given the low motivation people will have to provide accurate (and checkable) answers, such survey material would be most unreliable; moreover, irs panicular results would not
find a place in any systematic context. Far better to examine people's Copyrighted material 96 ..., CHOICE AND UTILITY behavior, what we might call their revealed beliefs, about interpersonal comparisons. Some will be shown in situations involving redistributive activities bur these activities will be a compound product of normative beliefs, actual interpersonal comparisons and self-interested disrorrions of these which feed into the normative beliefs ro one's own benefit. Though a positive theory someday will need to be offered of people's normative behavior, that area is far roo undeveloped ro build upon even programatically now. It will be more fruitful to consider behavior which, though mediated by people's judgments of interpersonal comparisons, is nor also mediated through their normative beliefs. We therefore turn to an area already lying within positive economic science, for which (positive) economists already have proposed explanatory theories: bargaining and threat situations. When two of us stand ro gain by reaching a mutual agreement or bargain, as compared tO the situation of our nor agreeing, yet various different agreements can be reached about which we have divergent preferences, what
determines which (if any) agreement will be reached? What determines which party will get more of what he wants? A particular case of this is the one of bilateral monopoly; between the highest price one party is willing to pay and the lowest the other is willing to accept lie a range of mutually beneficial agreements, about which the parries have conflicting preferences. It is well known that we lack a fully adequate theory, for these cases, of where on the contract curve a bargain will be struck. The formulated theories of bargaining mention many faccors which might enter here, such as bargaining skill. It seems plausible that one factor will play an especially crucial role, namely, which party will be hurt more by no agreement. And here, as before, we do not refer to absolute levels of illfare when there is no agreement bur rather increment in (what may be already existing) hurt due ro nonagreement. The question is: who diswants no agreement more? (I am aware that nor all theories of bargaining include such an interpersonal comparison, for example the theory (following Zeuthen) of John Harsanyi. Others already have criticized its mechanism of concessions. We might note that, within such a framework, it would be more plausible to develop a stochastic
theory, where the parries' maximum tolerable probabilities of no agreemenr do not determine which one concedes, but only 6x their respective probabilities of making the next concession.) On our view here, bargaining (and not merely arbitration) theories will utilize inrerpersonal comparisons of utility. The parry that stands to lose more if there is failure to reach agreement is (particularly) vulCopyrighted material Interpersonal Utility Theory ~ 97 nerable to a threat by the other to leave. Here lies a great incentive for the more vulnerable party, if not to accept the other's most preferred terms, at least to make some concession, to be the one to give a little in the process of negotiations. George Homans, following others, incorpo rates within his theory of social exchange the principle of least interest: That person is able to dictate the conditions of association whose interest in the continuation of the affair is least; the party to the exchange who getS the lesser reward from it is less likely to change his behavior in favor of the other than is the party who gets the greater; the person who is perceived by the other as the less intere,sred, the more indifferent to the exchange, is apt to have the greater power (Homans, 1974, pp.
73-74). What is the appropriate baseline from which to measure each person's gains from the agreement? The theories of bargaining differ on th.is point. In Nash's model, followed by Harsanyi, this baseline is the noagreement situation wherein both parties carry out their threats. Others have suggested the baseline is what the parties would receive if they treated the situation as a noncooperative game and maximized their security level, employing their minimax strategy. Or should the (shHting) baseline be the utility one party would receive from the other party's last offer? It is not our purpose to present a specific theory of bargaining situations here. We shall proceed as though one such is available, presupposing interpersonal comparisons; but our purposes here would equally well be served if a theory utilized different baselines for different types of bargaining situations. Let us imagine rhe foUowing, described vaguely. Each offer by one parry establishes a new status quo point for the other party, replacing (the previous one offered which replaced ... ) the original baseline point (whether no agreement at all or the security level). The cost to the other party of breaking off the negotiations and leaving is now what he would have received under this most
recent offer. If no agreement is reached, each of the parties stands to lose the differ ence between the last best offer rbey received and the baset;ne situation. The party that stand~ to lose the most makes the next conce.ssion. And he must concede enough to transform the situation into one where the other parry stands to lose the most; otherwise the first one simply will have tO make another concession immediately. Whether he makes the least concession sufficient to put the next move up to the other party, or he makes a la rger concession in an attempt to reach agreement right then, may well depend upon other factors. (Notice that, starting with Copyrighted material 98 ~ CHOICE AND UTILITY the stah<s quo point, this theory holds that the one worst off under this, or forgoing the most under this, will make the first concession, ooe sizable enough to pass the new vulnerability tO threat over to the other party. Compare how the Zeuthen-Harsanyi theory makes the identity of the concession maker dependent on who is willing to bear the greater probability of no agreement, where the concession is of a magnitude to shift this burden.) Be the details as they may, the important point for us is that the
process of bargaining, and the fina l result, will be importantly determined by the interpersonal comparisons the parries make. Each considers who gets hurt more, the other party or himself, and ac.ts accordingly. Notice that each wants it to be the case that the other parry will be burr more, so when one parry does make the concession, this reflects a particular interpersonal comparison he makes, and concedes, only reluctantly. It is in his interests not to acknowledge the fact that he is the one who will be hurt more, as it also is in his interests, if he is the one wbo will be hurt more by no agreement, to make a concession in order to reach agreement. So when he does make this concession, when both partie.~ know he must make rhe concession, then both are agreeing in the interpersonal comparison they make. One happily and the other sorrowfully, they both agree which one will be hurt more. What accounts for this agreement, so reluctantly wrung from one and acknowledged by him? One possible explanation is that he is the one who will be hurt more, and both discern this fact, and also the other's discernment of their knowledge, ere. To be sure, the other parry may be a skillful faker, dissembler, actor, or liar, yet the first party has much
incentive to see through this. Often, when both believe one will suffer more, and that one concedes, this is because they both know it is true a.nd act accordingly on their knowledge. Might not who concedes depend, nor upon who will suffer the greater increment of hurt if there is no agreement, bur upon who will suffer the greatest relative hurt as indicated by the ratio of hurtincrement to some measure of overall well-being? Might nor the conceder be be who suffers the greatest percentage hurr, so that the bargaining siruation cannot provide us with an interpersonal comparison of increment in utility? (I am indebted to Benjamin Friedman for raising these questions.) The intuition is that the percentage the hurt represents is what is important and will guide behavior. But Von NeumannMorgenstem utility is a measure of preference, of how much someone wants or cares about something, nor of intensity of pleasurable sensaCopyrighted material Interpersonal Utility Theory ~ 99 tions. Perhaps it is a certain ratio that people do care about in their choices. If so, that is what is rclle.cted in their utility functions and that is what is to be compared interpersonally. Such interpersonal comparisons of Von Neumann-Morgenstern utility are not designed to
capture some notion of "equivalent well-being" but only of "equally wanting." (A similar point applies ro a ratio-relativized treatment, rather than an absolute one, of the vertical axis in the utility curve for money.) We are encouraged by the shared judgment of interpersonal differences by the participants in the bargaining situation, especially since it arises in a context devoid of normative desires to redistribute. The reluctant concessions are the solidest rock on which to build. But can we make something of this shared judgment, can we elaborate it into a systematic procedure for making interpersonal comparisons? One procedure would involve overlap in separate bargaining situations. Two people, 8 and C, bargain separately with A, who is in the same situation both times with respect to his own potential gains and losses. If A concedes to 8 but does not concede to C (who instead concedes to him), then we can tentatively conclude that A would be hurt more than 8 would if no agreement were rc.ached in their bargaining situation, while A would be hurt less than C would if no agreement were reached in their bargaining situation. We thereby can conclude that 8 is hurt less in the first situation than C is in the second, even though 8 and C do not interact together. We can test this result by putting 8 and C in a joint bargaining situation together where their stakes are
exactly what they were in the previous separate situations with A. If the prediction that C will concede to 8 is borne out, then this provides reinforcing confirmation of that earlier interpersonal comparison. Tbis theory wiU utilize situations where the party's relationships are impersonal enough so that A can be in the very same situation, with respect to his own potential gains and losses, with 8 and C. Moreover, A in interacting with 8 doesn't know or take into account the fact that C will come. (His likelihood of external alteroatives is the same in the A 8 and the AC situations.) Just as these factors have to be controlled, so migbt others. Ideally, we would want to utilize a bargaining theory which rook account of all factors affecting bargaining behavior so that our mini-test bargaining situations (between A and 8, and A and C) would hold constant all relevant factors other than who gets hurt more by no agreement, so that this alone cou.ld be isolated and detennined. Alternatively, with a full and explicit theory, the other parameters might Copyrighted material tOO -&< CHOICE AND UTILITY be allowed to vary in known ways while still one could determine which
party gets burt more by no agreement. By chaining together people in a multiple network of bargaining, we may hope to determine (and test) an interpersonal fixing of the unit. Notice that this chaining together of different people in a network of bargaining situations need not provide sufficient information for full and general measurement of utility. That can be done by a standard Von Noumann-Morgenstem procedure. Given this individual utility measurement, our network of bargaining situations need only bring the different individuals into mutual interpersonal comparison. Unit: Utility of Money The discussion of bargaining describes how interpersonal comparison points might be extracted from people's behavior in bargaining situations, from their actions based upon their (implicit) judgment of interpersonal comparisons as they act self-interestedly. It would be desirable to have a procedure that directly establishes interpersonal comparisons, without going through people's judgments of such comparisons, even when the judgments are implicit and self-interested. (Such comparisons that do not go through people's judgment, we have said, are not needed for positive explanatory purposes.) Supposing interpersonal comparisons are possible, what would
we expect a comparison procedure to look like? It is not likely that some particular item, say oreo ice cream, will have the same utility for everyone and thereby demarcate an interpersonal comparison. What we should expect, instead, is that some structural fact will pick out the same utility for everyone, some salient structural feature of a utility curve. Again, it seems hardly likely that this will be a structural feature of the utility curve for a particular item, such as oreo ice cream, for it seems likely that here people will show the widest variation. The more general the item, the more widespread it is in a common role in people's lives and the less specific its purposes, the more promising it is as a candidate to exbibit structural features that will serve to peg interpersonal comparisons. If a general procedure for making direct interpersonal comparisons is possible, it will be based upon salient structural features of the utility curves for something thusly general. For example, money. It is to the utility curve for money (or someplace similar), if anywhere, that we should look to base our direct theory. Copyrighted material Interpersonal Utility Theory ~ 101 The St. Petersburg paradox and its generalizations led theorists to assert that the utility of money is not linear with its amount or
with the log of its amount and there is a finite upper bound to the utility of (any amount of) money. However, theorists have proposed different hypotheses about the shape of the utility curve for money. In 1948, M. Friedman and L. ]. Savage considered how the shape of the utility curve for money was related to willingness tO undergo risk, in particular how the convexity or concavity of the curve was related co willingness to gamble, or co buy insurance. Let money or wealth be measured along the horizontal axis, and the utility of the amount be measured, by a Von Neumann-Morgenstern procedure, along the vertical axis. A gamble between two amounts of money M 1 and M2 is a prospect of receiving M1 with probability p and Mz with probability (1- p). When amounts Mt and M2 have utilities u(Mt) and u(M2), we graph these factS as points a and b. A gamble between M1 and M2 will be located somewhere along the straight line between a and b. (See Figure 4.1.) Its precise location will depend upon the particular probability p (of M1) i.nvolved in the gamble. Note that it is not a trivial fact that the gamble falls along the straight line. The expected monetary
value of the gamble, p x M1 + (1 - p) x Mz, will fix its xcoordinate and its vertical location places it along the straight line because, within the Von Neumann-Morgenstern procedure for measuring utility, the utility of a gamble is its expected utility. Thus, the utility of this gamble is p x u(Mt) + (1 - p) x u(Mz). u(M) u(M,) u(M1) Figure 4.1 M, M2 M, M We now can ask which the person would prefer, the gamble or receiving (for sure) the expected monetary value of the gamble. Which does the person prefer (or is he indifferent): the gamble of receiving Mt with probability p and Mz with probability (1 - p), or definitely receiving an Copyrighted material 102 -<> CHOICE AND UTILITY amount of money MJ which is equal to p x M1 + ( 1 - p) x M2? If the person prefers the expected monetary amount M3 for sure to the gamble, then M3 has a higher utility to him than the gamble, and therefore the util ity of MJ is above the straight line connecting points a and b, for instance at poim c. If, however, the person prefers the gamble to M3 for sure, then the utility of amount M3 is below the straight line
connecting a and b, for instance at point d. Therefore a person will not gamble if or where his utility curve for money (around his present income) is convex from below, and so above the straight line connecting tbe endpoints. However, preferring sure things to gambles, he will insure within this range. lf the utility curve (around his present income) is concave from below, and so below the straight line connecting the endpoints, the person will not insure but will gamble. Friedman and Savage (1948) proposed that the uti lity curve for money was first convex, then concave, and then convex again, with two inflection points (see Figure 4.2). According to their hypothesis, if the person's present income is within the first convexity, at M1 say, he will not gamble, and similarly for tbe last convexity at M3, while if it is at M2, within the concavity, be will gamble rather than insure. u(M) Figure 4.1 M, M Soon afrerwards, Harry Markowitz (1952) proposed a modification of the Friedman-Savage hypothesis. Markowitz pointed out that individuals at all income levels may choose both to gamble and to insure. Moreover, according to the Friedman-Savage hypothe.sis, people
whose cunent income is in the middle concave range should like large symmetCopyrighted material Interpersonal Utility Theory ~ 103 ric bets giving them a 112 chance of g.reat wealth and a 112 chance of poveny; and also people "almost" at the height of the rightmost convexity should like a fair bet that either moves him to that or (if he loses) lowers him to the height of the first convexity. To avoid these FriedmanSavage consequences, neither of which is observed, Markowitz proposes another curve. The utility function for levels of wealth above current wealth is first concave and then convex. The utility function going from present wealth downward is fint convex, then concave. Putting these rwo together, we get the Markowitz hypothesis (Figure 4.3). There are three inflection points, with the middle inflection point being present (or "customary") wealth. The function is concave immediately above pre.sent wealth and convex immediately below it. (Markowitz also suggests that people avoid symmetric bets, so that the curve will fall faster to the left of the origin than it rises to the right.) u(\V) Figure 4.3 So things stood for many years, with the Friedman-Savage and Markowitz hypotheses receiving major attention and minor wrinkles.
Recently rwo psychologists, Amos Tversky and Daniel Kahneman (1979), made yet another proposal, where the origin is a neutral reference point. (T versky and Kahneman also present hypotheses about the "framing" of a decision, to account for where that neutral reference point will be placed.) Their curve is S-shaped, concave about the reference point and convex below it, with response to losses more extreme tl1an response to gains. Notice that each of these proposals concerns the general shape of the curve. It is not suggested that we each exhibit exactly the same curve, with dollar amounts filled in. Ten thousand dollars may be worth more to you than to David Rockefeller. What has been suggested is that different people's utility curves for money have the same general structure, though they will differ at which amounts along the xaxis the inflection points come. No doubt, other theorists will propose curves in Copyrighted material 104 _., CHOICE AND UTILITY addition ro these, and we may hope someday ro possess an adequate derailed theory of the contours of the uti lity curve for money. Perhaps it will be discovered that nor everyone fits the same-shaped curve, or rhat the shapes change during the life cycle. I leave these complications aside
for now. To proceed ro a particular procedure for interpersonal comparisons, we would need to utilize a particular curve, accepted as descriptively adequate. We can, however, say how we wo11/d proceed, even though we lack a final theory. Our procedure, we hope, is robust enough to take account of furure advances in the fields of bargaining theory, utility theory and rhe uriliry curves for money. It is reasonable for our sketch of a program for interpersonal comparisons to build on the best extant theories, even while recognizing their inadequacies. When we know the details of fully adequate theories we then can check whether the program is indeed feasible. These remarks clearly apply ro our use of utility curves for money but also to our use of Von NeumannMorgenstem utility theory. If following a similar program with a mathematically restrictive non- Von Neumann- Morgenstcrn utility theory and its salient structural features led ro different and incompatible comparisons, then we would have ro check which best fit other features (intuitive judgments, general maxints, etc.). If both jibed equally well with different portions of the rest of our beliefs, we might conclude we had discovered two distinct real phenomena our there. Which each approach
demarcates would depend, of course, on the details of those two theo nes. If everyone does conform to the same-shaped Von NeumannMorgenstern uti lity curve for money, this is a striking fact. Though located differently, each person's curve will exhibit the same structural fearures. We may utilize some of the,se salient structural features as (proposed) points of interpersonal comparison. The measurement of utility for an individual is unique only up to a positive linear transformation. Therefore, only those strucrural features (of the utility curve for money) rhar are invariant under positive linear transformations are candidates for interpersonal comparison points. The rest aren't even starters in the contest, e.g., being an amount of money where rhe curve has a slope of 1/2. That amount will shift for one person, depending upon how we scale his utilitie.s. Only those structural points that are fixed stably for one person, being invariant under positive linear transformations, possess the stability needed to possibly mark points of interpersonal companson. Copyrighted material .. ----Interpersonal Utility Theory -t> lOS An especially salient feature of the curves proposed by
FriedmanSavage, and by Markowitz, are their inflection poims. These poims of inflection are invariant under positive linear transformations, and it seems plausible ro propose these as pegs for interpersonal comparisons. (Those more sophisticated mathematically than I might focus upon other features which are invariant under positive linear transformations, and consider their intuitive suitability as pegs for interpersonal comparisons.) Supposing the Friedman-Savage hypothesis, for the moment, we equate in utility that amount of money M1 where your curve first inflects with that possibly different amount Mz where my curve first inflects. The utility of M1 for you is the same as the utility of M2 for me. Similarly for the second inflection points N1 and N2; these too contribute a point of interpersonal comparison. The Markowitz hypothesis involves three inflection points, bur these are not all equally plausible as pegs for interpersonal w1it comparison. (His middle point represents current income, which we would not want to hold must represent the same utility level for everyone. If it represents instead the same difference in utility for different people, that difference will be zero.) Thus, we propose Markowitz's first and third inflection points as interpersonal comparison points. The Tversky-Kahneman proposals about the curve.s, thus far only roughly formulated, present no inflection points as
candidates for interper.sonal comparison pegs. lf further refinement of their proposal incorporates salient structural features, invariant under positive linear transformations, then these might be utilized in a proposal about interpersonal comparisons. Earlier we mentioned the possibility that the different-shaped curves might 6t different folks. Our proposal about interpersonal comparisons rhus far have concerned only intrashape.d comparisons, not intershaped comparison. When precise specifications of the different curves are available, perhaps inter.shape comparison poinrs (or mappings) then will suggest themselves; if curves vary also in a systematic way within the person's life cycle, then inrrapersonal intertemporal comparisons, if they could be made, might then be used to suggest interpersonal intershape comparison points. Some may think it not really plausible to pick inflection points on the utility curves for JDOney as interpersonal comparison points, even though these are salient structural features invariant under positive linear transformations and they are the type of peg there would be if there is any interpersonal comparison peg at all. For it may seem that the only thing the location of these points reflect is people's attitudes
toward risk. Copyrighted material 106 -t> CHOICE AND UTILITY But people's attitudes toward risking money may well depend upon the utility of that amount of money to them. Using (comparative) behavior in a risky situation to indicate the (comparative) utility of money is not in principle more objectionable than using one person's preferences among risky alternatives, probability mixtures, to indicate the strength of his preferences among simple noorisky alternatives. To be sure, the Von Neumann- Morgenstern procedures for individual utility measurement, if they are infected by attitudes toward risk, are infected by e,ach person's own attitudes aod presumably hold that sornewhar constant in their scaling of that individual. The interpersonal case is more hazardous. Yet tbe Von Neumann-Morgenstern procedure does not allow attitudes toward risk free leeway; certain attitudes are eliminated by the axiomatic conditions themselves, upon wbicb the utilit)' measurement is based. For example, the condition that if .t ly then px, (1- p)zlpy, (1 - p)z rules out giving special weight to possible failure to get anything. If a person were indifferent between $2,000 for sure and 1/2 {$8,000), 112 ($0) only because he's s.re to get something
good in the first case, this feature will be altered wben both alternatives are embedded in yet another gamble, for example 1/10 ($2,000), 9/10 ($0) vs. 1110 (1/2($8,000), 1/2 ($0)], 9/10 (SO), where now he's nor sure ro get something good on either alternative. The above Von Neumann-Morgenstern condition excludes preferences based on attirudes about risk insofar as the relevant risky feature or distinction can vary or disappear when the alternatives are uniformly embedded in yet wider gambles. T{ tbe Friedman-Savage and Markowitz curves are indeed Von Neurnann-Morgenstern utility curves for money, then many differing attitudes about risk already are excluded in the individual case and so cannot later arise to contaminate the interpersonal comparison. Nonetheless, one would welcome an adequate theory of individual choice that explicitly builds in parameters representing attitude toward risk. U thi.s were obtained, our proposal about interpersonal comparison pegs could conrrol for varying risk attitudes, as independently specified and measured in separate individual choice situations. Alternatively, one could allow these attitudes toward risk to vary yet utilize the theory to take full account of how this known variation affects where the inflection points on the utility curves for money will fall and introduce
interpersonal comparison pegs only after first correcting for any distort ion in the inflection point's position that was introduced by attitudes toward risk. Copyrighted material Interpersonal Utility Theory ~ 107 Utilizing inflection points of the utility curves for money seems especially weU swted (though nor guaranteed) to yield another intuitive con sequence. It should turn out that, ceteris paribus, for positive amounts of money or gains, the rightmost inflection points for wealthy people will occur at greater amounts of money than the structurally corre sponding inOection points for poor people. We should, that is, discover it to be a consequence of the proposal about interpersonal comparisons, that a given fixed amount of money will have less utility for a wealthy person (since his inflection points come higher up) than for a poor per son. Yielding this intuitive consequence supports the proposal about interpersonal comparisons whereas having the opposite consequence, or the one that there is no difference in the utility of money between the wealthy and the poor, would disconfirm a proposal about inter personal comparisons. (lf the rich and poor person's utility curves
for money crossed somewhere in the upper right quadrant, we then might have to refine our simply put intuitive maxim.) The philosopher David Gordon has suggested to me a further prediction, using both bargaining and monetary situations, namely, given a plausible measure of bargain ing power intrinsic to the structure of bargaining situations and their outcomes, that generally a poor person will have less bargaining power than a rich person-either than a rich person with whom he bargains or than a rich person would if he were in that situation. We now have before us, given the proposal about a common zero point, two routes towards a common unit: one based upon conces sions in bargaining situations and the other upon the utility curve for money. These routes are independent of each other; neither depends upon the rationale for, or the results of, the other. When both proposals are made, as stared earlier, their conjunction has empirical and testable consequences. When the bargaining procedure specifies as interpersonal comparison points P1 for the first person and P2 for the second, we can empirically deterrnine where the first person's inflection point M1 on the
utility curve for money is in relation to his point P1 Then we can check whether the second person's corresponding inflection point Mz stands in that very same relation to his own Pz. For example, using the zero point and the Archimedean condition we can discover what particular probability mixture of the outside alternatives, with probability r, is indifferent to the preferentia lly middle one as a sure thing. With that particular probability in hand, we can now predict the relationship of the second person's corresponding inflection point M2 tO the zero point Copyrighted material 108 ~ CHOICE AND UTILITY and his interpersonal comparison point P2 from the bargaining situation. For that second relationship will involve the very same probability r that was observed in the earlier case. If we usc the bargaining procedure more fully, along with standard Von Neumann-Morgenstern individual utility measurement, to fully calibrate e.ach individual's utility scales and bring them into comparison, then we can make a prediction, knowing one individual's inflection points on his utility curve for money, about the monetary amounts at which the other person's curve will inflect. The failure of this
prediction would refute the conjunction of the two proposals, while the prediction's success would confirm (though not definitely prove) the conjunctive theory. When they mesh, there is reason to believe both theories demarcate the same real external phenomenon. To be sure, there remains the theoretical possibility that though they each portray some real-world phenomenon, they do so in a distorted way and in the same distorted way, distorted by the same ratio or positive linear transformation. However, this will be an implausible coincidence if the two theories are independent in their motivation and angle of approach. This, then, is the appropriate time to mention one worry about the independence of the two particular procedures I have discussed. The money utility curve procedure is direct, not based upon anyone's judgments of interpersonal comparison (other than our own in proposing the procedure, but these judgments aren't utilized rvithin the procedure). The bargaining procedure is indirect, based upon people's concessions due to implicit and self-interested judgments of interpersonal comparison. Mightn't the bargainers themselves make these judgments as estimates of the first procedure's direct facts? If these bargainers choose to concede or not depending upon their (implicit) beliefs about each
other's utility curves for money and where they inflect, then the jibing of the two procedures will not indicate that ao independently existing phenomenon is demarcated. To be sure, the bargaining situations often will involve oor monetary amounts bur orher objects or joint activities and projects. (Who will be hurt more if we can't agree about what movie to go to see and instead go our separate ways?) But if these too are implicitly calibrated by being equated with monetary amounts, and judged by relationship to inflection points, then the indirect procedure will have been contaminated by our particular direct procedure and hence not be independent of it. Clearly, it would be desirable to have a conjoint theory of interpersonal comparisons which utilizes two direct independently motivated Copyrighted material Interpersonal Utility Theory -t> 109 and mutually supportive procedures. I have no other direct procedure to propose now. Perhaps the extension of economic analysis to topics outside its uadirional domain, such as education and family behavior, will uncover still other general utility curves displaying salient structural features suitable to pe.g interpersonal comparisons. This essay has not presented a specific procedure for making interpersonal
comparisons, but only sketched a program toward that goal. We now can see how it is possible to have an empirical theory of interpersonal comparisons. Others (in accordance with our earlier warning, perceptive others!) will have to formulate the necessary prerequisite theories of bargaining and the utility curves for money, fill in the details, propose still other independent procedures, and perform the empirical tests. When the topic of interpersonal comparisons is thereby brought within the domain of empirical theory, I hope it will please us all, equally. Copyrighted material 5 ON AUSTRIAN METHODOLOGY The major figures of the Austrian tradition in economic theory are Carl Menger and Frederick von Weiser, originators of marginal utility theory, Eugen von BOhm-Bawerk, and in this century Ludwig von Mises and the co-winner of the 1974 Nobel Prize in Economics, Frederick Hayek.1 A framework of methodological principles and a.ssumptions, which economists in other traditions either do not accept or do not adhere to, shapes and informs the substantive theory of Austrian economics. I shall focus on the most fundamental features of this framework, the principle of methodological individualism and the claim that economics is
an a priori science of human action, and upon two issues at the foundation of Austrian theory within this framework: the nature of preference and its relationship to action, and the basis of time-preference. I shall be forced to neglect the farthest reaches of the theory, for example, the Austrian theory of the business cycle, where still the fundamental methodological theses intertwine. I also shall leave untouched other illuminating distinctive emphases and approaches of Austrian theory, e.g., the constant awareness of and attention to processes occurring in and through time, the study of the coordination of actions and projects when information is decentralized, the realistic theory of competitive processes. Nor shall I be able to detail the intricate interconnections of the different Austrian themes. Copyrighted material On Austrian Methodology ~ lJ 1 I. Method.ological Individualism The methodological individualist claims that all true theories o.f social science are reducible tO theories of individual human action, plus boundary conditions specifying the conditions under which persons act.2 Methodological individualists are reductionists to the extent of
their claim that true theories of social science are reducible to theories of individual human action, but typically Austrians oppose other reductionist clai.ms, e.g., that theories of human action are reducible to neurophysiology, chemimy, and physics, or that social science is reducible to these in a way which bypasses human action. This raises the question, which I shall not discuss here, of whether the anti-reductionist arguments wh.icb Austrian methodological individualistS wish tO usc against the possibility of reduction from below (physics and neurophysiology) also can be wielded against methodological individualism itself, by nonindividualist social scientistS who doubt the reducibil ity of social science to the level of ind.ividual human action. We now must state the thesis of methodological individualism somewhat more precisely. Consider the question of whether we can reduce the theory of two-person interaction to the theory of individual human action. Economists who discuss individual human action often use the example of Robinson Crusoe, so we might call the theory of such individual action "Robinson Crusoe theory." Our question then is whether the theory of the interaction of Crusoe and Friday can be reduced to Crusoe theory. Crusoe theory is rbe theory of Crusoe's interaction with the
inanimate and nonhuman animate environment. Crusoe faces scarcity, allocates time and resources to some uses and forgoes others, does what he prefers, satisfies the principle of diminishing marginal utility, exhibits rime-preference, saves from current consumption to increase future consumption, and so on. Crusoe theory wi ll include the theory of individual decision under risk and uncert.ainty.3 It will talk of various dimensions, such as actions, alternatives, expected consequences, estimates of likelihoods of expected consequences, and, I believe, expected urili!J, bur nothing need be made of this now. Does the theory of two-person interaction merely specify the previous Crusoe theory and apply it in particular circumstances, or doe.s it introduce something new and irreducible? This question is interesting, and it is not clear wh.at the answer is. Before discussing it further, I must Copyrighted material 112 -& CHOICE AND UTILITY contrast it with a trivial and uninteresting question and answer. Suppose it is asked whether the theory of two-person interaction is reducible to the theory comprised of aU psychological truths about ittdiuiduals, and that an affirmative answer is proposed, on the grounds that it is a psychological truth about Crusoe that when be interacts with another in a cenain situation he behaves in a certain specified way.4 This is
a truth about Crusoe, bur what kind of truth is it? In particular, is it a truth that follows from or is a specification of the tmths about Crusoe itt noni11teractittg situations? Having distinguished the interesting from the trivial question, let us notice why the answer is not obvious. Crusoe can't just treat Friday's actions as events of inanimate nature, form certain expectations about them, estimate the probabilities of the different things which Friday might do, and act accordingly and rationaUy. The problem is that what Friday will do in some situations depends upon his estimate of the probabilities of what Crusoe will do, where Friday does 1101 merely treat Crusoe as an inanimate object, but as someone who also is estimating what he (Friday) will do. Each treats the other as a rational agent whose action depends upon (his estimate of) the rational course of action of the other, mutually realizing that the other's act depends upon his own in this way. Such situations are treated in the li terature of game theory.5 Is the theory of action in game-theoretic situations reducible to, that is, derivable from the general theory of isolated action (Crusoe theory) plus a statement that the people are in, and (mutually) realize they're
in, a game-theoretic situation? Or, is the theory which is true of people in game-theoretic situations a new and irreducible theory? We cannot say merely that people will do what they think is best io this situation, taking this claim from Crusoe theory. The question is whether what they think best in this situation is determined by the Crusoe truths as applied to this situation. The answer to this question is not obvious, and we should nore that theorists who work within game theory develop conceptS (e.g., "equilibrium strategy") which do not seem merely to be specifications of some notion(s) within the Robinson Crusoe theory. (That is, they do not seem to be merely notions of Crusoe theory restricted by parameters of two-person interaction.) To say that since all there is are individual people, the Robinson Crusoe theory must be sufficient, would be to answer th~ trivial question. ("It's true of h.im that in a game-theoretic situation he does such and such.") The interesting one would remain. Copyrighted material On Austrian Methodology -<> liJ However, I do not wish tO deny that the inanimate environment might mimic game-theoretic situations. For example, suppose Crusoe interacts with a mechanism whose states causally depend upon the actual
reasoning Crusoe goes through. Crusoe's reasoning is mirrored in the states of his brain, and these stares causally iniluence the mechanism whose motions then affect Crusoe. If Robinson Crusoe theory includes the theory of Crusoe's interactions with this bit of inanimate nature, it may well include game theory. Or perhaps without such a mechanism a person may (consider himself to) be in a choice situation wherein he must anticipate and perhaps thwart the choices of his future self (who will nor remember the earlier choice). So, again, the theory of one individual's behavior may come tO include game-theoretic or simi lar principles. It is plausible to try to demarcate such situations S as involving an agent's belief that outcomes are contingent (in a way which must be specified) upon how his environment reacts to his own actions. Given this demarcation, Robinson Crusoe theory would be the theory of an isolated individual's behavior in non-S situations. On the other hand, if the demarcation cannot be drawn, so that by this route the theory of an individual's behavior includes game theory, Kantian moral considerations, ere., then the present epicycle about n-person interactions will
nor be needed in order ro stare rhe methodological individualist position. Is the theory of 3-person interactions reducible to Crusoe theory plus the theory of 2-person interactions? Three-person situations involve the possibiliry of coalitions of two against one, of a member of the majoriry coalition being lured away into a new coalition with the outsider, and so forth. Suppose, for example, that $10 is to be divided up among. three persons so that if (at least) rwo of the three stay agreed to a particular division, it gets instituted. Crusoe and Friday say, "Let's each take $5, and freeze Defoe out." Defoe says to Crusoe, "I'll give you $6 and rake only $4 for myself, so you're better off in a coalition with me than with Friday." Crusoe agrees. Friday then says to Defoe, "Let's divide the $10 by giving $5 ro each of us, and freeze Crusoe out. Tbat way we both benefit." And so on.6 Once again, it is nor obvious whether 3-person theory is reducible to 1- and 2-person theory, or whether instead new principles emerge wbich govern the new phenomena that are possible. Our questions have been of the form: are the laws of 11 + 1person interactions
reducible to the laws of n-person (inrer-)acrions? The methodological individualist is committed, I believe., to saying Copyrighted material 114 -t> CHOICE AND Ul'ILITY (a) that there is an n above which there is no change in the laws; all laws about interactions are reducible to the theory of n (or fewer)-person interactions; all social science is reducible to the laws(< n) of human interaction pl11s a specification of the situation (to which the laws are applied)7 and (b) that 11 is small. I admit that "small" is an imprecise way of delimiting the claim. How small must 11 be, if methodological individualism is to hold true? Three? Anything less than ten? However, such imprecision does not undermine the contrast of methodological individualism with the view that, for example, a new theory (as opposed to a speci6cation and application of the previous theory} is needed for the explanation of behavior of (and in) crowds. 8 The methodological individualist denies that there are specialized and irreducible psychological truths to the effect that a person in a revolutionary situation and conditions C does A, etc. It is logically possible that there be such emergent truths, irreducible to the g~meral theory of n-person interaction.9 (In this case, the
completely general claim of the methodological individualist would be incorrect.) How does the methodological individualist treat institutiotiS? Let us take as an example (admittedly a favorable one) the institution of money. Menger offers us the following account. 10 To avoid the disadvantages which direct baner involves of not being able to trade for something you want from Crusoe who wants nothing you have, and therefore having to search for an intermediate Friday who has what Crusoe wants and wants what you have, people will tend to hold, trade for, and be willing to accept in exchange those goods they know others are (more) likely to accept in exchange; the more this is known, the more will such goods be traded for, and so there will be convergence upon a small number of goods which will, for obvious reasons, be initially valuable, portable, easily divisible in varying quantities, and homogeneous. Thus a medium of exchange precipitates out of the exchanges of individuals i.n a baner situation, each attempting to improve his own situation while taking account of the likely actions of others. Menger thus provides us with an explanation of the creati011 and of the mai11tenance via individual actions of a particular institution. II Existing institutions also shape and affect the actions of individuals.
They affect the opportunities available to people, and they shape Copyrighted material On Austrian Methodology ~ 115 people's utility functions. Furthermore, institutions are transformed and altered into different ones, and sometimes they are overthrown; tltis, roo, is the result of individual actions. These statements are not very controversial. The methodological individualist adds that in explaining each such transition and affecting, it is a general theory (of < 11person interaction) which is specified and applied, and not some specialized theory which fits only that social situation. Our construal of the thesis of methodological individualism has several virtues: under this construal, the thesis is an interesting one, it has real content, and could be false (emergence could be true), and at present its truth (l believe) is unknown. Furthermore, the explanations which methodological individualists offer, and view as satisfying their thesis, aU fit the thesis so construed.12 Must methodological individualists speak of institutiOitS at all? Rather, won't they view institutions merely as the sum o the actual acrs done within them, so that any institution is nothing more than aL"tual acts done by particular people (who occupy particular institutional roles but are otherwise identified)? However, institutions are not
merely sums of particular act tokens, when if the person hadn't done those particular act tokens, he would have done other similar ones, or when if this person didn't occupy an office or role, another would have who would have acted similarly. In such a situation, the subjunctives must be explained also, not merely the indicative facts of which tokens were done when. lr is the existence of such subjunctive facts which prevent,s the identification of an institution with the particular act tokens done (as we say) under it. But though tltis identification is blocked, the methodological individualist can proceed to (try to) explain the subjunctives as well as the act tokens on the basis of previous act tokens, and so nothing about institutions need be left unexplained. t.l Let us glance further at the patterns whereby actions C(Cate and maintain institutions which shape actions (see Figure 5.1 ). An institution is self-sustaining if it shapes actions which maintain it (and are sufficient ro overcome actions which tend to alter it). An institution is selfdestructive if it shapes aCtions which alter and transform it. 14 There will be a regularity wherein Institution 1 is followed by Institution 2 if 1 shapes actions which create 2. Seeing this explanation will suggest what type of exception,s to the generalization there will be. The
methodological individualist derues there is a law that Institution 2 must follow Institution 1, and that becatise of the law Institution 1 will willynilly shape some actions or other which lead to lnstirution 2. To explain the Copyrighted material 116 ~ CHOICE AND UTILITY current situation of institutions and actions, we would show how current actions are shaped by yesterday's institutions and actions, and bow today's institutions are yesterday's institutions altered by intervening actions (and maintained by some), with a cultural lag. Institutions don't disappe.ar overnight, for they are embodied i.n modes of behavior which don't alter overnight. lostirutioo 1 shapeS Actions Aetions Lnstitution 2 c.reo~~re ;.md maintain Figure 5.1 If each time we explain the current situation as arising from previous actions in a certain institutional setting, then why are actions prior? In this apparent chicken and egg situation, wby aren't we equally methodological institutionalists? Why think in terms of Figure 5.1 rather than Figure 5.2? Do we eventually get back to a starting point of actions and
no institutions? lns-tirurion lnstirurio.n Actions Actions Diagram 1 Actions Actions Figure 5.2 Institution Institution The first people came from organisms with mating patterns, group relations, territorial patterns, and so on. We have no need to pre.ss Copyrighted material On Austrian Methodology ~ I I 7 the term institution. Perhaps institutions are (by definition) transmirted and maintained cultl;ra/ly, whereas these far-past parterns were (initially) passed on biologically, as a result of evolutionary selection.15 Given these facts, does the methodological individualist win or lose? He wins, I think, if he can explain all with the theory of human action plus evolutionary biology. ltiStitutions are then, initially, a dependent vuiable. However, the process of evolution may have instilled desires wh.ich themselves refer to institutional or particular social situations. l.n this case, even though non-macro-social explanations can be offered from the beginning, social scientific explanations of current behavior would have to admit (innate) desires or reinforcers for which macro-
social reference would be needed to specify either what is desired or the conditions under which the desire is operative. The strongest and most general thesis of methodological individualism would then be false, and though the facts mentioned would surely not greatly perturb the methodological individualist, it is difficult to see how to modify his thesis so as to make it compatible with such biological facts and s.imilar ones as yet unknown, while still avoiding the trivial thesis discussed earlier.16 We haven't yet focussed upon the part of the scheme where institutions shape actions. The methodological individualist, we have seen, is distinguished by his claim that it is a general theory of action which gets applied, not particular theories which are irreducible to general laws of action. But what are the general laws whereby institutions shape actions, what are the general laws of the shaping of utiliry functions? The Austrian tradition has devoted lirtle attention to this question, perhaps because it was thought that all substantive features of utility functions (or preferences) were. matters either of biology or of free choice. If methodological individualism is true, there will not be irreducible laws
of the form: people brought up under institution I (tend to) have suchandsuch a type of utiliry function. The fundamental laws will not mention a particular kind of institution; rather, there will be general laws of utility shaping, which are then applied to the situation of a particular institution, in order to show how it shapes. Institutions (the stable patterns of others' actions) provide opportunities for doing various things, and a certain parterning of rewards and punishments for actions. One general theory of the shaping of utility functions might state how opportunities, rewards and punishmentS, and contingencies of reinforcement combine to shape utility functions and preferences. The framework of Copyrighted material 118 -&- CHOICBAND UTILITY this sort of theory is provided by the theory of operant conditioning. 17 This theory has been much attacked by libertarian writers, but it is important to see that methodological individualism requires some general theory (it need not be this one) of how utility functions are shaped in institutional environments. Supposing the thesis of methodological individua lism is true, what are its consequences for the practice of social science? The fact that all true tbeories of social science arc reducible to general theories of human
action and imeraction ( < n) does not mean that all true theories already wi ll have been reduced, or that it will be obvious how the reduction is to be carried through. Therefore, a social scientific theory formulated at the macrolevcl cmmot be condemned merely because it is (as yet) unreduced. It could be condemned (supposing the thesis of methodological individualism were known to be true) if it were known that it was impossible to reduce it to theories of human action. But it is doubtful in tbe area of social science that a proof of tbe irreducibility of some theory will be forthcoming. (Though perhaps one could show tbat a contrary theory at the macrolevel follows from a specific consistent theory of human action.) Does it follow from the thesis of methodological individual ism tbat the proper way to construct a social theory is to start with the theory of human action, and to work one's way up from it? (This is the procedure of the Austrians.) Compare the situation of someone who believes that the true theories of biology are (ultimately) reducible to physics and chemistry. He will not condemn the biologist for failing tO do physics and chemistry, and he will not condemn biological theories which bave not yet been reduced (even though he does believe that whatever true biological theories are discovered will
be reduced to physics and chemistry eventually). Social scientists should be the last to claim that there cannot be a similar division of labor in their intellectual world. Thus, it appears tbat no consequences r~eed follow from the thesis of metbodological individualism about what our attitude should be to any given as-yet-unreduced macrotheory of social science. Methodological individualists, like everyone else, will have to assess the truth of such theories by examining the evidence and arguments for and against them. This conclusion is so much at variance with the usual Austrian methodological individualist view that it suggests an attempt be made ro formulate the thesis of methodological individualism so that it is r~ot merely abour the reducibility of theories. If I knew how else to formulate it, I would. Copyrighted material On Austrian Methodology -& 119 II. The A Pri~tri Science of Human Action and Its Applications One branch of Austrian theorists (Mises, Rothbard) holds that (many of) the most important general laws of human n-person interaction are derived within (as they term it) an a priori science of human action. This science, it is held, begins with an examination of the essence of action
(that is, purposeful behavior) and, in elaborating what is contained in th is essence, is led to a body of necessary truths holding of human action as such. Note that holding there is this a priori science of human action is logically independent of accepting the thesis of methodological individualism. For a methodological individualist might hold that all true social scientific theories are reducible to laws of human action, but that none of these latter laws are necessary truths. And someone might hold that there is a body of necessary truths ahout human action, but that these do not suffice to reduce and explain all true theories of social science. 18 The idea of elaborating what is contained in the essence of human action is certainly an interesting and challenging one. Such a project, involving as it does synthetic necessary truths, would have been ruled out by the logical positivistS as impossible and empty. But the positivist position and arguments on synthetic necessary statements, as much else within the positivist position (for example, the verifiability criterion of meaning), has fallen upon hard times. Recently, there has been a resurgence of interest in such statements, and it is fair to say, I thin.k, that there are no arguments generally acknowledged to be
compelling against the possibility of such synthetic necessary truths.19 This is not the appropriate occasion to investigate the general possibility of such truths. Instead, we shall consider the implications for the practice of economics if there were such a body of truths. Later sections of this essay will consider particular such (purported) truths. I shall consider ooly some of che (purported) derivations from the essence of action, but I should mention that there is a need for a dear, precise, and consistent statement of the content of the (a priori) theory within a specified vocabulary of primitive terms and with explicit definitions and axioms. Such a statement should make clear whether preference is initially over actions or outcomes; it would fix one notion (preference, satisfaction, desire) as primitive and define the others in terms of it; it would be sophisticated enough to take account of the considerations Copyrighled malerial 120 -b CHOICE AND UTILITY of the theory of choice under uncerta inty and consistently distinguish in the presentation of rhe theory what it is expected will happen from what it is thought might happen; and it would avoid Mises' unfortunate tendency to speak as if the outcome of an action is preferred to
the current situation (it need not be) rather than to what would obtain if the action weren't done. More care also is needed in stating the future orientation of action, for the point of an act may merely be ro do it, or to be continuing a previously started plan, or to be foUowing a previous commitment. Thus, it is a mistake, I thll1k, to speak as Mises does of acting man necessarily ignoring sunk costs. It may be irrational to consider them, but people in restaurants, for example, whose food has come and turned out 10 be poor-tasting certa inly often speak as though the reason they're going ahead 10 eat it is that money already has been committed to it. True, one might explain their sticking with this food rather than ordering another (preprepared) dish in the same way we explain why the newest, most modern equipment is not immediately installed in every factory. But, on the other hand, it is not impossible that letting something he's paid for go uneaten has disutiliry for that person. Economic theory can assume entrepreneurs ignore sunk costs, not because doing so is part of the essence of human action, but because the market tends to filter out entrepreneurs who take account of sunk costs. FinaUy, I would want the theory to be formulated so that even
though preferring is a subjective psychological state, the ultimate rhings which are preferred one to another need not themselves be subjective psychological states (such as felt satisfactions or dissatisfactions, or removals of such things). This, I realize, is a more complicated question, but I need not linger over it here, since I have stated relevant considerations else.where. 20 I must make only these brief remarks about formulating the theory, since I wish to examine in somewhat more detail the connection of preference, choice, and acrion, and the theory of rime-preference. But first, in this section, consideration must be given 10 the application of a body of necessary truths about. action. Actions are a certai.n r.ype of behavior, namely, purposive behavior. Not all behavior is purposive. There is caused bodily movement which is not even behavior, such as the movement of a body when dropped unconscious from a height. There is behavior which is not action; for example, simple unconditioned reflexes. By traioing, such behavior might come under the control of choice, but if such training didn't take place, the reflex behavior wasn't action (though the earlier "nontraining" may have been). What about the conCopyrighted material
On Austrian Methodology ~ 121 ditional reflexes of classical Pavlovian conditioning? The ringing of rhe bell has been paired with food, and now (with no food present) the bell rings, and the dog, or the human being, salivares. Is this salivating behavior an action? I do nor recall a discussion specifically on this point in the writings of the Austrians, bur their answer, I believe, would be that such Pavlovian reflexes are nor actions. More intere.sting and difficulr is opera>ll behavior. Is behavior which falls under the principles of operant conditioning, action? Processes of operant conditioning by use of rewards and punishments, positive and negative reinforcements, on various contingencies of reinforcement, raise and lower the probability of various behaviors, and introduce various patterns of extincrion.21 Is such behavior, which is subject to and shaped by reinforcement, action? Austrians might quarrel with applying principles of operant condirioning to people and to theorizing about people in this way, for the behavior rhe Skinnerian psychologist applies his principles to is the very behavior rhe Austrian theorist would call "purposeful," that is, "action." The Skinnerian psychologist applies his principles to animals as well, which raises the question of
whether the Austrian theorist counts operant behavior done by, e.g., rats and pigeons as action. Again, I do not recall a speci6c discussion of this question by an Austrian theorist, but I would think they would be reluctant to apply to animals many of the categories that go with actions, e.g., expectations, plans, images of a desired end. Recall that principles of operant conditioning have been observed to operate quite low on the evolutionary scale. lf such operant behavior of these other organisms is nor action, then we have an interesting example of a type of behavior to which the categories of human action do nor apply. The principles of operant conditioning, let us assume, apply to these other organisms. (Note that there is an evolutionary account of the adaprive advantage of operant conditioning; being operantly conditionable would be selected for in the process of evolution among organisms that were merely subject to classical conditioning.) Might those principles also be true of us, of large segments of the behavior of human beings, as an a posteriori theory? Do we know a priori that its categories and statements don't apply to much of human behavior?
There is this alternative, relatively elaborate conceptual scheme which does not appear to talk of or within the Misesian categories, yet whether it is true of people or nor appears to be an empirical question, to be decided a posteriori. (Did these principles just stop being true at a certain Copyrighted material 122 ..,. CHOICE AND UTILITY point on the evolutionary scale? Does human consciousness and intelligence provide new means through which these principles work, or instead an area in which they don't work at all?) It may be that Skinnerian theory and human action theory are compatible, and can both be true of the same behavior_ If so, even if human action theory is known a priori, it may be that it is not the best rheory, for the Skinnerian rheory actually may explain more, enable us to predict more, and so forth. In this case, the possession of an a priori theory would not stop us from pursuing and developing, formulating and resting, a body of contingent truths about behavior. Possession of an a priori theory cannot tell us that there won't also be a better, more predictive, wider-ranging a posteriori theory, and so cannot bar the pursuit of empirical science. (This point, obviously, does not depend on anything about Skinnerian theory, but it seemed useful to make it by considering an actual alternative framework, very different from Mises', which
functions within an a posteriori theory currently under active investigation.) On the other hand, it may be that the two theories cannot both be true of the same behavior, that the truth of one excludes the truth of the other. Many Austrians write as though there is no set of conceptS, other than theirs, via which one conceivably might understand much of human behavior_ The existence of Skinnerian theory casts this in doubt, and supposing this theory to be incompatible with human action theory, it raises the delicate question of which of the two bodies of theory is true of most of our behavior in the world. (In Austrian terms, it raises the question of whether that behavior is action.) Note the importance for the Austrian view that there be nothing else close tO action, nothing very similar to it. On that assumption, human behavior wb.ich isn't action can easily be distinguished. It would not be a delicate matter involving much observation and testing to decide which theory holds true of some behavior which appears to be action. On the other hand, suppose there were behaviors similar enough to human actions so that they could not be distinguished at one or even two glances. Suppose further that such nonaction behavior occurred frequently,
often intermixed with the actions of persons. The nonaction behavior does obey different laws, though, e.g., the laws of operant conditioning. Even if there were an a priori theory of human action, the question of whether this theory fit a particular situation would have to be determined by detailed empirical investigation, complicated statistical procedures, etc. That is, the claim that the theory of human action Copyrighted material ---------------- -----On Austrian Methodology ~ 123 fits a situation would require as much evidence as the claim that some other a posteriori theory firs a situation. The a priori character of the theory of human action would be of 110 help in deciding whether it was true of a behavioral situation, and economics would be carried on in as empirical a manner as any other science. (Even if somehow in this siruation someone could know a priori that it was an action he was performing, others couldn't know dt.is a priori.) Thus, it is important for the Austrian view of how economics is to be carried on to add to their claim that there is an a priori science of human action, the claim that there is a great enough gulf between human action and other human behaviors to make it easy to identify which behavior is action. Otherwise, the only way to determine
whether or nor behavior is action might be to determine which laws it satisfies. And this might require as much empirical investigation as goes on in an a posteriori science of human behavior. Can the existence of sucb a gulf be known a priori; is it a necessary rrutb that no behavior mimics action closely enough to require detailed investigation to separate them? I do not see that it is, and the existence of an a posteriori theory of this sort (e.g., Skinnerian theory) bas the consequence that the science of human behavior must, to a great extent, be carried on as an empirical science. Even if it were known a priori that actions satisfied certain laws, it would be an empirical question whether particular bits of behavior are actions and, hence, whether they satisfied those lawsP Does the method of verstehe11 give us knowledge of which conceptual structure applies to people's behavior?23 Critics of verstehen have claime.d that it is at best a route to thinking up hypotheses, and not a way of cont.ing to know which hypotheses are true.24 Current views on the testing of hypotheses describe a role other than the suggesting of hypotheses or the showing of which one is true, namely, the assignment of plausibilities to hypotheses or (on the Bayesian view of
statistics) of prior probabilities to these hypotheses. Such plausibilities or prior probabilities will affect what it takes co show a hypothesis is true or false. tn an unpublished essay, Hilary Putnam makes the interesting suggestion that the process of verstehen come.s into this prior assignment, and so legitimately plays a role (though not a conclusive one) in the process of accepting and rejecting hypotheses. Even if we grant Putnam's point, we should step back to examine what might underlie the use of verstehen. Consider the enterprise of what, following Quine's use of rhe rerm radical translation, 25 we might call radical verstehen, rhe understanding of the behavior of organisms Copyrighted material 124 ~ Cl:IOICB AND UTILITY of another species or with whom we have no biological links since they evolved on another planet. Can we vcrstehen such organisms? Empathetic understanding might suggest hypotheses about their behavior, but these would have to be te.sted further. The general hypothesis that we are able to verstehen their behavior would be tested in the process of testing the specific hypotheses about them which empathetic understanding suggests.26 What is clear is that we could not know a priori that we
have the ability to empathetically understand the behavior of these organisms. If we do have this ability for those organisms, we will have to come ro know this a posteriori, presumably by appropriate procedures of psychological science. It might turn our that we have this ability with only some types of their behavior; though empathetic understanding suggests with equal plausibility hypotheses about other types of their behavior, further testing might show our success rate with the.se latter hypotheses to be no better than random. Our reliability with different types of their behavior will be determined a posteriori. Closer to home, consider the claim that you can verstehen the behavior of other persons. That, roo, it seems, is nor known a priori. Presumably, you will have confidence (if you do) in your ability to so understand the behavior of others because those of your everyday attempts ro do so which had funher predictive consequences have borne up pretty weU. Bur such a nonrigorous rest of the hypothesis cannot show that an alternative conceptual scheme is inappl icable when this alternative scheme often would have roughly the same fun her predictive consequences about the behavior of others in the situations of everyday life. ln that case, the apparent success of our episodes of empathetic understanding might be only apparent.
Bur can't we know, without intricate resting, which concepts apply to our own behavior, that is, can't I know which apply to mine, and you know which apply to yours? We each cenainly do apply a parricular conceptual scheme to our own behavior with great conlidence.27 But on what basis do you do so and, in panicular, do you know a priori that this scheme applies to you? Perhaps one wantS ro say that you know which concepts apply ro your behavior since you're doing it, after all, and doing it in a way which infuses those concepts into the behavior which you produce, and which also leads )'OU to know that those concepts are so infused. You know you're making it happen as an action. As it stands, this is too murky to be helpful. Clarifying it is unlikely ro show that the way the concepts get infused into behavior you produce while under the sway of the concepts, makes the theory Copyrighted material On Austrian Methodology ~ 125 utilizing those concepts the best explanatory theory of your behavior. (And there still would remain the problem of knowing what concepts to apply to the behavior of other persons.) Still, there is the fact that we do apply, with confidence, the
categories of human action. Another theory, e.g., the Skinnerian theory, will fail tO generally undermine this confidence until it is combined with a theory of concept formation so as, from its own perspective, to explain the (cross-cultural) fact that people confidentially hold to the usual (non-Skinnerian) categories. If the Skinnerian theory could explain why, though the usual theory is (as it claims) false, almost all people adhere to it, this would make it more difficult to retain undiminished confidence in either the usual theory or in the process of verstehen, which uses its categories. Howe\er, even in this case it would not be impossible to continue to believe the usual theory, and the Skinnerian theory itself may predict this will happen, since belief in the usual theory was reinforced on a schedule which makes it highly resistant to extinction. Indeed, we may question how faithfully operant condi tioning theorists adhere to their own theory when they try to argue the public away from the product of long and varied processes of reinforcement! Verstehen cannot, we have seen, help to eliminate all empirical questions about the application of Mise.s' or any other theory of human action. However, we should not leave the subjec.t of verstehen
without remarking that it is possible that people (or some people) do possess the capacity of empathetic understanding, for at least some types of behavior, and that this capacity might be trained so as to become even more reliable. Whether this is so or not is an empirical question. Suppose that investigation by the procedures of empirical science identifies some persons as especially reliable verstehers of at least some types of behavior; as it turns out, they rarely or never are inaccurate. Even if we did not possess a theory of how this capacity operated, we might use such re liable verstehers as instmments by which to learn various things, even as we use nonhuman detectors of facts. Tests might yield the result that particular verstehers had a reliability as high as that we attribute to our usual empirical procedures of theory construction and testing, and in that case we would no more hesitate to say that verstehen bas given us knowledge than to say that the rest of science bas. The standard procedures of the rest of science would be what established the reliability of verstehen, but once this were shown, there is the possibility that social science might then be carried on very differently. Copyrighted material
126 -<> CHOICE AND UTILITY One final remark about the appl icability of a priori laws. Questions about this arise due to considerations within. the theory as well, for example, for the principle of diminishing marginal uti lity. It is often objected t.O this principle tbat an additional unit of X (say, the sixth) may have increased utility since, now, along with the previous five units, tbe person is able to acquire something previously unavailable. The Austrian reply points out that in this case the relevant size of the unit is 6X; the principle is meant to apply to situations where each additional unit of X can acquire only rhose rhings which previous units also were able ro bring (and together with the others can bring only a larger conjunction of previously available things).28 The size of the units must be chosen so this condition is satisfied. Now, it is logica lly possible, of course, that for any choice of a finite size of the unit X, whenever a new ntb unit of X becomes available, rhere is also a new total use of rhe nX's (that is, of the new X and the previous ones together) which ranks higher in the person's preferences than the sum of any n of the previously available things (and than the sum of the conjunction of something obtainable
with one X along with anything obtainable with 11 - 1 X's rogether, ere.). Let us ca ll such a situation an expa11ding universe. A person's universe is locally expa11di11g if ir is an expanding universe over as many units as he has; that is, for any choice of the size of unit, so that he has at lease one. The principle of diminishing marginal uriliry is not false of a person in a locally expanding universe; it just doesn't apply to him. The combination of his preferences and the opportunities available to him does not satisfy the condition of applicarion of the principle. Since whether or nor a person is in a locally expanding universe is an empirical question, it is an empirical question whether or nor the principle of diminishing marginal utility applies to a particular person in a particular situation. I II. Preference, Choice, and Action Austrian writers put forward a number of interconnected theses about the relations of preference, choice, and action: (1) lf a person does an action A, then that person preferred doing A co doing any other act which (he believed) was available t.O him at the time. (2) No evidence can establish that a person prefers A ro 8 in the face of a choice of 8 when A was available ro the person.
Copyrighted material On Austrian Methodology --~> 127 (3) The notion of preference makes no sense apart from an actual choice made. 29 Does all action show preference? (That is, is 1 above true?) Mightn't the person be indifferent between what he did and some other alternative available to him? We might have the following picture: of the acts available to a person, some he is willing to do, and others he is unwilling to do. He prefers A to 8 if and only if in a choice between A and B be is willing ro do A and unwilling to do 8. Let us grant that if be does A, tbis shows be was willing ro do A. But doing A does not show be was unwilling to do B. He might have been willing to do B also. Thus, doing A doe-s not show the person preferred doing it to doing B. Writers sometimes speak of weak preference, of the person we.akly preferring doing one act to doing anotbeL We can understand "the person weakly prefers doing A to doing 8 " as: the person prefers doing A to doing 8, or the person is indifferent between doing A and doing B. In terms of this relation of weak preference, a person is indifferent between doing A and doing B if and only if he weakly prefers A to 8, and he weakly prefers B to A. A person strongly prefers A to B
if and only if he weakly prefers A to B and be does not weakly prefer B to A. Choosing an act may well be a sufficient condition for weakly preferring doing it to doing one of its alternatives. But it is not a necessary condition for that and, hence, it is not a sufficient condition for strong preference. There are other indications of indifference; for example, the person flips a coin between doing A and doing B and acts on the ourcome of the flip; or, the person uses a random device to choose between doing A, doing 8, and flipping a coin ro decide between doing A and doing 8, and so forth. Indeed, the Austrian theorists need the notion of indifference to explain and mark off the notion of a commodity, and of a unit of a commodity. If everyone or one person prefers one homogeneous batch of stuff to another homogeneous batch of the same shape of the same stuff (perhaps they like to choose the left-hand one, or the one mined first), these are 110t the same commodity. They will have different prices. Particular things x and y will be the same commodity (belong to the same commodity class) only if all persons are indifferent between x and y. Without the notion of indifference, and, hence, of an equivalence class of things, we cannot have the notion of a commodity, or of a unit
of a commodity; without the notion of a unit ("an interchangeable unit") of Copyrighted material 128 ..,. CHOICB AND UTILITY a commodity, we have no way to state the law of (diminishing) maiginal utilicy:10 Does the existence of behavior, when a choice was made, show what was chosen and what was preferred? The behavior done does not show to what something was preferred, for it does not show what behavior was rejected. That depends upon the person's beliefs about what alternatives were available to him, and these are not shown by his actual behavior. We might think it is shown by his talk, but why can talk show beliefs about the alternatives available yet not show preferences? We should not answer, "because beliefs issue in talk and preference.s issue in action." For, don't beliefs sometimes issue in action, and preferences in talk? Rather, aren't talk and action each the product of beliefs and preferences both? Do we at least know what was the preferred alternative? Note that the behavior performed can be described differently, e.g., " travelling to the 6rst floor; "stepping in this particular place," "exerting a certain
pressure." What did the person prefer doing to something else? Preference, it seems, does not apply to things directly, but applies only via descriptions, via something linguistic-like. We cannot just say "that behavior," and point. For which is that? We might try to narrow things down by noting that what he preferred was something he knew about; hence, the description has to be one he knew of at the time he was acting or choosing. But which of these descriptions he knows of is connected with his preference? Perhaps the answer is, all that he knows of: he prefers the combination of all the descriptions he knows of for the behavior he does, to the combination of all the descriptions he knows of for each of rhe alternative behaviors he doesn't do. Bm he may have neglected and even repressed many descriptions he knows of. He knows of many he didn't consciously think of at any point in the process of deciding; he thought of some which he forgot, but could recall with effort, and so forth. It might be suggested, instead, that he prefers the combination of descriptions that (he thought) fit the behavior done which he was actually thinking of, as he decided, to the combination he was then actually thinking of which (he thought) fit the behavior rejected. But mightn't he then have been thinking only of the worst description, and then decided,
" I'll do it nonetheless"? Need he have been thinking then of a description at all? Couldn't he have gone through deliberations, and then when the time came, just acted without consciously thinking about it? Copyrighted material On Austrian Methodology ~ 129 What is preferred, it seems, is hidden from an observer. The most an observer can say, for sure, is that when action rake.s place something or other is (weakly) preferred ro something or other else; more fully, something or other connected with the behavior done was (weakly) preferred ro something or other else not connected (as closely) with the behavior done. lr is often said that an external observer cannot predict behavior on the basis of preference. Here we see that he cannot retrospectively describe preference, even on the basis of the behavior done! Must it be, at least, that the person who acts knows what he himself preferred? Can't he say, "I thought about many things about what ro do, and then just acted later without thinking about it then. I don't know what about it I preferred" ? Is this impossible? or just rare? "Bur he knows he preferred doing it!" Need he know what the it is which he
preferred? "Well, he did it!" True, bur who is denying that? Let us now tum to the strongest of the above statements about the connection of preference and action, viz. (3): the notion of preference does not make sense apart from an actllal action which exhibits it. Notice that this is stronger than saying that preference is somehow connected ro choices, to what a person wo.,ld choose under certain circumstances. It says the notion is inextricably tied to acmal choices. It is nor clear that the Austrians can consistently rnauuain th.is strong thesis. Consider, for example, Mises' view of action as exchange.31 An acrion A is done instead of another act B. What is nor done, B, is the price of doing A. The value of B is rhe cost of doing A. Mises then goes on to talk of the profit of an act, and also of the (I assume, ex post) loss. Now the cost of doing act A cannot be the value of everything forgone, that is, it cannot be the s"m of everything forgone. If it were, we almost never would have profit. Suppose there are five alternative acts, closely bunched, and you do the one act, A, which is the best. The cost of A is not the Sllm of the values of the other four. (This would greatly outweigh the value of A.) The cost of A is the
value of the best one of the rejected alternatives. The best one of the rejected alternatives is the one you strongly prefer ro all the others, or (if there's a tie at the cop of the rejected alternatives) one you weakly prefer to each other rejected alternative. Notice that we are speaking of a person's preference apart from any actual choice of the preferred alternative. The next best alternative ro A, call it B, is preferred ro any of the others C, D, E, tho.,gh B itself is not chose11. If we are ro speak of the cost of A, when there is more than one other alternative rejected, it must make Copyrighted material 130 ~ CHOICE AND UTILITY sense to speak of preference apart from an actual choice or doing of the preferred alternative. If that doesn't make sense, then neither does the notion of the cost of the action which was actually chosen. 32 How can it make sense to speak of preference without an actual choice? Granting that the Austrians are wrong in denying that it makes sense, how is it possible? Let me present a sketch of a plausible account. Preference is connected with a subjunctive: to say a person prefers A to 8 at a time is to say he would choose A over 8 if he were given a choice between (only) A and 8 at that time. There's something about
him in virrue of which he would choose A over 8. It is not, for example, a random event. That an event actually happens does not show that a subjunctive is true. Consider photons emitted toward a screen with two slits, and suppose that it is truly a random event which slit any photon goes through. If two people bet on which the next photon will go through, one saying it's one slit and the other saying the other, the winner is the one who picked the slit which turned out to be correct. The indicative statement is shown true by the event. In contrast, if someone says, "if a photon were emitted in the next second, it would go through the righthand slit," and one is emitted, and does go th.rough the right-hand slit, this does not show he was right. The subjunctive statement is not shown true by the event. Perhaps this can be made inruitively more obvious via a recent account of subjunctives. Under this account, a subjunctive of the form "if p were true, q would be true" is itself true if and only if in all those pos sible worlds where p holds true which are closest to the actual world, q also holds true. The minimal change from the actual world necessary to make p true brings the truth of q along with it (in those closest
worlds).33 Now it might turn our char those closest worlds in which p is true are neither uniformly q nor uniformly not-q. In that case, in the actual world neither of the subjunctives "if p were true q would be true" and "if p were true, not-q would be true" will hold true. If it is truly a random matter wh.ich slit the photon goes through, then before the photon is emitted, for each possible world in which a photon is emitted and goes through the right (left) slit there is another possible world, equally close to the actual one, in which a photon is emitted and goes through the left (right) slit. More precisely, the "phoron is cmitted"neighborhood of the actual world contains worlds where the photon goes through the right slit, and it contains worlds where the photon goes Copyrighted material On Austrian Mrthodology ~ 131 through the len slit. The subjunctive "if a phoron were emitted it would go through that slit" is not true. To say a person did A rather than B because he preferred A to B is to say he did A rather than B because it's rrue of him that he would choose A over B. This is not a very potent explanation, but it does put the action in a pattern; it says the action flows from some underlying
disposition (which might be very fleeting) ro choose A over B. The occurrence of an action does not demonstrate that the subjunctive is true; also the person might have been indifferent. But often an action will stem from ao underlying preference aod will indicate that there is such a preference. On this view of preference as constituted by the subj unctive to the effect that a person wollld choose A over B if he were given a choice between the two, we can see how to make sense of preference without an actual choice of the preferred act. It might be that a person would choose A over B in a choice between the two, but since he didn't have to make a choice between the two, chat preference was not exercised; he did not actually choose A over B, and he didn't do A. A substance is soluble in water if it would dissolue if placed in water. Substances which are never placed in water may yet be watersoluble, even though they never actually dissolve. "Prefers A co B" is li.ke "soluble"; "chooses A over B" is like " dissolves." The claim that it makes no sense to say a person prefers A to B unless he's actually chosen A over B is like the claim that it makes no sense to say something is soluble unless
it already has actually dissolved. Both claims arc mistaken.34 The cost of an action, we saw, is the value of the most preferred alternative not chosen. To say that some unchosen alternative is preferred ro every other one is to say it ruould be chosen over each other one io pairwise choices. And this might well be true. If the Austrians were correct in talking of scales of values as existing only in actual choices, there couldn't be a particular cost of a choice. Subjunctives give us the possibility of making sense of rationality conditions on preference, e.g., the condition that preference be transitive. To say that preference is transitive at a timet is to say that for any three acts X, Y, and Z if (1) it's true that the person would choose X over Y if given a choice at timet between (only) X and Y and Copyrighted material 132 -b CHOICE AND UTILITY (2) it's uue that the person would choose Y over Z if given a choice at timet between (only) Y and Z then (3) it's true that the person would choose X over Z if given a choice at time 1 between (only) X and Z. It is possible that (l) and (2) are both true for some X, Y, and Z, yet (3) is not; rather, ( 4) might hold: (4) it's true that the person would choose Z over X if given a
choice at time 1 between (only) X and Z. Also, iris possible that no subjunctive, neither (3) nor (4), is true of the choice between X and Z. In this case, also, preference fails to be transitive. It is possible that (4) is true along with (1) and (2), because each of these subjunctives talks about different hypothetical circumstances. Since (1) and (2) do not talk about a choice between (only) X and Z, they don't, even together, settle what would happen under that condition. However, they arguably do settle what should happen, and so we have room for a rationality condition.JS Mises considers and objects to the requirement that preference be transitive (Human Action, p. 103): The attempt has been made to attain the notion of a nonrarional action by this reasoning: If a is preferred to b and b to c, logically a should be preferred tO c. But if actually c is preferred to a, we are faced with a mode of acting to which we cannot ascribe consistency and rational ity. This reasoning disregards the fact that two acts of an individual can never be synchronous. If in one action a is prderred to b and in another action b to c, it is, however short rhe interval between the rwo actions may be, not permissible to construct a umform scale of value in
which a precedes b and b precedes c. Nor is it permissible to consider a later third action as coincident with the two previous actions. All that the example proves is that value judgments are not immutable and that therefore a scale of value, which is abstracted from various, necessarily nonsynchronous actions of an individual, may be selfcontradictory. The acts cannot be synchronous, but the subjunctives can hold true at the same time. So we can make sense of nontransitive preferences at a time. (Still, it does not follow that the action is irrational, only that the structure consisting of that action plus those preferences violates a normative requirement. Which should be changed is left open.) Leaving Copyrighted material On Austrian Methodology ~ 133 preferences aside, what are we to say about the corresponding nontransitive subjunctives? Isn't someone irrational of whom (1), (3) and (4) bold true ?36 If preference is specified by choices that would be made in pairwise choice siruations, there is room for normative conditions in addition to the transiviry one_ For there will be conditions connecting pairwise choices with nonpairwise choices. For example, and
[f, if there were a choice only between X and Y, X would be chosen, then if there were a choice between X, Y and some other a.lternatives, Y would not be chosen. If in a choice among the members of set S, X would be chosen then in a choice among the members of a subset T of S of whjch X is a member, X would also be chosen.37 Why does Mises think it so important to argue that the structure of preference.s cannot be irrational? Perhaps because he doesn't want anyone interfering with choices on the grounds that they arise from irrationally structured preferences. That, however, is another issue, and one might think it possible for people to have irrationally structured preferences and also hold that this is within their rights, so that they may not be interfered with on the grounds that their preferences are irrationally structured (unless they have explicitly contracted into such a scheme of interference). I should not neglect to mention that the view of preferences as embodying (and embodied in) subjunctives about choice opens up the possibility of formulating further conditions on preference among probability mixtures of outcomes (in the manner of Von Neumann and Morgenstern) or among uncertain actions (mappings of states of the world onto consequences, in the manner of L. ). Savage) so as to yield measurement of utility on an interval scale.38
Of the three statements connecting preference, choice, and action which we listed above, we have argued that the first and third are incorrect. What about the second? Can evidence establish a preference of A over 8 in the face of an opposite choice of 8, when A was available to the person? Can there be indirect evidence that a person would choose A over 8 if given a choice (only) between A and 8, so that we conclude he does prefer A to 8, despite the fact that he has chosen B when A was 3\ai]able? The person might have chosen 8 over A and over lots of other things, over so many things, in fact, that he lost sight Copyrighted material 134 -b CHOICE AND UTILITY of A. Here would be a case where he would choose A in a choice (only) among A and B, yet in a choice among A, B, and many other things, he would choose 8, because he would lose track of A. Or perhaps there's no "would" about the matter; he just gets confused and acts; no subjunctive is true of the wide choice. The person actually chose and did B (violating one of the rationality conditions listed earl ier), )et this doesn't show he didn't prefer A to B. We might stand by the claim that he does prefer A to B, offering a special explanation of why he chose B that tune. These examples, which are designed tO cast doubt on (2) above,
involve choice among a large number of alternatives. There cannot be a similar counter-example which involves only a pairwise choice, if preference goes with a subjunctive involving "would always do A rather than B." lf, however, preference goes with a subjunctive involving "would ... un less ... " then there is the possibility of a pairwise counter-example to {2). There remains, however, another difficulty with (2)'s claim that no evidence can establish that a person prefers A to B in the face. of a choice of B when A was available to him. From a person's doing B, we cannot know he believed A was available to him, or how he conceived the rejected alternatives. If our earlier discussion of whether we can know to what something was preferred was correct, we cannot possibly, no matter wbat we conclude, fly in the face of a choice of B over A. For that is never a fact which faces us. Rationality conditions are conditions which it is possible to violate. In contrast, it is difficult to see how preference could fail to be irrellexivc and asymmetric. These seem co be parr of the notion of preference, and it is a virtue of the subjunctive account tbat it makes no sense of instances where preference is reflexive or symmetric. However, though (in contrast to constitutive conditions) rationality conditions are
conditions it is possible ro violate, they had better not be violated too much, without special explanation. For if some organism violated tbern just about always, it would be unclear that it was preferences which the organism possessed, even tbougb the binary subjunctives did bold of it. (If alpha-centurions always have an intransitive structure of subjunctives hold true of them, is it preferences that they have?) An organism has preferences, it seems, not merely because and when binary subjunctives are true of it. The binary subjunctives wbicb hold true of it must bang together in a reasonable fashion. The organism must show some modicum of rationality to be counted as having preferences at all. Given a being whose subjunctives usually and normally fail to rationally cohere Copyrighted material On Austrian Methodology -t> 135 (with no special explanation of this failure), it is doubtful that it is preferences which the being has.39 IV. Time-Preference Time-preference plays such a central role in the Austrian approach (to interest, capital theory, the business cycle, etc.) that we cannot close without giving it a brief examination. BOhm-Bawerk's famous three reasons for time-preference have frequently been examined,<0 and we turn
to more recent v1ews. We can deal briefly with an argument of Rothbard's (Man, Econom)l a11d State, p. 13, his italics): A fundamental and constant truth about human action is that mm prefers his end to be achieved in the shortest time. Given the specific satisfaction, the sooner it arrives, the better. This results from the fact that time is always scarce, and a means to be economized. The sooner any end is attained, the better. Thus, with any given end to be attained, the shoner the period of action, i.e., production, the more preferable for the actor. Tl1is is the zmiversal fact of time preference. This argument does not demonstrate time-preference on the basis of the scarcity of time. lime is scarce, and we want to economize it, to use less of it in achieving our goals. Given a way of achieving our goal that takes two minutes, and another way that takes five minutes, we will choose the two-mi.nute way, thereby leaving the three extra minutes for the pursuit of some other goal. However, time-preference is not the same thing as economizing time. Suppose there are two acts A and B, which each take five minutes ro do, and yield the same goal, but one delivers its goal earlier than the otheL Suppose A delivers seven minutes after
it's done, and B delivers one year after it's done. However, each taking five minutes to do economizes rime equally. Time-preference, therefore, can.not be derived from economizing time. ft also may conflict with it, as when the action which delivers its goal sooner take-S longer to perform than the action which delivers the goal later. Mises connects time preference with action in a different and more striking way (Human Action, p. 484): Trme preference is a categorial requisite (lf human action. No mode of action can be thought of in which satisfaction within a nearer period of the future is not-ocher things being equal-preferred to that in a later period. The very act of gratifying a desire implies that gratification Copyrighted material 136 -<> CHOICE AND UTILITY at the present instant is preferred to that at a later instant. He who consumes a nonperishable good instead of postponing consumption for an indefinite later moment thereby reveals a higher valuation of present satisfaction as compared with later satisfaction. If he were not to prefer satisfaction in a nearer period of the future to that in a remoter period, he would never consume and so satisfy wants. He would always accumulate; he would never consume and enjoy. He
would not consume today, but he would not consume tomorrow either, as the morrow would confront him with the same alternative. On this view, an action shows time-preference because if the person didn't prefer doing it now, he wouldn't do it now. Furthermore, if the person never preferred doing it at a time rather than at a later time, he would never act at that time, and so would never act at all. Let us scrutinize this argument more closely. First, a person might be indifferent between doiog some act now and doing it later, and do it now. ("W'hj not do it now?")41 So action now can show time(weak) preference, but it need not show time-(strong) preference. Second, a person might act now to get a particular satisfaction, without caring whether it comes sooner or later. He acts now because the option of getting the satisfaction is a .fleeting one which wiU not be available later. Thus, a person can have a reason, other than time-preference, to act now; to prefer satisfaction sooner rather than later is not necessary in order to act now. Third, the fact that we act constantly cannot show that we always have time-preference for all goods. At most, it shows that when a person actS (and rhe option also is available later) he has
rime-preference then fo[ the particular good he then acts to get. This is compatible with an alternation of periods of time-preference for good G, and periods of no time-preference for good G. The person acts to get G during one of the periods of rime-preference for G. This is considerably weaker than general time-preference, as might be seen by considering what the dteory of interest would look like if there were only rh.is weak form of time-preference. Finally, even if Mises' approach yielded the strong conclusion he envisages (which, I believe, it does not), it would leave time-preference mysterious. Action shows timepreference; but why is there time-preference? Time-preference would sti ll stand in need of explanation. 42 Let us approach time-preference by considering apparently unrelated propositions, which Austrian theorists usually list as additional (non-a priori) assumptions, viz., leisure is a consumer's good, and labor has Copyrighted material On Austrian Methodology ~ 137 dis utility. 43 It should be possible, I think, to derive the first of these from other deeper facts (which will not themselves be nece.ssary truths, though). Note first that leisure need not be a consumer good, for
each person. II depends what the person's labor is like. For example, suppose you have no desire for privacy, and a psychologist wants to study someone's life. He wishes to watch you from afar, unobtrusively, and he will pay you a fixed salary per houL There is no limit tO the amount of time you will gainfully labor, doing as thou wilt. Do not say that this is not labor, because it has no disutility, for that would make it a necessary truth that labor has disutility. Musm't labor, at least, be an activity? The hired person is always acting. What he sells is a right; the right to watch him. It might even be more particular labor he does; his employment might require him not to do some trivial act A he cares nothing for. Notice how the particular labor I have described differs from almost all other labor. It is a featuJ"e of almost all labor that it is incompatible with the simultaneous doing of other activities which the person wishes to do and which have some value for the person. Some consumption takes time, and cannot be done simultaneously with some other activities; for example, listening to Beethoven quartets and working a steam drill, lolling on the beach and teaching a philosophy class. Therefore, a
person with multiple desire-s, some of which cannot be satisfied s.imultaneously with the particular labor he does, will want leisure time in which to satisfy these other desires. The value of an hour of leisure to the person will be the value to him of satisfying the most highly ranked of his desires or combinations of desires which can be satisfied in an hour, whose satisfaction is incompatible with the simultaneous doing of his particular form of labor. We do not need the additional assumption that labor has disutility. Even if labor is a good, we can still have tradeoffs of labor and leisure; that is, tradeoffs of labor with the other goods that can be had only in that person's oonlaboring time. (Thus, Austrian writers need not deny that work may itself have intrinsic satisfying quality.) These are empirical facts of a high level of generality: (a) there are multiple goods; (b) most people's labor (what others will pay them for) is incompatible with their simultaneously achieving some other goods (as they judge them). These other goods cannot be achieved (as well) simultaneously with the labor; (c) in this area, beneficial tradeoffs are possible. Copyrighted material 138 -<> CHOICE AND UTILITY Now consider time-preference in the light of these.44 If a person knew
the moment of his death, and knew each of the desires that he could satisfy, then the incompatibility of various desires (the impossibility of their simultaneous satisfaction) would require him, if all the desires are to be satisfied, to stack desires back from the last moment, far enough so that they would all fit in. This, however, will nor constitute timepreference. Suppose uncertainty is introduced; the uncertainty usually mentioned in this context is uncertainty about the moment of one's death ("eat, drink, and be merry for tomorrow we might die"), and about whether a currently possible consumption will rontime to be possible in the future. But it is another uncertainty I wish to emphasize here (if only because it is generally ignored), that concerning which other current desires (or which desires to be acquired in the future) you may come to be in a position to satisfy in the future. This uncertainty will lead a person to prefer satisfying some desire or other now rather than postponing all their satisfactions until later. For it might tum out later that another stronger desire can only be satisfied then, which is incompatible with the satisfaction then of any of the desires which it is currently possible
to satisfy. And this might be true for all the "laters." (Add to this the uncertainty about whether it will remain possible to satisfy one of the current desires.) If you don't satisfy one of the desires now, you might never do so, for later it might be outran.ked by an incompatible desire. In these terms, there is nothing to gain by postponing the satisfaction of a desire, and there is something which might be lost. Satisfying some desire now, therefore, weakly dominates (in the sense of decision theory) postponing the satisfaction of all desires until later. Hence we have the rationality of an analogue of Mises' time-preference, though not of that which involves a preference of greater current consumption and less fu ture consumption over uniform consumption (summing to the same as the previous alternative) throughout time. The opportunity cost of failing to satisfy any desire now is the value of that desire, if any, current or future, which gets crowded out later. Since there might be such a crowded-out desire, there might be this opportunity cost. Given rwo desires which can be satisfied now or later, the greater need not be satisfied first; the opportunity costs of satisfying it 6rsr might be larger than those of satisfying it second. The opportunity
cost analysis explains the sequencing of the satisfaction of desires. There is no opportunity cost to satisfying a particular desire now rather than Copyrighted material Figure 5.3 I On Austrian Methodology ~ 139 II Time Early Reward Ill Later Reward none, unless we imagine a reason to want it later rather than now. But starring, as we suppose, with no such preference either way, dominance considerations give us a time-preference. But d:tis does not appear to be the whole story about rimepreference. Such preference bas been discovered in animal experiments, where the effectiveness of a reward declines with its distance forward in time in accordance with some concave curve. U the curve is more concave than an exponential one, for example, if it is a hyperbola, then we have the possibility of two such curves crossing. (See Figure 5.3.) The height of a curve at each point in time represents how valuable the later reward is
tO the organism at that point in time. At each rime an organism chooses to occupy the highest curve. Thus, in the diagram drawn, at time Ia person would choose to wait for the larger later reward rather than take the earlier smaller one which excludes getting the larger one, at time n the organism would choose the earlier smaller reward even though if it forgoes this and waits, the larger one will become available, and if the organism can somehow get beyond the time of the smaller reward, it will choose and reap the larger one. Thus, hyperbolic curves which cross are able to explain in an illuminating fashion the making and later breaking of resolutions (the first at time I and the second at time D), problems of self-control (at time II), etc.45 The time-preference found in animal experiments is not, I assume, to be explained by their performing rational calculations, even implicitly. How is it r.o be explained? If my earlier dominance argument about satisfying some desire or other in the present is correct, then an organism Copyrighted material 140 -<> CHOICE AND UTILITY which does th is and exhibitS that sort of time-preference will, on the average, satisfy more of its desires. There would be other arguments
as well which show that in a wide range of situations, timepreference would lead to greater total desire satisfaction. In an uncertain world, where contingencies frequently interfere with obtaining particular future satisfactions, time-preference enables organisms unable to reason about such contingencies to satisfy many of their desires. Supposing such a time-preference tendency arose by random mutation and was transmitted genetically, and that the desires involved themselves were connected with survival to reproductive age, ability to protect progeny, etc., then time-preference would be evolutionarily adaptive, and would be selected for in the process of evolution, once it appeared. If some such explanation accounts for its presence in lower organisms, it is reasonable to think that we too have some genetically based time-preference. The evolutionary process bas built time-preference into us, for within that process the rationality of timepreference is reflected as adaptive value. My account here has been very sketch), but some evolutionary account of time-preference in lower organisms must be given, and presumably we will be, in these ways, continuous with these other organisms. Economists should no more he$itate to use this biologically based fact
about people than Mises hesitated to incorporate the general (non-a priori) statement that labor has disutility. However, it does leave us with a puzzle. The evolutionary process builds time-preference into organisms who do not calculate, as a (rough) rule of thumb to approximate what calculation would lead to. Discounting of future goods is a surrogate for the calculations of the dominance argument (plus perhaps other maximizing arguments). Let us now consider the situation of organisms who do calculate, and who can in their deliberations take into account various future contingencies. If these organisms (call them people) do such explicit calculation, and also feed into these calculations magnitudes of (future) desires which have been discounted to take such calculative considerations into account already, then isn't there doublecounting, or rather, do~<ble-discountit~g? Time-preference first discounts, and our later calculations explicitly take into account factors and lead, in effect, to explicit discounting. When we ourselves can take various contingencies explicitly into account, do we want to be bound by the type of discounting found adaptive as an average rule of thumb i.n the evolutionary history of the spe.cies? Shouldn't we try to correct for such discounting as is built into us, and put all discounting into our
calcuCopyrighted material On Austrian Methodology ~ 141 lations? Or should we instead eliminate all such discounting from our calculations? The problem of double-discounting arises when our calculations utilize the current value to us of the future goods, which value we explicitly discount in order to take account of the uncertainties of the goods' being realized. The discounting is double because the current value to us of the future good exhibits time-preference, an implicit discounting of the future selected for in the evolutionary process because of the adaptive advantage of discounting for future uncertainties and other such factors. This problem of double-discounting would be avoided if our calculations, instead of utiliz.ing the current value ro us of future goods, utilized some measure of how much the good would be worth to us at that rime in the future when it would be realized. There would be no doublecounting in explicitly discounting such future values which have nor yet been discounted even implicitly, for uncertainties. However, a measure according t.o your preferences now of how va luable a future good will be to you then does not fit well with Austrian theory's dose typing of preference to actual choices, or even to
rhe loosening in terms of subjunctives which we suggested earlier.'6 So the problem of double-discounting is an especially pressing one for Austrian theory. Having built-in time-preference and also realizing what the evolutionary explanation of such time-preference is, how should we, as rational beings, behave? Copyrighted material Copyrighted material PHILOSOPHY AND METHODOLOGY Copyrighted material Copyrighted material 6 SOCRATIC PUZZLES Ignorance Socrates claims he does not know the answers to the questions he puts, and that if he is superior in wisdom this lies only in the fact that, unlike others, he is aware that he does not know. Yet he does have doctr ines he recurs to (that it is bener to suffer injustice than to do it, that no one does wrong voluntarily but only from ignorance, 1 that a doer of injustice is bener off being punished for it), and he shows great confidence in these judgments. When Polus says it is easy to refute the view that the unpunished unjust man is more wretched than the
punished one, Socrates replies that it is "impossi.ble. For the truth is never refuted" (Gorgias, 473b). And Socrates tells Callicles that the doctrine that it is more evil, ugly and shameful to do injustice than to undergo it " is secured and bound fast ... with arguments of adamant and iron" (508e-509a). Is this supremely confident Socrates merely being ironic when he elsewhere denies that he knows? How are we to understand what Gregory Vlastos terms "Socrates' central paradox, "2 his profession of ignorance? Socrates does not (generally) deny knowledge of the Socratic doctrines (it is bener to suffer injustice than to do it, ere.) but of the answers Copyrighted material 146 ~ PH I LOSOPHY AND METH ODOLOGY to the "What is F?" questions he pursues with others. His superior wisdom then resides in his knowing that he doesn.'t know, and cannot pro duce, the correct answer to these "What is F?" questions. Not knowing these thiJ1gs is compatible with his knowing other things. There is no contradiction in his saying he k11ows be does not know the nature of the relevant F's, for that knowledge he has is not itself the answer to one of his "Whar is F?" questions.
This limits the scope of Socrates' claim not to know, bur what precisely does he mean by this profession of ignorance? It has been sug gested that Socrates does believe he has true beliefs about the matters he discusses, but that he lacks something in addition to true belief that (he thinks) is necessary for knowledge or for a certa in kind of knowledge. Terence Irwin holds that Socrates claims only true belie.3 Vlastos objects (didn't Socrates think some of his beliefs justified?), and he proposes instead a distinction between two kinds of knowledge: elenctic knowledge, which one has when one's (true) belief has survived repeated t.esring in the process of elenctic inquiry; and certainknowledge, which one has only if one knows with infallibility and certainty. Socrates (says Vlastos) is consistent: when he clainls knowledge, be is claiming only tbe first kind; when he denies he has knowledge he is denying he has the second.4 Those who heard Socrates say that he doesn't know the answer would have taken him to mean that he doesn't know the truth, whereas Irwin and Vlastos interpret him as saying that he doesn't know the truth. Their interpretations might rescue the literal truth of Socrates' denial but they leave him generating misleading inlplicatures. I suggest that when Socrates says he himself doesn't know the
answers to these questions, he means not just that he doesn't know the truth but that he doe.sn't have the truth; he doesn't even have true belief. One instance where commentators have suggested a part icular whatisf truth that Socrates possesses concerns piety:1 Socrates leads Euthyphro to the suggestion that piety is performing a service to the gods. Vlastos suggests that Socrates would have gone on to speak of a service to the gods that the gods want done and would do by themselve.s if they could. What specifically is this service? Vlastos suggests that Socrates would say it is improving our own souls, and helping to improve the souls of our fellows.6 (VIastos states that Socrates has arrived at a new and religiously revolutionary, nonritualisric conception of piety.) We can add that a pious act would be one that does this improving not by accident or solely for some different reason but (in part) beCiluse it is desired by the Gods. This would make Socrates, pursuing his ordained mission, Copyrighted material Socratic Puzzles -& 147 the most pious person in Athens. Should we call it another paradox of Socrates that he, the most pious person, was prosecuted for impiety? (The first paradox is that the founder of Western rationality was
set upon his course by an oracle's statement.) Or is this Socratic irony? Can Socrates offer a satisfactory account7 to answer the question: what is piery? We already have suggested one component in addition to the ones Socrates mentions or hints at: doing the pious act because it is desired by the gods. But the investigation into the nature of piety cannot stop here. Further questions will arise. (How precisely do we delineate the needed connection between our doing the act and its being desired by the gods? Will any desire by the gods for something excellent do, or does it matter why the gods have this desi.re? Would it nevertheless be pious if rhe gods desired something bad of us? What is the nature of the good to which the gods adhere? Must a pious act concern what the gods literally cannot do for (or accomplish by) themselves? Precisely what counts as an improvement in the soul's state? And so on.) In the same paragraph where be credits Socrates with reaching (and leading Euthypbro up to) a new conception of piety, Vlastos says (pp. 175176): To derive from this a de1inirion of piery, Euthyphro would then have
had tO generalize, contriving a formula that would apply, not only in Socrates' case, but in every possible case of pious conduct. This is a tall order, and it is by no means clear that Socrates himself would have been able ro fi ll it. But this technical failure would not shakewould scarcely touch the ceotral insight inco the nature of piery ... Whether or not a fonnula could be devised to encapsulate this insight in an elencrically foolproof definition ... Let us consider another concept., that of knowledge, a concept not discussed in the early dialogues bur one which (as we have seen) the recent commentators lean upon. The history 1 shall rehearse is familiar but this particular example will serve us lateL Is knowledge true belief? No, says Plaro in the Theaetetlls. Is it true belief with an account or true belief with adequate reasons or justified true belief? No, shows the Gettier counterexample. Is it a rrue belief that p that is caused (or explained in part) by the fact that p? No, in view of the example of the person floating in a tank brought by psychologists to believe be is floating in that tank. Is knowledge belief thar tracks the truth, that is, (1) a bel ief that p, where (2) p is true, (3) if p were nor true the person would not believe it, and (4) if p were true the person
Copyrighted material 148 .... PHILOSOPHY AND METHODOLOGY would believe it? Leaving aside additional complications about methods and ways of knowing, I said yes in Philosophical Expumations, but others have raised objections. a So what is knowledge, what are the necessary and sufficient conditions for knowledge that p? A reasonable philosopher today might say that, in view of the difficulties thus far encountered, he just does not know. And by this he means not that he is prepared to put forward conditions that he believes are correct without being certain that they are, but that there are not conditions he is prepared to put forth, not conditions he believes will not succumb to further counterexamples and complications. This reasonable philosopher may believe, on inductive grounds, that since all previous accounts have fallen so will any one he can formulate, or he may already know of example.s or objections that he cannot yet handle. He knows that he cannot yet formulate "an elenctically foolproof definition." He does know something else rhat more naive people do not; he knows that knowledge is not just true belief, or justified true belief, or ... Moreover, be may know more than they do, in the sense that the
account he would put forward if forced to mention the most adequate one he knows of, while itself false and inadequate, is more adequate, is closer to the truth (whatever that true account is) than the earlier accounts or than the general accounts that would be proposed by those who do not know the history of successive accounts. Socrates himself, I suggest, was in this position with regard to the topics he discussed. I do not claim, anachronistically and absurdly, that he knew all of the difficulties revealed in the future discussions of knowledge or of any other topic. But I do claim that he was more thoughtful and more acute than his Athenian contemporaries in probing various topics. He knew more than his fellows, he knew precisely where their accounts failed, and he could formulate a more adequate account that did not fall to those objections. However, he also knew of or believed there were objections that would show his current best account to be inadequate, that is, false. He didn't know what piety, courage, justice, etc., were. He didn't have a completely adequate true belief about any of these ropic.s. He didn't have accurate necessary and sufficient conditions for any of these notions, much less a completely accurate understanding
of how aU of these notions were interrelated. Hence, Socrates was not being ironic when he denied knowing the answers. He did not know. And he knew he did not. Copyrighted material Socratic Puzzles ~ 149 Blench us In his discussions with others (and presumably in his own internal thinking) Socrates uses the method of elenchus. Inconsistencies in belief are uncovered, thus impelling the person to modify his beliefs. An inconsistency reveals that we must have some false belief, but we do have many false beliefs anyway-few of us believe all of our beliefs are true-so why is inconsistency especially bad? Provided we are careful not to deduce any consequences by forms of inference that depend upon the beliefs being inconsistent, why should inconsistency trouble us especially? An inconsistency or mistake about the important (evaluative) con ceprs Soc.rares discusses is dangerous, either because it itself constitutes a defect in rhe state of one's soul, or because it easily can lead ro one. Other apparently good rhings (wealth, power, etc.), without guidance by a correct knowledge of the good, can lead ro one's harm.9
The Socratic search for the definition of a concept F is a search for necessary and sufficient conditions that provide a standard rhar can be utilized to decide whether F applies in any given case. However, when people often are able to appl) the concept F to particular cases, judging whether or not it applies, why is an explicit standard necessary? An e.xplicit standard can help to judge difficult cases or borderline ones. Moreover, explicit understanding of what makes something F can prevent being thrown into confusion by arguments that seem ro show that F is impossible or never applie.s. Even someone who by and large can classify cases correcrly as being knowledge or not, when faced with the skeptical argument, "you cannot know that you are not floating in a tank that generates all the experiences you have had; therefore you know almost nothing," might conclude rhar indeed he does nor know most of what he thought he knew. An explicit understanding of what constitutes knowledge (or F) makes your beliefs about knowledge (or F ) more stable and less prone to being upset by inadequate arguments. About the important evaluative concepts, connected wirh the nature of the good and the best srate of one's soul, it is especially
important not to be misled, given Socrates' view that knowledge is vinue. Morover, the stability of one's correct belief might itself be a component of a wellordered soul. What makes Socrates' method of elenchus a method of teaching the truth? What must hold true if elenchus is to be an effective way of Copyrighted material 150 -~> PH ILOSOPHY AND METHODOLOGY reaching the truth? In "The Socratic Elenchus,"10 Vlastos offers a very illuminating discussion of this matter. First, it is possible that I hold a false belief about some matter though my beliefs are not inconsistent.. I believe that a particular undergraduate student comes from Seattle. Elenclms will not help here, in the absence of further information which I don't yet have. (For example, I might also believe that that student does know what city he comes from. lf you now tell me, and I believe this, that the student believes he comes from Minneapolis, then my beliefs will form an inconsistent trio.) Socratic eler~chus doesn't get a grip to impel a change in belief unless there are inconsistencies. Or so the commentators tell us. We can imagine that a questioning procedure also is able to impel changes
because of tensions, disharmonies, and complications that it uncovers, even when these do not constitute strict inconsistencies. A person can be made uncomfortable if his beliefs in combination are very improbable, or if to harmonize them he has to assume some other very improbable (or inordinately complex) funher group of statements, etc. The discussion and probing of issues can utilize norms of belief, in addition to the norm of avoiding inconsistency, to impel change in belief. What must hold, thougll, for the stringent elenchus that works only upon inconsistencies to do irs work? When a person holds a false belief, what guarantees there will be inconsistencies? Socrates assumes (suggests Vlastos) that about the evaluative matters he discusses, the person will hold some rrue beliefs inconsistent with the falsehood. (B) For any person and for any false bel ief q about the requisite subject matter, the person entering into elenchus also will hold some true belief p, inconsistent with q Y' (How did he acquire these true beliefs? By recollection?) An inconsistency is enough to get the elenctic process of change going, but what points that process toward the truth?12 When an inconsistency is discovered among a person's beliefs p and q, he can remove that inconsistency by denying the true one p or by denying the false one q. What stops him f.rom denying the true statement p?13 If the
belief that p and the belief that q are each merely beliefs, they seem on a par. What, then, prevents elenchtts from leading him further from the truth? I suggest the following. (Here, perhaps, I trespass upon Plato's topics rather than Socrates' assumptions.) Elem;hus is a reliable vehicle for getting roward the truth wheo a person knows (and doesn't merely hold as true beliefs) truths about the subject matter sufficient to introduce Copyrighted material Socratic Puzzles ~ 151 inconsistency with any false beliefs he holds. So we strengthen Vlastos' assumption B to: (K) For an)' person and for any false belief q about the requisite subject matter, the person entering into elenchus also will know some truth p, inconsistent with q. But why does this knowledge make (it more likely that) the process of removing inconsistencies (will) move toward the truth, and not further in the direction of consistently false beliefs? The former will happen if knowledge has more stability (tenacity, stickiness) than belief, true or false. What this means is the following. (R) When a person is faced with an inconsistency among statements, and resolves this inconsistency by giving up one statement, he will (tend to) give up a false statement that he believes rather than a
true one that he knows. When beliefs conflict., something about knowledge makes the person adhere more to what is known than to a false belief. 14 If this is so, then the elenctic process will (tend to) lead someone toward the truth. Even if he just is more likely to give up a false belief rather than something known, and so sometimes does not do that, still, provided he has enough of such knowledge so that conflicts with his false beliefs will continue, the higher stickiness of knowledge will (tend to) win out in the end. In this elenctic competition, knowledge is more fit than false belief, and so tends to be selected for. Do the necessary and sufficient conditions for knowledge guarantee its greater stability or stickiness, or is that just knowledge's fortunate concomitant? In the tracking account of knowledge described earlier, the fourth condition states that if p were true, the person would believe it. The belief that p is stable under small enough perturbations. In terms of the modelling of subjunctives by possible worlds, he continues to believe p in those p-worlds, the p -band of worlds, closest to the actual world. These include the worlds where p comes into conflict with a false belief in q; even there the person believes that p. Thus this fourth tracking condition gives us (a portion of) the requisite stability
and stickiness of knowledge. I do not want to insist that this particular tracking condition is a part of the (correct) analysis of knowledge. But if it is, or if some modification is, or if some different condition is that Copyrighted material 152 ~ PHILOSOPHY AND METHODOLOGY gives knowledge greater stability than merely false belief, then the crucial condition wi ll hold that makes elenchus an effective route toward the truth. I have said that Socraric ignorance consists in not knowing (and not having a true belief stating) the answer to the "What is F?" question. And I have not denied that Socrates believes that he has reached the truth about some evaluative matters, e.g., concerning the doctrines that it is better to suffer injustice than to do it, that no one does wrong voluntarily but only out of ignorance, that it is better for someone that he be punished for an injustice he has committed. Elenchus is an effective route to reaching such evaluative truths, on this account, only for persons who satisfy conditions (K) and (R), that is, only for people who already do have some knowledge about these evaluative matters. So elenchus (if it is reliably effective) cannot be the only route tO knowledge about these matters.
Why is elenchus not also an effective route to reaching a correct answer to "What is F?" Is it that we do not have the requisite knowledge concerning the nature ofF to satisfy condition (K)? We wiU have some knowledge about F, though, for those concepts F that are involved in Socratic doctrines that elenchus can yield. Another possible answer would be that elenchus is able, in principle, to reach such a correct answer after a finite number of humanly takable steps (proposa ls, objections, modifications, etc.), but that number is so large the process is so long and arduous, that no one has yet reached that resultant stage.15 Particular evaluative truths are less complex than the complete truth about rhe nature of the evaluarive concepts; they can be reached early in the elenctic process, whencefotth they remain stable. Why are the answers to the what-is-F questions so complex? Analytic philosophers should not tax Socrates with these questions when they themselves lack a satisfactory explanation for why their own stock of correct analyses is so meager. Improving Souls In "The Paradox of Socrates," Gregory Vlastos accuses Socrates of being so strongly wedded to his method of elenclms as to evidence "a failure of love."
I have already argued that he does care for the souls of his fellows. But the care is limited and conditional. U men's souls are to be saved, Copyrighted material Socratic Po n ies ~ !53 they must be saved his way. And when he sees they cannot, he watches them go down the road to perdition with regret but without anguish. Jesus wept for Jerusalem. Socrates warns Athens, scolds, exhorts it, condemns it. But he has no tears for it ... One feels there is a last zone of frigidity in the soul of the great erotic; had he loved h.is fellows more, be could hardly have laid on them the burdens of his "despotic logic," impossible to be borne.16 Does Socrates know of (or at least, believe there is) some other way souls can be improved? If not, then in adhering to his way he is, at worst, making a mistake due to ignorance. Indeed, it is not enough for Vlastos's charge that Socrates know of another way to improve souls. That other way must fit Socrates' particular talents and skills, it must be one that Socrates can follow (whether exclusively or in mixture with his own way) with at least equal success and effect. That Buddha too had a way of improving people- suppose that Socrates knew of this waydoes
not mean that Socrates willfully follows his own. Of course, Vlastos is aware that firm Socratic doctrines seem to leave Socrates (or anyone else) no other way.17 To improve the state of someone's soul is to make him more virtuous. That requires (and requires only) bringing him into a state of greater knowledge of the good. (Knowledge is virtue; vinue is knowledge.) True belief about the good is not sufficient. Perhaps knowledge of the good is an intrinsic pan of a virtuous soul. In any case, true belief is unstable, hence an insufficient guide and gyroscope. Moreover, Socrates cannot simply tell people the truth; they cannot reach a knowledge of the good through the sayso and authority of anyone else. At best, that will give them an unstable true belief, open to upsetting by the next contrary "authority." The only way to aid their knowledge of the good is to get them to think through issues about the nature of the good and its relation to other things, that is, to engage them (or to get them to engage themselves) in elench11s concerning these matters. Socrates can give others a method of inquiry only by showing it to them in acrion; thereby he launches them upon their own path of inquiry. Given his doctrines and views, Socrates is not insisting upon saving his fellow citizens IJis way. There is no
other way. Inquiry arises because of puzzlement, john Dewey said. People who are quite confident of the truth of their very extensive views are un likely to engage in probing inquiry about these matters. The first step for Socrates, then, must be to show these others that they need to think about these matters, that is, to show them that what they already are Copyrighted material 154 ~ PHILOSOPHY AND METHODOLOGY thinking (or unthinkingly assuming) is quite definitely wrong. Why will someone who is confideD! of his views enter such discussions? TravelLing reachers may do so to show their wares. Others may have tO be shamed or bulldozed into such discussions, for their own good (and the good of the onlookers). Even apart from its further effects, Socrates may hold that it is better for a person with false views about these matters to be refuted, just as it is better for someone who has committed injustice to be punished .. 18 To improve people's souls, then, Socrate.s will have to make a pest of himseU, buttonholing unwilling conversants, forcing them into physical and then inrellectual corners.19 Being a pest, an expression of his caring, puts Socrates into danger, or increases the danger he already is in.
Socrates has doctrines but what he teaches is not a docrrine but a method of inqu iry. (He does think or hope that sustained inquiry will lead others closer to his doctrines.) He teaches the method of inquiry by involving others in it, by exhibiting it. Their job is to catch on, and to go on. Socrates shows something more: the kind of person that such sustained inquiry produces. lt is not his method alone that teaches us but rather that method (and those doctrines it has led him to) as embodied in Socrates. We see Socrates within his inquiries and his inquiring iDleractions with others; we see the way his inquirie.s shape and infuse b.is life, and his death. Socrates teaches with his person. Buddha and Jesus did as well. Socrates is uoique among philosophers in doing this. Socrates teaches us beyond the arguments he offers. The arguments presented in the Crito are weak, after all.20 They are insufficient, individually and jointly, to show that Socrates has an obligation to the city not to escape. Yet who would say that Socrates should have accepted Crito's offer? Socrates is right that scurrying off to another city to gain a few more years would be the wrong thing for him to do. To escape would show an unseemly concern with death. It would not improve the state of his soul. The unflinching courage and integrity and c.almness
and clariry with which Socrates faces and meers death teach us how death is to be faced. This incarnation of rationa(jty and courage and justice uplifts and ennobles our souls; we are the better for admiring and loving him. About the best state of the soul, too, Socrates shows us more than he says. Arguments are presented in the Gorgias rhat it is better to suffer injustice than to do it, but these involve an aesthetic component. The unjust soul is disordered and unlovely; it is ugly. Socrates invokes here Copyrighted material Socratic Puzzles ~ 155 (without explicitly developing) a general notion of value within which the moral and the aesthetic fit. According oo this general conception, souls cao be more or less lovely, more or less beautiful. We learn what this beauty of soul is, as did his hearers, not by being presented with an explicit theory but by encountering Socrate.s.21 Socrates' reaching and irs convincingness depend significantly upon the embodiment Socrates is, and the way in which this is displayed. That is not something that Socrates himself would have said or believed. His reasoning and arguments were supposed to bear the full weight, at least when combined in elenchus with what the other person already knew or believed. This additional way that Socrates
teaches us---<::all .it the method of embodiment-stands alongside his method of elenchus, and also is worthy of sustained attention. Copyrighted material 7 EXPERIENCE, THEORY, AND LANGUAGE For the past twenty-five years Quine's powerful and deep philosophical views have been, deservedly, the focus of more attention, discussion, and controversy than those of any other philosopher in the Englishspeaking world. His powerful reductionist philosophy attracts those who wish to fit things into a sleek, limited, tight framework (quantification theory, loved by Quine and central to his philosophizing, also exemplifies the virtues he seeks in his philosophy as a whole), while it repels those who think it excludes much that is important about our intellectual lives or ourselves. I myself, strongly subject ro both the attraction and the repulsion, am ambivalent about his view, and about such reductionist views. The basis of these contrasting pulls and pushes is itself an interesting and different topic, but to metaphilosophize on the sources of our (my?) philosophical ambivalence would be inappropriate on this occasion. Instead, I shall discuss some of Quine's themes, sometimes critically assessing his positions, and sometimes sympathetically
defending or extending them. This will result in a clouding of rhetorical force, as perhaps befits an essay written out of ambivalence. Copyrighted material Experience, Theory, and Language ~ 157 Quine is the theorist of slack." Data underdetermine theory, there is leeway about which component to modify when data conflict with a theory, and translation is indeterminate. Also, theory underdetermines the world (the doctrine of ontological relativiry). These particular theses ramify throughout philosophy, and the fact of slack itself is of great philosophical interest. We cannot strip away the conceptual trappings sentence by sentence and leave a description of the objective world, but we can investigate the world, and man as a part of it, and thus find out what cues he could have of what goes on around him. Subrracting his cues from his world view, we get mao's net contribution as the difference. This difference marks the extent of man's net conceptual sovereigntythe domain within which he can revise theory while saving rhe data.' I. Experience and Language Duhem and Experiential Meaning What explains why, vide Duhem, our statements don't face the rri bunal of experience singly? (Is it cowardice?) Would a language whose statements did this, a nonDuhemian language, be impossible for
us? If not, in what ways would such a language be inferior? I do not ask these questions from the standpoint of what Quine calls first philoso phy. Within our current theory, what explains Dubem's facts? Our experiences are dependent upon the state and condition of our sensory apparatus, on our orientations in the world, and on what the world is like, including those aspects of the world which are distin guished as conditions of observation. Furthermore (one is tempted to continue), our experiences of the world depend upon many interconnected facts about the world. Since experience.s are dependent upon so many things, which may vary independently, a being would have to know enormous amounts to have any hope of formulating statements with individual experiential content. But this pu.rported explanation dissolves when we notice that we have no clear language-independent notion of number of facts. We count facts in our language. Our view that our experiences depend upon many Quine's view coumenances slack galore; how then c-an some have held it to be rigid? It CJCdudC$ what on other '<"iews fixes things in determinate place. {Apparend)'t there is nor enough slack ro admit, as compatible with the data, a theory of fixed meanings.)
Copyrighted material 158 -t> PHILOSOPHY AND METHODOLOGY facts derives from our not having single statements with individual experiential content. We cannot explain why our statements do not singly connect with experience by the multiplicity of facts upon which experiences depend. Nor can we appeal to how complicated the facts are. For if we measure the degree of complicatedness of a sentence, languageindependently, by the nature of its connection to experience, then long conjunctions which uniquely map onto experience will come out as quite uncomplicated. The statements (in English, complicated conjunctions) which (arguably) singly have experiential content wn athwart those which are important to us, and have been in the course of evolution. The precise experiential content of " there is an unrestrained tiger in front of me" is less important than avoiding knowledge of the experiential content of "I am being eaten by a tiger," a coment which once known is unlikely, in any event, to be conveyed to others. About such dangers as there being a tiger present, type-two errors (accepting an hypothesis when it is false) were preferred to type-one errors (rejecting an hypothesis when it is true) and were selected for. Better to Aee unnecessadlv than to Stand and refine the experiential content of one's statement!
The interest of many other statements stems from the diverse ways they 6t into people's plans. Trees before you can be walked around, hid behind, climbed up, and cut down. What one wants to know is whether there is a tree there. That multi-purpose Statement does not have the separable experiential content had by conjunctions about trees and lighting conditions and the state of one's eyes and lack of nontransparent obstruction berween you and it, and so forth. What it has is utility. The facts that evolution seleccs for attentiveness co, facts which have multi-purpose utility in our lives, lie athwart those with specific experiential content. If language is geared to the former facts, and its component sentences state them, then these sentences will not ha,,e (be exhausted in) separate experieotial content. Relative to the grid set by our evol utionary past and activities, sentences with separate experiential content will be very complex, and our experiences will depend upon the conjuncrion of many (language-identified) facts. Since language is geared, for evolutionary reasons and reasons of practical activity, to facts which lie athwart (those facts which correlate with) experiences (as represented perhaps by 6rings of nerve endings), it would be surprising if some significant unit of language was
marked off via experiential content. In particular, it would be very surprising if Copyrighted material Experience, Theory, and Language ~ 159 the unit of meaning of the language were those entities (whatever they might be) with separate CXJ)<'riential content. Why should something have evolved or been formulated or come to be, whose units were that? U we were designing a language to be taught to organisms who had not developed their own, in order to aid them in surviving and acting in the world, would it be organized around units of distinguishable experiential content? Would distinguishable experiential content be a theoretically significant notion for a language with that purpose? The claim that something with separate meaning must have separate experiential content is a philosophical imposition which rests on considerations about learning a language, to which I turn below, and on rhe question "how else ca11 something have meaning?," which I take to be a symptom of being trapped by a picture. The explanation of why Duhem's thesis holds, makes implausible the view that the units of meaning are the units with separate experiential content. That the facts that language is geared to state lie athwan those with
separate experiential contell! is understandable and ex plicable; that the sentences sta.ting those facts which language is geared to state should lie athwart the language's units of meaning is not to be believed. Not only are our individual sentences without separate experiential content;2 there is no reason to think our total scientific theory has it either. Given the range of factors (as counted by our current partitioning of the world) upon which our experiences depend, including the facts of sensory psychology, one would have to know all of these (and that they were all), as well as the theory which connected them to our experiences, in order to know that one's total theor)' had experiential content. (And would there be open ceteris paribus clauses in one's body of theory?)3 In the history of mankind, when did we reach the point of having theories extensive enough to have separate experiential conrent? Before then were all sentences without meaning? Through assuming separate experiential content must be somewhere in the language if language is to be meaningful, and not finding it in individual sentences, one might think it must lurk in the total set of sentences we would assere. Not there either. U our total theory did have separate experienrial content, we
might think to assign experiential coment to its individual sentences as follows. Consider the set of sentences of rhe language of length < n (setting 11 high enough so that we don't exclude as too long any sentences we are interested in), and take the power set of this set. Within each set in the power set order the sentences alphabetically, and then order these Copyrighted material 160 -t> PHILOSOPHY AND METHODOLOGY sets alphabetically. For any sentence S and any set X within the ordered power set, we may consider the experien6al content (observa6onal consequences) of (S] U X. (S may already be contained in X, or its denial may be, or there may be no observational consequences of this particular union.) Against the structure provided by the ordered power set, we may now specify the experiential content (observational consequences) of any sentenceS as O, = ({TJ), (T2I , ... , (T.I), where T1 the observational consequences of {S] u the ith member of the ordered power set. The observa6onal consequences of S would be specified as irs observational consequences relative to all other sentences, and translation between languages would depend upon isomorphism between the components of such structures. Is this then a way to formulate separately the experiential content
of individual sentences? Actually not all (or even many) possible sets of sentences are conditioned to experience; instead only one evolving set is (that is, a sequence of sets). Since it is this actual process of conditioning to experience which, on Quine's view, gives the tie to e.xperience, we cannot assume that other large conjunctions do have this tie to experience. For this assumes the conjuncts come already with their own individual (relative) observational content and that the content of the conjunction results from fitting together the individual contents. (How else would a conjunction, itself unconditioned ro experience, have experiential content?) Quine, therefore, would reject this method whereby the individual observational content of the sentences we accept is to be specified relative to an ordering of all other sets of sentences, I" conjecture. For these total sets (all bur one) will nor, on his view, have observa6onal conrent since they will not have been actually conditioned to observation. Duhem and DetermitUJte Content Since those facts we are interested in, for evolutionary reasons, lie athwart facts with specific experiential consequenc.es, a language limired ro stating these latter facts would fail to serve our purposes. But what would such a language be like? Would this be a oon-
Duhemian language: the applied lower functional calculus with one atomic predicate 'n fires at 1 ,' where 11 ranges over nerve endings and 1 over instants (or intervals) of rime? Label all nerve endings and specify the person. This language is suggested by Quine's remarks about information (Philosophy of Logic, p. 5), directed ro showing that we caonot Copyrighted material Experience, Theory, and Language ~ 161 specify the meanings of our English sentences individually in this language. All sentences in this language have determinate informational content (Philosophy of Logic, pp. 4-5). But, whether in the first or the third person, these are highly theoretical statemems of sensory psychology which are themselves subject to Duhem's thesis. (Is this very sophisticated language of psychological theory also the observation language? With community-wide agreement about its sentences?) These sentences may perform for the "naturalistically inclined" epistemologist some of the functions of introspectible sense data, in that their obtaining underlies a person's observation sentences; but they are unlike sense data statements in not being incorrigible. being theoretical, and being subject to Duhem's thesis. Being subject to Duhem's thesis does not, apparently, insure lack of determinate informational
content or failure to state a separate fact of the matter. Therefore, Duhemian considerations do not show that sentences do not have separate meaning. (Such considerations would do so only along with a verification theory of separat.e meaning, but that theory would be difficult to square with Quine's remarks on the objective information of sentences.) Quine also considers the possibility of specifying the coment of sentences by specifying which distributions of elementary particles they are true in, and notes that we cannot do this (Philosophy of Logic, pp. 4-5). But there are sentences which are already, without analysis, sentences about the distribution of elementary panicle.s. These sentences of physics would have determinate informational (factual) content surely, for if it were possible to translate other sentences into these, this (Quine says) would show that the others had determinate content. Yet still these sentences are subject to Duhem's thesis. For the statement that a certain distribution of panicles obtained could be rested only in conjunction with many other sentences. Let us say that a language's sentences have uniform determinate content if and only if there is some matrix of alternatives into which all the senrences are uniquely translatable. (Tbe one matrix provides the uniformity,
the uniqueness provides the determinateness.) Failing this, will the sentences of a language lack individual meaning? That would be too stringent a criterion. For even if two languages each satisfy the criterion (relative to two different matrices which do not "interfere"), it would exclude their union as containing individually meaningful sentences. Yet to admit any language which was the sum of languages which did satisfy the condition would exclude no language, without some constraint Copyrighted material 162 ~ PHILOSOPHY AND METHODOLOGY upon admissible matrices to prevent every atomic &agrnent of the language from providing its own matrix. It might be objected that when Quine says that some statement S about distributions of elementary particles or firings of nerve endings does have separate content, he is speaking from within a theory which contains and utilizes a sub-theory about particles, nerve-ending firings, ere.; the sentences of this theory, when viewed from the outside, however, have no separate content and are subject ro Duhem's thesis. Bur from withi11 a theory each of its statements (excluding the ambiguous) has a determinate content. Yet Quine does not merely say (leaving us to realize he's speaking &om within a theory) that every sentence has a
determinate content. "We can never hope to arrive at a technique for so analyzing our ordinary sentences as to reveal their implications in respect of the distribution of particles" (Philosophy of Logic, pp. 4-5). Our ordinary sentences are nor geared to talk of what may micro-reduce what acruaUy is their subject matter. Our interests lie elsewhere. (Would we know what someone who spoke such analyses of ordinary sentences mea11t?) The distributions of particles does nor constitute a privileged matrix into which we must translate our ordinary sentences. It is notable, though, that Quine thinks that sentences would have determinate meaning ("objective information") if we could analyze their implications in respect of the distribution of particles. One might think that determinateness is the crucial criterion of separate meaning, rather than separate experiential content, which the particle statements lack. But Quine, I think, is concerned with and favors two distinguishable things, determinate content and experiential content. So he would view with some favor a particle language whose individual sentences separately have the first though nor the second. Since separate experiential content is 011e way of providing uniform determinate meaning, his concerns converge in his treatment of determinate experiential content. Bur he has not yet given us a
rationale or motivation for being concerned with uniform determ.inate content. We have no argument for a sharp criterion connecting meaning with experiences. (Why is Quine so sanguine about holistic verificationism? " Epistemology Naturalized," p. 81.) Quine grants, moreover, that some sentences have separate informational content even though they are subjeer to Du.hem 's thesis and are nor separately testa ble, and we would expect the unit of meaning not to be connected with experiential content anyway. Hence, it seems unlikely that any interesting conclusion about Copyrighted material Experience, Theory, and Language ~ 163 the theory of meaning will emerge from Duhcmian considerations. 4 But something interesring might emerge from asking how we would expect a language to be organized, given the considerations adduced above to explain why Duhem's thesis holds. Language Learning and Indeterminacy One might think that considerations about language learning do show that the matrix of nerve-ending firings constirutes a privileged matrix into which our ordinary sentences musr be translated. For it is through such firings that we learn language; how can what we leam
transcend how we learn? And if meanings are anifacts shaped by us, will they not be fired in the kiln of our nerve endings? Comfort for viewing meanings as going beyond stimuli might seem to come from a much studied phenomenon: on the basis of a few firings we generalize along pre-existing lines in quality space, thereby transcending the stimuli from which we learn. (Psychologists speak of generalization gradients; these vary from species to species.) However, we do not thereby transcend that type of stimulus, which can be used to check that we do generalize along the same paths. It is striking that we do generalize alike past particular data in those cases we can check with other data. One might speculate that we move similar!)' beyond the data in other uncheckable cases, but unlike the former case, direct evolutionary selective pre.ssure would not explain the latter. Another way to detach (panially) meani11g from stimuli offers more promise. Consider specifying (not defining) X as that thing or mechanism which actually explains data D. 5 Here the reference of X is fixed by D and by the world. D may have separate experiential content, or it may have been originally explained, as X here is, but in terms of data
with separate experiential content, and so on. Let us suppose that X stands in the ancestral of the explains-relation to something with separate experiential content. Cannot the meaning of X be 6xed in this way, has it not been given separate meaning, even though clearly it itself does not have separate experiential content? It might be said that if we fix the reference of X in this way, we can't k11orv what X means or refers to. And it might seem as if the whole truth about the meaning or mechanism of reference of our terms must be exhausted by psychological truths about us. How can there be more to what I mean, to the meaning of the term as I use it, than is contained Copyrighted material 164 ~ PHILOSOPHY AND METHODOLOGY in my intentions, beliefs, etc.? There are numerous counterexamples to this apparently compelling view, e.g., Wittgenstein's on how what the person means by chess is fixed outside of his intentions and beliefs, Kripke's causal story about proper names, Quine on dispositions; and recently Putnam, in his illuminating talk about division of linguistic labor, has shown how uninruirive it is, how inefficient it would be, if we each had to carry around with us the complete meaning of
what we say.6 One might think to use the way in which external facts fix meaning to resol.ve Quinean indeterminacies. But before looking at details, note bow such views undermine one objection to Quine on indeterminacy. Many who object to Quine think it's perfectly certain what they mean, treating this as some incorrigible psychological fact about themselves. On the external determination view, of course, they can be wrong. (Add to this Wittgenstein's erosion of viewing meanings and intenrions as i.nternal psychological pelletS, thereby undercutting a platform on which one might think to stand in confidently attacking Quine.) So bow can they be so certain about what their fixed meaning is? A tale will be useful here, which also will serve to combat the view that within neurology lie.s determinacy of translation. Suppose we send rwo ships off in opposite directions, containing infant children and teachers. One group of teachers will reach the children English, speaking it to them. The other group of teachers on the other ship will speak stage-English, which is the stage-language of English. (This is the language, described by Quine, whose terms denote temporal stages of
enduring objects. It bas nothing special to do with the theater!) The words of stage-English are some trivial phonetic transformation of the corresponding English words, "Blegg" instead of "egg, n etc. Locked in a safe in the captain's cabin is a dictionary which stares the meaning of stageEnglish words and sentences in English. Now suppose almost exactly the same things happen on the rwo ships; for each sentence heard on an occasion on the English-speaking ship, a corresponding sentence wi ll be heard on the stage-English ship on the parallel occasion. (They are almost parallel worlds.) The children on the first ship will learn to speak English, and the children on the second will learn to speak a language which, let us suppose, is stage-English. The only neurological differences between a child on one ship and a corresponding child on the other ship will be due to the trivial phonetic differences; otherwise they are neurologically identical. Copyrighted material Experience, Theory, and Language -t> 165 Indeed, it was not necessary that the children on the stageEnglish ship be taught a phonetically different language. If the phonetically different stage-language was somewhat longer, couldn't they be taught an abbreviation of it (recorded also in the captain's cabin), and mighm't this abbreviation be phonetically identical to English? So their
neurology would be idet~tiCtJI to that of English speakers, though of course there would remain the separate dictionaries in the different captain's cabins (and perhaps interested social scientists telemonitoring from afar). Whatever neurological configuration realizes one of the languages will also realize the other; any neurological model of English will also be a neurological model of stage-English. Given the formal relations between the two languages, how could it be otherwise? Perhaps we can reduce the psychology of a speaker of English to oeu.rology, identifying his believing there is a table before him (believing that semantic content) with a panicular neurological state N. But there can be a parallel identification of his believing this corresponding stage-English sentence with the very same neurological state N. We can parallel one reduction by another; no determinacy lurks in neurological reduction. But did the children on the second ship learn, as we supposed, stageEnglish? Though an attempt was made to teach them that language (perhaps their teachers even silently translated i.nto English whenever they spoke, or silently said "that was stage-English; in English it means . .. "), perhaps the children really learned English, for perhaps a stage-language is unlearnable as a first language. But we have
seen that the grounds for saying this cannot lie in neurology; wbat tben can they be? The view that the children on the second ship learn stage-English firs in very well with theories of the fixing of meaning by external facts. Here, the central external facts are the setting up of the experiment and the book in the captain's safe. Suppose these children are taken to an island, settle, have children, etc., and two hundred years later the misplaced captain's safe is discovered and opened. Would what the descendantS learn then be important? What would they have to do to change their language to a non-stage language? Suppose instead that the children on the two ships were settled on the same island. Would the differences between t.hem, as they fluently spoke together, be linguistically or semantically important? The children grow up, intermarry with those from the other ship, and raise their children who learn language from Copyrighted material 166 -& PHILOSOPHY AND METHODOLOGY their parents and from others around them in the normal way. These latter children have learned language equally from persons from each of the ships who themselves were unaware of any linguistic
differences among them. Do these children speak English or stage-English? Is there a fact of the matter about which they speak? The mingling together of two distinct external sources produces an example of "no fact of the matter" about which language they speak. They just ... speak. Someone who thought there a:mldn't be a case of "no fact of rhe matter" can, via the external fixing view, be brought to accept the possibilit)' as making sense. Now, how important are our differences from those people for whom there is no fact of the matter? (And are you sure that. we are different, that. our language wasn't such a commingling of different streams?) How important a fact is it about us that we are. in our group and not the one in which there is no determinate propositional content? Should we stop our children fTom intermarrying with theirs, so that our descendants wiU speak determinately, so that there wiU be a fact of the matter about which they mean? Whatever differences there are seem trivial, hardly worth making central co a philosophical position. We have supposed, until now, that tbe fixing of meaning by external facts would fix determinate meaning. (Even on that supposition, we
were led tO conclude that the differences between determinate meaning and no determinate meaning were trivial.) But how will the intentions of people in the initial fixing situa tion bave gained determinacy? And for any factual relation R between y (data, things, use of a word, etc.) and the x which is "fixed" by standing in that relation R toy, there will be an x' and R' such thar x ' R' y, or a y' such thar x' R'y', and rhere will be no apparent way to resolve rhe indeterminanc)' between terms denoring R and rhose denoting R' (or berween those denoting y and chose denoting y'). The ropic of the rel.arionsh ip of language acquisition to meaning bas led us to discuss inrroducing a term X to denote what explains clara D. This mode of rerm introduction clear!)' does not attach the meaning of X to firings of nerve endings. (On the other hand, we have argued, it also does not fix X so as to narrow Quinean indeterminacies.) We cam10t get a cogent argument for experiential content being wbat fixes the unit of meaning from considerations about language learning. The example of introducing terms via the explanation-relation which shows this also provides a special explanation of why Duhem's thesis
holds for some sentences. Introduced via the notion of 'explanation,' the term X Copyrighted material Experience, Theory, and language ~ 167 will live for us upon the notion of 'best explanation,' which is a nonlocal notion, looking at large portions of theory. Stiml4114s Meaning Quine's own positive account of the relarionsnip of meaning and experiential content centers on the notion of stimulus meaning; "stimulus meaning ... may be properly looked upon still as the objective reality that the linguist has to probe" (Word and Object, p. 39). The stimulus meaning is the ordered pair of affirmative and negative stimulus me.anings, and these meanings "of a sentence (for a given speaker at a given time) are mutually exclusive" (p. 33). A stimulation o belongs to the affirmative stHnulus meaning of a sen renee S for a given speaker if and only if there is a stimulation o' sucb that if the speaker were given <1', then were asked S, then were given o and then were asked S again, he would dissent the first time and assent the second. A stimulation <1 belongs tO the negative stimulus meaning of a sentenceS for a given speaker if and only if there is a stimulation o' such that if the speaker were given o', then were asked S, then were given <1,
and then were asked S again, he would assent the first rime and dissent the second. (Word and Object, p. 32) The immediate difficulty is that since each of these definitions uses an existential quantifier (" there is a stimulation <1"'), it is left open that for the same <7 and S there are two different stimulations <1 ' and <1" which serve to place u both in the affirmative and in the negative stimulus meaning of S. To have these be disjoint., we would need the truth of (3u')(if the speaker were given <1' then were asked S, then were given <7 and then were asked S again, he would dissent the first time and assent the second) :::>- (3u")(if the speaker were given u" then were asked S, then were given <1 and then were asked S again, he would assent the first time and dissent the second), and also of the statement which is gotten from this by interchanging "assent" and " dissent." \Y/e have been given no argument for these very strong statements, and they appear to be fa lse. First, an intuitive c.ase. Let <1' be the auditory stimulation "Please say 'no' to the first question and 'yes' to the second." If this is an admissible stimulation, then (<1) (S) (<1 belongs to the affirmative stimulus meaning of S); and we can proceed similarly with
negative stimulus meaning. Excluding such stimulations7 would make it difficult Copyrighted material 168 -& PHILOSOPHY AND METHODOLOGY to ascertain (or define) the stimulus meanings of sentences like "He has just spoken well" or "He has just spoken loudly." . For any CT which would appear to fitS, let CT ' be a stimulation which fits S and which leads the person to believe that though CT occurs, notS; e.g., the evolving ocular radiation which occurs when he sees people construct and place a brown papier-mache rabbit alongside a white rabbit. CT is the ocular radiation of seeing a brown rabbit with no white rabbit around. He assents to "gavagai" the first time and dissents the second (because he thinks he's still looking at the papier-mache rabbit), so that, according to Quine's criterion, CT is in the negative stimulus meaning of "gavagai." And so on. Intuitively, what one will assent to on the basis of stimulation depends upon one's other beliefs, and even though Quine means to include this by speaking of stimulus meaning for a particular person at a particular time, different preliminary stimulations cr' can change the person's beliefs in different ways. Can the tasks Quine sees for the (as we have seen, defective) notion of
stimulus meaning be accomplished without its use? Quine adduces two areas of use: for the philosophy of science, to demarcate the intersubjective tribunal of evidence for science (see "Epistemology Narural.ized," pp. 86-89) and for linguistics, within the theory of a child's (or scientist's) language learning and within the theory of translation. (We shall consider linguistics in the next section, after taking up here the use in the philosophy of science.) The widespread agreement among scientists, Quine holds, stems from there being sentences (the observation sentences) "on which all members of the community will agree under uniform stimulation" ("Epistemology Naturalized," p. 87). This notion gets Quine into problems about intcrsubjectively uniform stimulation, the homology of nerve endings, etc. Why not focus instead upon those sentences for which each person can arrange (or have arranged) circumstances (or stimulations, if you wish) to prompt him to assent? There is no need that the circumstances or the stimulations be intersubjectively the same. There is agreement enough for science to proceed if everyone can be brought to assent; there is no need for them to follow exactly the same path to this agreement. The explanation of why each assents in some arrangeable situation, for Quine, will speak of each individual's history
of conditioning of language to situations. To be sure, there must be enough perceive.d similarity in the circumstances so that they all think they're agreeing (rather than merely using homonyms) and hence supporting each other. Such would be the case if each sees the circumstance Copyrighted material Experience, Theory, and language ~ 169 in which another assents as one in which, with some slight modifications, he also would be led to assent to the sentence. Each need only see the situations of the others as (modifiable into) ones which present a discriminative stimulus for himself to utter S. It is fortunate that the concept of stimulus meaning is not nece.ssary for the philosophy of science, for, in addition to the difficulties with Quine's definition, there is further trouble. What a linguist treats as assent and dissent {and a query also?) is underdetermined, not merely inductively but "on a par with the analytical hypotheses of translation that he adopts at later stages of his enterprise" ("Replies [to Critics]," in Words and Objections, edited by Donald Davidson and Jaakko Hintikka, p. 312; does this indeterminacy about assent and dissent also begin at each of our homes?). Even if the "on a par" puts it too strongly, it would seem there is strictly speaking no fact of the matter about
whether something is assent {or dissent) or something else, and hence, strictly speaking, no fact of the matter even about stimulus meanings. 8 II. Language and Theory Why Linguists? If "stimulus meaning by whatever name may be properly looked upon still as the objective reality that the linguist has to probe. "9 and stimulus meaning crumbles, why do we need linguists? What is it that linguists are trying to do, and what is their subject matter? Let us consider some possible answers, and criticisms of these answers of a sort that would appear to be congen.ial to Quine. Later we shall face the question of whether similar criticisms don't apply to Quine's own answer. {a) The job of the linguist is to produce a theory that explains the linguistic facts about the sentences of the language: grammatical facts, and semantic facts, 10 e.g., that some phrases and sentences are synonymous, that some words are semantically similar to each other in a respect in which other words aren't, that some words are antonyms (have incompatible meanings), that some expressions are meaningful and others are semantically anomalous, that words or sentences have multiple senses, that some pbrases have senses that contain superfluous information, that some sentences are true by virtue of the fact that the meaning of the subject contains the property expressed by the predicate, that Copyrighted material 170 _. PHILOSOPHY AND METHODOLOGY
some sentences are false just by virtue of the fact that the meaning of the subject contains information incompatible with what is attribute.d to it in the predicate, that some sentences are neither true nor false on the basis of meaning alone, that some pairs of sentences are inconsistent, that a sentence sometimes follows necessarily from another one by virtue of a certain semantic relation between them, that some interrogatives presuppose the truth of some declaratives, that some sentences are possible answers to a question whereas others are not, that some questions answer themselves. The 'facts' in this list are heavily theoretical; they are not what, in the first place, a linguist has to explain, though in explaining what he must he might be led (or forced) to a theory which countenances such facts, and so have to explain them also. But what is it that the linguist must explain in the 6rst place, what sets him his original task? (b) The job of d1e linguist is to explain how we communicate to each other via language, how one person transmits his inner thoughts to another by encoding them in t.he form of external observable acoustic events which the other decodes thereby obtaining for himself his own inner represemarion of the speaker's thoughrs. 11 Once again, this can hardly be the data which the linguist starts out to explain, tnough be might be led to such a theoretical picture in the course of explaining, e.g., people's talking to each other.
(c) The job of the linguisr is to explain linguistic performance, that is, how someone utilizes nis linguistic competence to communicate with his feUow speakers in actual speech situations, that is, how someone utilizes what an ideal speaker knows about the grammatical structure of his language, stated as the system of rules that formally represents the ideal linguistic structure of natural speech, abstracting from memory limitations, noise, what situation the person is (inrerested) in, etc., ro communicate with his fellow speakers in actual speech situations. Again we have a heavily theoretical description of the linguist's task. Linguistic performance is what is to be explained. The hypothesis is that a theory explain.ing this performance will incorporate and attribute to the speaker-hearer linguistic competence in the form of (tacit knowledge oO an interoalired system of rules which describe (or perhaps constitute) the language. "[n general, it seems that the study of performance Copyrighted material Experience, Theory, and Language ~ 171 models incorporating generative grammars may be a fruitful study; furthennore, it is difficult to imagine any other basis on which a t.heory of performance might develop" (Noam Chomsky, Aspects of the Theory of Syntax, p. 15). Can it turn out that the power of this internalized set of rules is beyond the power of the organism who internalizes it? "A model
of the speaker's knowledge of his Language may generate a set so complex that no finite device [models which represent actual performance, Chomsky holds, must necessarily be strictly finite] could identify or produce all of its members. In other words, we cannot conclude, on the basis of the fact that the rules of the grammar represented in the brain are finite, that the set of grammatical structures generated must be of the special type that can be handled by a strictly finite device." 12 Thus, it is mistaken to think (as is often done) that the results of formal linguistics are supposed to apply directly to psychology, by showing how powerful a psychological model of us must be, on the assumption that we must be at least as powerful as the language we speak. We should be wary of concluding that our language (our language) is more powerful than we are, solely because such a more powerful language is most elegantly described. ("In genera l, the assumption that languages are infinite is made to simplify the description of these languages ~; Noam Chomsky, Syntactic Stmctures, p. 23.) Since we have no need to attribute to the organism itself knowledge of the description of its linguistic output (and so need no1 worry about its handling such complexities), we may choose the less elegant but psychologically more realistic
description, striving for theoretical elegance in our psychological theory of what produces its speakings (and other behavior).13 In any case, it is puzzling to encounter the confidence that a psychological theory of our speaking and acting on heard speech will incorporate the more powerful abstract structures (admined to be beyond our power) that linguists present. (d) The job of the linguist is to (help) explain linguistic performance, whkh is "the actual use of language in concrete situations. ~ 14 Do we now, ar last, have a sufficiently theoretically low-key statement of the initial task of the linguist? What could be more down to earth than trying to explain the actual use of language in concrete situations? The use of what? If on this view language is the internalized rules, physically realized, or (to avoid having the language transformed when its speakers die) an abmact entity of some sort,l5 then we are back to case (c) above. We do no1 want to interpret 'the actual use of language in concrete situations' as committing us to '(3x) (3y) (3z) (x is a person, Copyrighted material 172 ~ PHILOSOPHY AND METHODOLOGY y is a language, and z is a situation, and x used y in z)'. In particular,
we don't want to be committed to quantifying over entities which are languages in explaining the task of the linguist, in explaining why there's a job for him to do. Instead, we want the task to be something like helping to explain those events in the world which are people's ralkings and their behavior before and after producing or hearing talkings. I am indebted to conversations with Burton Dreben for suggesting that Quine would criticize these specifications of the linguist's task which are offered by Katz and Chomsky in the above ways, as being heavily laden with philosophical assumptions and theories, and would favor instead descriptions of the task and data of linguistics which were less theoretically committed. Quine on the Linguist's Task However, cannot this line of criticism be turned against Quine's own description? Even if there can be no wholly neutral description of the data, isn't Quine's description also more laden with theory and assumptions than need be, especially as assessed from a more minimalist standpoint Quine should find congenial, that of Skinnerian psychology? Quine offers the following: (e) The job of the grammarian is to demarcate the class of all the strings of phonemes that get uttered or could get uttered
within the community as normal speech without a bizarrene.ss reaction.16 Why does the grammarian try to demarcate the class of all strings of phonemes that could get uttercd? 17 Perhaps the picture is as follows: the grammarian wants to (help) explain what does get uttered. The metahypothesis is that the explanation of a particular utterance will have two parts or components: first, a component which specifies everything that could be uttered (by anyone, anytime, under any conditions), and secondly a selective component which operates on the specification of the set of all possible sentences (on the first component) to select the one which actually gets uttered. This meta-hypothesis that the explanation of the actual utterance of a particular sentence will refer to the set of all possible utterances is a very strong hypothesis. Will the explanation of a particular home run by Hank Aaron (that particular swing, trajectory, landing place, and number of RBI's) involve a specification of all possible home runs by him, and some further bit of theory which explains Copyrighted material Experience, Theory, and Language ~ 173 why he hit a (that) particular one of those? Nor is there any reason to think that the explanation of any particular home run will refer to the set of all actual home runs he hits during his baseball career. The
explanation of a particular knockout by Joe Louis does not incorporat.e reference to the set of all knockouts he could have made or did make, or to the ser of all punches he could have rhrown or did throw. It is certainly not generally true that the explanation of a particular activity will refer ro and utilize a specification of all possible (or actual) activities of that sort. An explanation of a particular activity may, however, yield (as opposed to utilize) a specification of all actual or possible activities of that sort. From the story of how a particular one was (or all the actual ones were) produced we may be able to see which ones could be produced; a specification of that set might drop out of the theory. But the explanarory task would nor be to produce that specification, and the specifications need play no explanatory role in the explanations one does give. 18 Thinking the primary task of rhe linguist is to demarcate and study the ser of sentences of the language, the ser of sentences rhar could be uttered, fits in nicely with rhe un-Quinean conception of language as an existent entity to be studied, knowledge of the totality of which plays a role in the production of individual sentences. But there is also another
route ro thinking the linguist must specify the set of all sentences which could be produced, that on the face of it might be more congenial ro Quine. This route runs as follows. For sentence S a person does uner, it might be said, evidences a disposition ro utter it in some circumstances C and perhaps in other circumstances as well. This disposition will be an internal (neurological) condition of the organism, in virtue of which the person utters S in C. Had the person not been in circumstances C he still would have had the disposition ro utter S in C. There is something about him which makes him an S-utterer-in-C. The set of sentences a person is disposed ro utter under some circumstances or other will be the sentences he co11ld uner, and wiiJ correspond to facts about his internal neurological condition. The job of the grammarian, on rhis conception, is ro specify rhis ser of sentences, as an aid to specifying these occurrent neurological facts. However, this view encounters dif6culties.19 The set of all circumstances a person could be under is illspecified. Does it include someone's saying "repeat after me," in which case the ser of sentences will include all sound combinations the person can (be raughr ro) produce?20 Does ir include rhe application of
instruments of Copyrighted material 174 -t> PHILOSOPHY AND METHODOLOGY torture? The meeting of extra-galactic visitors and all they might tell us? Does it include thinking for years and coining new terminology to deal with theoretical problems, e.g., "category," "muon," "stimulus meaning"? A child of ten will not be disposed to utter "I am twenty years old" until approximately ten more years have passed. Shall we count it as disposed at ten to utter "I am twenry years old" under the circumstance of ten years passing? For at age ten, it seems there is something about the child which, when ten years pass, eventuates in irs saying "I am twenty years old." But then an infant would be disposed to utter aU the sentences of English which it can learn. For isn't irs internal state such that under certain circumstances it will utter those sentences of English? Also, whicb language you speak depends on how you were raised; did you as an infant have the disposition to utter all sentences of every language? Would the grammarian aid the student of your behavior by specifying the set of sentences of every actual, and possible, language?2l Consider a uniform homogeneous pinball about to be launched
on a stationary pinball machine without flippers. The path the ball will fol low will depend on the force appl ied and the surface of the machine, irs frictional forces, placement of obstacles, their elasticiry, irs placement of holes, etc. Of course, even given all this, the path also will depend upon the nature (mass, shape, elasticiry) of the ball. ln virtue of these facts about the ball, it will follow the path it does in tbose circumstances. It bas the disposition, we might say, to follow that path in those drcurnstances. Similarly it has another disposition to follow another path in other circumstances (on another machine, say); rbis disposition (as int. ernal structure and properties of the ball) though distinct from the first disposition is not discrete from it,22 for it is the same physical state as the first. Suppose now it were suggested tbat since the ball does behave as it does, and lawfully, and could behave differently under other circum stances and lawfully, the theorist of the ball bas tbe job of listing irs possible paths, the different ones it is disposed to follow under different circumstances. After all, these all correspond to facts about the ball, be ing (internal) states of it. Yet it does not follow that this is the
theorist's task. Listing of dispositions would be a central task for the theorist on.ly if it were thought that (usually or often) distinct dispositions are dis crete. Otherwise, the theorist should give a statement of the one or few states of the entity, and the laws governing its behavior. Copyrighted material Experience, Theory, and Language -& !75 Linguists a11d Psychologists An alternative conception of the linguist's task is that of aiding the psychologist in the construction of a theory of verbal behaviorP with no initial commitment to the theoretical centrality of language as an entity or even to sentences as entities. l.t is not obvious that to say "there is not language, there are nor sentences" is to be entwined in pragmaric contradiction. Quine has already rejected one traditional explanation of talking behavior and the changes in behavior it leads ro, viz., the explanation that uses the idea idea, the myth of the mental museum, etc. The language story is another theoretical story, also open to scrutiny. If it is an old story, so is the museum myth; if the theorists who treat language as a structural entity trace elaborate patterns among its parts, so do the theorists of relations among ideas or semantic markers. Wby then
does Quine not begin with the Skinner of Verbal Behavior, what leads him to a more traditionaUy oriented language-position? My aim here is not to endorse Skinner's view, bur to show how close it is to Quine's views, and so to press rhe question of whether it isn't really his view. This explanation of each individual's behavior would be an individual explanation; other persons and their actions would occupy the position of external conditions.24 This is compatible with everything that Quine says about language being a " social art." If we view the linguist as aiding in the explanation of an individual's behavior, treating other individuals merely as sources of input for the individual in question, we avoid problems about the homolog)' of nerve endings and worries about being "unable even r.o negotiate the A-B-Cs of behavioristic psycholog)'. ll What explanatory task is the "intersubjective equating of stimulations" needed for? (Defining some (surrogates for) notions of traditional semantics; but what explanatory function do these latter notions have?) Compare the situation in economics, where no notion of interpersonal comparisons of utility is needed for the explanatory tasks of positive
economics. Each individual's behavior is explained by his own utility function, joint behavior as the aggregate of individual behavior, perhaps coordinated, bur the explanation of this needs no interpersonal comparisons. Sometimes a person may act on a belief about interpersonal comparisons (e.g., he might make a threat only if he believes its realization will hurt another person more than it will hurt himself), but to explain his actions we only need a theory about how he makes such comparisons. The situation of psychology would seem to be the same, Copyrighted material 176 -& PHILOSOPHY AND METHODOLOGY with interpersonal comparisons of stimulations serving no useful funcnon. "But how can we make translations without interpersonal comparisons of stimulations or of something?" And Quine tells us that be is in favor of rranslation.26 But what are translations for? Will translations aid in the task of psychological explanation of another's behavior including verbal behavior? To the extent that we explain people's actions and speech by anriburing to them (in indirect discourse) beliefs and intentions, and we take what tbey say as indicating and sometimes stating their beliefs and intentions, then psychological explanation will involve translating. Our everyday psychological explanations do proceed by such arrributionsP But such explanations are not offered or
endorsed by Quine. Presumably he hopes they can be dispensed with. If they can, then insofar as translation rides on their back, it can be dispensed with also. The fundamental story of behavior and speech, for Quine, talks of conditioning, innate distances, and stimulation of nerve endings. Sentences attributing beliefs and intentions are rough and ready summaries of these more fundamental facts. A ttanslation of S tells us that S srands to stimulations of its believer or to discriminative stimuli for him roughly as that which translates it does to our stimulations or discriminative stimuli. A translation abbreviates fact, and is only roughly accurate. It serves the purposes of psychological explanation when its inaccuracies or distortions do not matter, when the relevant portions of our theory of behavior do not cut so .finely as to be perturbed by the roughness. Explaining by using translations is like navigating a ship on the geocenrric theory; the distortions due to placing ourselves at the center don't matter for our purposes, but the simplifications do. Within psychology, translations are convenient summaries; not only do they feed into a theory, the grain of the theory determines what is convenient and will not mislead. So a summary of facts about
discriminative sti.muli or stimulation of nerve endings, as presented in translations, need be only as fine, and accurate, as the theory connecting behavior to stimuli can make use of. Furthermore, we should expect that in using our everyday theory of belief, intention, and action, our psychological theory would, given its latitude, allow a great range to attributions of propositional attitudes. Psychology remains a theory of individuals' behavior. It need not provide a notion of same or similar stimulation so as to provide the foundation for linguistics or the theory of translation. Instead, translating Copyrighted material Experienc-e, Theory, and Language ~ 177 is the handmaiden of psychology, helping it conveniently to describe facts about discriminative stimuli and stimulations. Whatever rough and ready similarity translating brings wi ll help the psychologist describe how it is with that particular person ("it 's like . .. ~); no fact of stimulation-similarity is used as an explanation, there is nothing for it to explain. Other more practical purposes of translation, e.g., to enable us to have fluent interchange, social relations, economic relations with others, raise no new issues of principle that need concern us, especially since "going native" will he the hest route and need involve no
translating at all. The fact that we read literary and scientific works in rranslation and learn much from them raises no new problems of explanation special to translation; it presumably will be explained by judiciously combining our existing theories of learning from speech in your own language, and respondint: to writing in your own language, with our theories about the author's behavioL We may learn differently from djfferenr translations of rhe same text, and the question "which did the author mean?" will either he rejected or be transformed into the question of the psychological explanation of the author's behavior. Still, translations are made, read, and enjoyed. The theory of translators' behavior will be a part of psychology. Translators themselves will lean on learned aids, max.ims, traditional equatings, etc. to carry out their task to the satisfaction, instruction, and delight of their readers, but in all this there is no sub;ect matter. Why might one think that the concept of stimulus meanings aids the psychologist (even when it is limited to inrrapersonal use, thus avoiding problems of homology)? Perhaps one hypothesizes that a theory of
linguistic performance, a theory of speakings (and changes in behavior from hearings?), will be based upon stimulus meanings, in that: (a) Some speakings just are assentings after a which itself is after a" and dissenting; that is, just are the speakings in Quine's 'judicious experiment' ("Replies [tO Critics]," p. 308). (b) Some speakings are sayings of S after 0' which is in the affirmative stimulus meaning of S. Though it does nor logically follow that a person will sayS when 0'-stimulated if he wi ll assent if stimulated (0' ', S? and then) 0', S?, it may be assumed that a small bit of psychological theory will fill in here. However, if S is a standing sentence which the person believes, it will have a null affirmative stimulus meaning, as Quine defines this, for Copyrighted material 178 -<> PHILOSOPHY ANl> METHODOLOGY there will be no sequence of a', S? to which the person will dissent. If we drop the initial setting a', S?, altering Quine's notion to let a into the affirmative stimulus meaning of S if the person merely assents to S? after a, then every srimulation (which doesn't get him to give up believing S) will be in the affirmarive stimulus meaning of a standing sentence he accepts. In either case, as Quine is well aware, the notion of srimulus meaning would seem to have little use in explaining utterances
of a standing sentence, except perhaps in the situarion of being asked S?. Just as for the hypothesis that a theory of linguistic performance is to be based upon one of linguistic competence, we might interpret the concern with stimulus meanings as based upon the particular hypothesis that a theory of speaking is to be based upon stimulus meanings (cen trally utilize the notion of 'stimulus meaning'). (Note the parrial parallels to the mold of Chomskian theorizing.) The merits of this hypothesis are unclear; however, Quine does not present srimulus meanings within this framework of explaini11g speakings. Instead, I th.ink, he introduces stimulus meaning to see how much of traditional philosophers' talk of mea ning he can save. Quine is not, on this view, the revolurionary exciser of meaning be is reputed to be; rather he tries to conserve some of the tradirion with his 'strictly vegetarian imitations' (Word and Object, p. 67). Why bother? Isn't this a case where considerations of simplicity of theory outweigh conservatism (Roots of Referet~ce, p. 137)? Of Quine's tasks for the concept of stimulus meanings, we have discussed translation, and its use in tbe philosophy of science ro demarcate the observation sentences. We should mention its possible use in our theory of a child's (or scientist's) language learning. But if language
learning is learning mea11ings, we don't need a theory for this; if it is coming to use words so that others reinforce and don't penalize, we have Skinner's story of shaping and reinforcing in three-term contingencies. Is stimulus meaning "a device, as far as it goes, for explaining the fabric of interlocking sentences, a sentence at a time (Word a11d Object, p. 35)? What is it we hope for from such reconstruction? Not meaning; "net empirical import of each of various single sentences then (Word and Object, pp. 34-35). Is this notion of import a notion useful to explanatory psychology? If not, what philosophical theory is it that Quine wishes to rescue it for? And what would that philosophical theory explain? Whatever it would, we presumably should find it writ large in the phiJosophical theory of science, the philosophical use Quine mentions, but we have seen it to be unnecessary there. Copyrighted material Experience, Theory, and Language ~ l79 Even if it is granted that we can explain agreement among scientist,s without introducing the concept of ~imulus meanings, can a radical "no language" view explain the histOrical success of science, its predictions and its technological achievements? To do this we seem to need the notions of reference and tmth, yet how can we engage in
such metalinguistic talk if language itself vanishes as an entity to be studied. It might appear that only a realist philosophy of science (which says the scientific theory has a model, namely, the world) can explain the success of science. 28 Quine, standing within his theory, can say what the realist says (e.g., "'electron' refers to electrons, which we have discovered to have properties ... "),even though reference is a notion within the indeterminacy of translation (Roots of Reference, pp. 81-82).29 The theorist I am imagining would first try to show how the scientists' speech was shaped,30 second (using the first) tO show what explained the scientists' and technologists' actions, and then, speaking himself, would say why those actions led in the world to the results they did. He would have no more difficulty in saring why those results ensued than why any event, not having to do with language, has its results. "Yes, he could say those things, but wben be does, does he also think he's talking about the world, referring to the world?" He thinks his talk has been shaped in various ways in (by) the world. ("But when he says that, does he think he's talking about the world?" Loop back.) And he will offer to tell you how his, and your, use of "refers" and
"about" was shaped.31 My aim in this last section has been to draw out Quine's response to a hard-line Skinnerian view of language, that is, verbal behavior. There is reason to think Quine should embrace this view, yet he continually uses notions from traditional theories of language, e.g., "reference," which have no place or explanatory purpose within the Skinnerian picture. What scienti6c explanatory purpose does Quine see these notions as nece.ssary for? Or is there some nonexplanatory and more traditionally philosophical role they play? Is there a baby, or is it all bathwarer? Does the thesis of the underdetermination of theories undercut even Quine's notion of truth? If no theory is superior except as judged by canons whose coo.nection with truth is unclear, and aU theories would explain tbe data, then is there a fact of the matter about which one is true? Quine reassures us (Word and Object, pp. 24-25) that we can apply 'true' to a sentence which we see from within the theory which embeds that sentence, and which we take seriously as our own theory. Copyrighted material ISO ~ PHILOSOPHY AND METHODOLOGY
But why doesn't a full appreciation of the underdeterminarion of theories undermine the seriousness and earnestness with which we rake our own theories? A rheory of rhe truth of sentences of an object language, one might have thought, would have to describe and segment the world and show how the sentences of the object language connect with the world. Tarski saw tbat one could avoid independently describing and segmenting tbe world by using the object language's view of it. Translate rhe object language (perhaps homophonically) into the metalanguage, and speak about the world through the object language translated. But if underdetermination of theory raises the worry that there's no fact of the matter abour which of alternative theories is true, one will similarly worry about the factual status of their metalinguistic translations. Nor are things helped by additional worries about the determinacy of the translating! Let S and T be alternative theories underdetermined by, yet admissible for, the data, and let Sand T be particular sentences in them, respectively. To explain what it means to say S and T are true we trans late them into the metalanguage as s and r' respectively, and proceed a
Ia Tarski. But what makes s a correct translation of S rather than ofT? If there's no fact of the matter about this, then in what sense are we, by using s, affirming the truth of S as opposed to that of T? It does not help to say that the translation into the metalanguage takes place within a theory wbicb includes analytic.al hypotheses which translate its object language subcomponent. For simi larly, we would have rhe notion of unique translation, to be sure within a theory which includes fixed analytical hypotheses. Hence, this route would give us a notion of truth which is no more secure than the notions of 'unique translation' or ' unique mea ning'! ln wbat way does Quine see "truth" as better off than these other notions he has done so much to undermine? Is the problem solved "by acquiescing in our mother tongue and taking its words at face value" (Ontological Relativity, pg. 49)? What is it to take tbe words at face value? To translate them homophonically? And why don't we have tO iterate that translation, to ensure that we're raking it at face value? To use tbe same words i.n the translation in the metametalanguage for the truth definition of the metalanguage? (Can't all th.is be done without understanding the words at all?) To take s at face value is not to translate it in a certain way, for what it is
translated i.nto will raise tbe same problem. Quine offers one picture of understanding: one understands a language if one can translate it into some language. (Those monolinguals among us, less linguistic.ally agile than Quine, may Copyrighted material Experience, Theory, and Language -& 181 find themselves reduced to only the homophonic translation.) Clearly one must translate it into a language one understands- we can homo phonically translate languages we don't understand-and this notion of understanding will have to come down to something which is not translation; presumably, it will be fluent use and application, uncorrected by other fluent users and appliers. But all such use rests will be within the indeterminacy of translation, and will not distinguish s from 1, and so cannot fix s for the purposes of a truth definition.32 Copyrighted material 8 SIMPLICITY AS FALL-OUT Lf an indefinite number of hypotheses fit our data, we ncec;l some way to select which one to accept. Merely 6ning the data is not sufficient; the additional criterion philosophers put fortb most prominently is simplicity: we should believe the simplest of the hypotheses (we can formulate)
that fit (and would explain) our data. Why believe the simplest? Convenience in manipulating, remembering, transferring, and teaching the theory or hypothesis would lead us, perhaps, to utilize the simplest, but why believe it? Why think that, of the orherwise equally satisfactory explanatory theories, the simplest is most likely to be true? If simplicity is "relative to the texture of a conceptual scheme, "1 to the kind of "graph paper" on which we plot the world, then to show this connection between simpliciry and truth would involve the unpromising task of establishing our conceptual scheme or coordinate system as privi leged. The question about simplicity is not merely a skeptical conumdrum. It is difficult to think of any reasonable explanation for why including such a simpliciry maxim should help (and should have helped) make the institution of science successful or more likely to arrive ar tbe trurh. To explain how a simplicity maxim contributes to the success of science by saying that the world is simple says little, if i.ndeed it says anything Copyrighted material Simplicity as Fall-Out -& 183 other than that there is some simple true theory of the world. (Of all of it? And why think any particular simple theory, which
explains only some things, will be a component of the overall simplest theory?) So the maxim of simplicity and its connection with truth puzzles us. W. V. Quine and joseph Ullian2 find that natural selection "offers a causal connection between subjective simplicity and objective rruth . .. Innate subjective standards of simplicity tbat make people prefer some hypotheses to others will have survival value insofar as they favor successful prediction. Those who predict best are likeliest to survive and reproduce their kind ... and so their innate standards of simplicity are handed down" (p. 47). But maxims of simplicity direct us to choose the simplest of hypotheses in areas and at levels in physics, chemisrry, cosmology, molecular biology, none of which were believed or chosen by our ancestors. What explains why their subjective standards of simplicity, which have stood the evolutionary test for hypotheses about middlesized macro-phenomena, also connect with truths about cosmology or micro-phenomena? (Or is there no "comtection between subjective simplicity and objective rrurh" in these areas?) Perhaps the fact that nature is uniform, here interpreted as meaning that the same type of simplicity obtains throughout. If one could turn this into a precise statement about the world, the requisite sort of explanatory link between our standards of simplicity and truth would be forged. Quine and Ullian funher
note that "such standards will also change in the light of experience, becoming still beuer adapted to the growing body of science in the course of the individual's lifetime. (But these improvements do not get handed down genetically)" (p. 47). However, they do get handed down in the education of scientists. If scientists' standards of simplicity change to fit the theories they develop and accept, then these theories will seem to them and to later generations, who judge by these changed standards, simple. In "On Simple Theories of a Complex World," Quine artfully sidestepped the issue of simplicity's connection with truth by connecting it not with truth but with confirmation. A preference for simplicity is, so to speak, an artifact of our procedures of confirmation, so that the simpler of two theories is more likely to be confirmed and to hold up under the way we investigate theories. The question Quine raises, what the anifacts of our procedures are, is of extreme importance. However, to raise the question of whether something is an artifact of our procedures of confirmation is to raise the question of whether, in that respect, those Copyrighted material 184 -<> P H I LOSOPHY AND METHO D O LOGY procedures should determine our beliefs. If people who learn
science (strangely) got headaches when they considered a hypothesis containing three words beginning with the letter d, then no such hypothesis could get confirmed. That would he an artifact of the manscientificprocedures combination, but surely no reason to disbelieve such a hypothesis. Thus, questions of the justification, and not merely the causes, of our beliefs would be raised. Put somewhat differently, noricing the artifact and knowing the arrifacrual causal story might lead us to change our (artifacrual) beliefs.' Simplicity is usua lly introduced as an additional factor to help decide among the different hypotheses that fit (or would explain) the data D1 in hand at time r1 Another way to decide would he to gather more data. But this seems not to solve the current problem, for however much new data Dz we gather, and however many old hypotheses Dz eliminates, there will stiU be an indefinite number of hypotheses that fit D1 + Dz. And how are we tO choose among them? The feeling that no progress has been made by collecting oew datato be sure some old hypotheses are refuted-should set us to wondering whether rhe situation is being misconceived. Instead of thinking of a temporal cross-section of hypotheses and data, consider the following
very simple model of an ongoing process. At time 11 there are data D1 and already formulated hypotheses H1 . Hm which fit D1 New data D2 are gathered to select among these hypotheses. If only one hypothesis survives, believe it (tentatively), and go on to gather new data to test it against other hypotheses you now think up which fit D1 + Dz. "But how can it matter th.at a particular hypothesis was actually thought of earlier and tested? If another hypothesis H23 which we have just now thought of and which also fi ts D1 + D2 had been thought of earlier, then it would fit all the data we now have. Surely which hypothesis the data support cannot depend upon which hypothesis we happened to be thinking about as we gathered the data." U HLJ had been thought of earlier it would fit D1 + Dz now, but we don't know that it would 6t all the data we would have now, for if Hz3 had been thought of earlier, we would have garhered different data. Our process of data gathering is selective. Given hypotheses H;: y = f 1(x) we gather data at values of x where the hypotheses differ, and at a sufficient number of points x so as to leave only one of the original hypotheses surviving. If HLJ had
been one of that original number, we would have gathered data at some additional points x. Copyrighted material Simplicity as fall-Out -t> 185 Granted that the counterfacrual that purports to make the sequential procedure look ridiculous i.s not known to be true, still the question arises of why we should (tentatively) believe the surviving hypothesis rather than another that fits those data just as well. The answer depends on the way the underdetermination of theory appears to make it difficult to (strongly) support a hypothe.sis. Hypothesis H fits data D. But an infinite number do. We expect many to, and tbis is one of those which do. Whatever the data we found, we could dream up many hypotheses afterwards to fit them, and so with D we dreamed up H; is this any reason to believe H? Suppose someone says he will show you that he is a very skilled archer. U an arrow is shot at random at a large target, the probabiliry of its bitting a particular point is very small (bas measure zero). And that's true for each point. "So,~ he says, ~ if I shoot and succeed in getting the arrow to a particular point, it will show I am very skilled." He shoots, the arrow lands, and he says, "See how skilled I am; the
probability that the arrow would have landed precisely there at random was minuscule." "Hold on," we say, "the arrow had to land somewhere, and it landed there. Wherever it landed would have been an unlikely place. We'll believe you're a skilled archer if you first specify the minuscule area your arrow is going to hit, and the, succeed in hitting it." Similarly the data are going to land somewhere. That they land in a particular place is a reason for thinking a specification of that place is significant only if that specification was offered beforehand. Suppose the archer shot at random at a wall, and then went to where the arrow had landed, drew a circle around it, and said "Bull'seye!" Suppose when the data land somewhere, we draw a curve through it and say "Bull's-eye." A bull's~ye is when the data nit the hypothesis, not the other way around. Since bull's-eyes give support, ro believe the best supported hypothesis is (some complications aside) to adopt the method of tenacity.> Now to connect simplicity with the sequential selective data gathering and testing procedure. We (tend to) think of the simpler hypotheses, relative to our conceptual scheme. So the simpler hypotheses get
into the &ay early. Suppose only one hypothesis Hs &om the initial batch of hypotheses we formulated survives our gathering of data Dt and, in accordance with the procedure, we tentatively believe it. Either: (1) The initial hypotheses6 were the simplest possible wbich would fit 0 1 ; no other possible hypothesis that fits 0 1 is simpler than any one of these. With all but one of the original hypotheses eliminated Copyrighted material 186 -<> PHILOSOPHY AND METHODOLOGY by Dz, Hs is rhe simplest possible uneliminated hypothesis that firs D1 Since whatever fits the large data set D, + Dz must also lit D1, Hs is the simplest possible hypothesis that firs D1 + Dz. So we already are (tentatively) believing the simplest hypothesis compatible with our data. Or (2) we did nor originally formulate the simplest hypotheses that lit D., and there are hypotheses simpler than Hs compatible with D1 + D2. We formulate funher hypotheses, and enter the next stage, gathering the (selective) data D3, and we (tentatively) believe the surviving hypothesis. Either it's now like case 1 above: our formulated hypotheses were the simplest that fit D1 + D2 (that is, their subset bounded by this stage's
survivor contained the simplest), and so we're now believing the simplest hypothesis compatible with all of our data, or else (loop) case 2 a gam. This sequential testing procedure fails ro get us believing the simple.st hypothesis (of those we can formulate) compatible with our data only if there is an infinite descending sequence of more and more simple hypotheses, and we start up on it and move slowly down. If we sequentially test hypotheses that we've thought of, tentatively believing the survivor, and if we do (tend to) think of simple hypotheses (judged relative to our conceptual scheme), then the result will soon be that we are believing the simplest hypothesis (of those we can formulate) which is compatible with all the data we have. (Even simpler hypotheses, of course, may have been eliminated earlier_) We need no rule or maxim that bids us to believe the simplest hypothesis compatible with our data. It's just a consequence of the operation of that sequential procedure that we will end up doing so. And the fact that simplicity seems relative to our own conceptual scheme, to our background concepts, coordinate system, type of conceptual graph paper, etc.-a fact that blocks attempts to connect simplicity with truth-/irs naturally into this stOry. The more simplicity fits however we
happen to tend to think, the more likely it is that we will think that way, and early. To guarantee we'll end up believing the simplest hypothesis compatible with the data, it would have to be assumed, not merely that we tend to think up simple hypotheses, but that by some time we have thought up all the simplest_ This assumption is too strong, but then again science provides no guarantee of convergence to the simplest. It is no defect for our theory of simplicity to explain no more than t.he facts. Simplicity is not called in as an additional criterion for hypothesis selection which is made necessary by underdetermination of theories. Rather, underdetermination of theories makes necessary, in order to Copyrighted material Simplicity as Fall-Out ~ 187 have hypotheses supporred, a sequential process of data gathering and testing, and out of that process drops simplicity (as judged by us).7 But what is the explanation of the fact that we do find hypotheses that fit the data at all, and that not all those we think up are always eliminated at the very next stage? Scientists choose to work on problems and in areas where they think they can get results, and leave intractable problems
aside. It would need explaining if they were never able to succeed! Our argument about how simplicity precipitates out of the sequential process would apply to any situation in which: (1) lnstances of some type of thing Tare generated. (2) The instances can be ordered along some dimension D. (3) Those T 's which are generated e.arly tend ro cluster around and eventually exhaust one end of the dimension; they are D-er than any T's not yet generated. (4) At each stage, T's that fail to satisfy some criterion C are eliminated. (5) The ordering of the T's along the dimension D does not depend upon their relation to the criterion C. Since these conditions do not focus especially upon simplicity, any process with a dimension D that satisfies 1 to 5 will eventuate in the survival of the 0-est.s The theory we have presented yields simplicity as fall-out from the process, but it gives simplicity, qua simplicity, no role as input, and so does not account for the way we utilize simplicity, seek it, and consciously favor it. For example., the sequential process may involve rejecting previous data on grounds of simplicity, if there is a relatively simple hypothesis that firs 0 2 and most of o., and only very complicated hypotheses fit Dz and all of D1. Some of 0 1 may be rejected in order
to accept the simpler hypothesis. Or suppose we came upon beings on another planet with a different simplicity ordering and, hence, with a different history of hypothesis testing. They had never tried out what (to us) is the simplest hypothesis about the operation of their environment. Would we be impressed by the history of one of their hypotheses? Before we tested theirs against the simplest one of ours that fit their data, which would we believe? If ours, even though it hadn't yet been tested by them or us, that also would show that there's something more to simplicity than merely the fall-out result of the sequential procedure. Our story about simpliciry as fall-out can be elaborated to give simplicity some role as input to the process as well. Once we notice that Copyrighted material 188 _,. PHILOSOPHY AND METHODOLOGY we are believing the simplest hypothe.sis that fits the data, we can come to pursue a policy of doing so. The fal l-out provides the basis of an induction. Here Quine and Ullian's natural-selection argument serves. We find ourselves having simple beliefs about ordinary middle-sized narural macro-objecrs and situations because those (who were contemporaries of our ancestors) to whom these truths seemed complicated left no descendants among our contemporaries. Also, since the process of
sequential testing and tentative acceptance leads us to end up believing the simplest hypothesis compatible with our data, a retrospective look will find that simplicity is successful, that the successful are simple. Even more so because, as Quine and Ullian note, our culturally transmitted standards of simplicity rend ro change so as to fit more neatly what we acrually have ended up accepting.9 Since the past exhibits a correlation between the simplicity and the success of a hypothesis, a modest induction- to be sure, a simple one, but that's how we tend to think- leads us to conclude that these do go together and, hence, to rely upon simplicity. There are various ways to imagine the induction; one is as an inference to the best explanation.10 We start with an observed connection between simplicity and acceptance by scientific procedures, and we explain this by positing connections between each of these and truth: we posit that acceptance of hypotheses by scientific procedures is correlated with their truth (there is independent reason to believe this, since scientific procedures eliminate false hypotheses that conflict with the data) and that the simplicity of hypotheses is connected with their truth (that the simpler of competing
hypotheses is more likely to be true). The.se last two connections, if sufficiently tight, would imply and explain the observed correlation. One, therefore, might be led, in order to explain the observed connection, to infer a real connection of simplicity with truth. This account is admittedly rough, but fortunately we require no precise and rigorous inference for a plausible account of how one might actually arrive at a trust in simplicity. lt might appea~:, however, that this account is vitiated by circularity. For the conclusion that simplicity and truth really are connected is inferred in order to explain the observed connection of simplicity with acceptance by scientific procedures. Yet there are other more complicated explanations of this observed connection. So in making the inference to the connection of simplicity with truth (and not some inference to a more complicated explanation) isn't a simplicity maxim already being used? A simple inference was made, but simplicity was not consciously pursued. Since our purpose is not to ;ustify simCopyrighted material Simplicity as Fall-Out -& 189 plicity but to explain how it might come to be consciously pursued as a goal, we legitimately may place the wisdom of pursuing simplicity as the conclusion of an inference which instances simplicity
without itself pursuing it. Starring only with simplicity as fall-out, we end up trusting in simplicity and using it in the process. Seeking simplicity affects the sequential procedure in two ways: first, we try to think up especially simple hypotheses and, second, we consciously use simplicity as a criterion in selecting among hypothe.ses. Thereby the sequential process even more effectively yields simplicity.11 This lovely picture, unforrunately, is marred by the fact that our fall out explanation of simplicity, if correct, undercuts the induction that leads to trusting in simplicity. If the fall-out tale is the best explanation of the observed correlation between success and simplicity, then the inductive extrapolation that leads to trust in simplicity is blocked; that is, it is unreasonable. But still, it may have occurred (and not unreasonably then, since no one then had the fall-out explanation), and that induction may be the actual explanation of our current trust in simplicity. Does and should accepting all this undermine trust in simplicity? It would be pleasant to bring things full circle by having the trust tenaciously hang on. After all, the hypothesis that there is a real connection between the simplicity and the success of a hypothesis entered the field before my explanation did; that hypothesis fit past data, and new data
were gathered which the hypothesis also fit, and so that hypothesis came to be accepted. In accordance with the method of tenacity, that hypothesis and the accompanying trust in simplicity will and should continue; that hypothesis should not be displaced merely because some new hypothesis has been thought up which also fits the data. But even if we granted this application of the method of tenacity, our confidence in simplicity might well be short-lived; for the next step, now that another hypothesis has entered the field, would be to use the hypothesis to generate different predictions, to test them selectively, to discover which survives, and to end up believing that one. However, I find myself already believing the explanation I have of fered. This might be because the two explanations, mine and the one positing a real connection between simplicity and truth, do not fit exactly the same currendy available data, because mine already explains more. But another view of the matter is more fun. My explanation involving fall-out and the induction based upon the fall-out is, I believe, simple, elegant, forceful, and lovely. More so, Copyrighted material 190 ~ PHILOSOPHY AND METHODOLOGY surely, than the reigning hypothesis of a real connection between simplicity and truth. The simplest and most elegant hypothesis is that there
is no real connection between simplicity (and elegance) and truth. Now if I accept this hypothesis as true (partly) because it is so simple and elegant, and the proponent of a real connection between simplicity (and elegance) on the one hand and truth on the other rejects this hypothesis as true because it denies any such connection, with whose petard is each of us hoist? Copyrighted material 9 INVISIBLE-HAND EXPLANATIONS In Nozick (1974) I described how, if people entered into mutual protection agreements and firms offered buyers protective services, a dominant protection agency would arise by legitimate steps, and this would constitute at least an ultra-minimal state. No one need have intended to produce a state. A pattern or institutional structure that apparently only could arise by conscious design instead can originate or be maintained through the interactions of agents having no such overall pattern in mind. Following Adam Smith, I termed such a process or explanation an invisible-hand process or explanation and offered a list of examples to make the phenomenon salient. These included evolutionary explanations of the traitS of organisms and populations, microeconomic explanations of equilibria, Carl Menger's explanation of how a medium
of exchange arises, and Thomas Schelling's model of residential segre gation. (Edna Ullman-Margalit, 1978, is a later attempt to define the concept.) Two types of processes seemed important: filtering processes wherein some filter eliminates all entities not fitting a cenain pattern, and equilibrium processes wherein each component pan adjusts to local conditions, changing the local environments of others close by, so the sum of the local adjustments realizes a pattern. The pattern produced by the adjustments of some enrities might itself constitute a filter another Copyrighted material 192 -& PHILOSOPHY AND METHODOLOGY faces. The opposite kind of explanation, wherein an apparently unintended, accidental, or unrelated set of events is shown to result from intentional design, I termed a hiddtm-hand explanation. The norion of invisible-band explanation is de.scriptive, not normative. Not every pattern that arises by an invisible-band process is desirable, and something that can arise by an invisible-hand process might better arise or be mainrained through conscious intervention. Economics typically explains patterns in rerms of the actions of rational agentS. However, a disaggregated theory of the agenr herself, wherein patterns that seem to indicate a central and unified
directing agent arc instead explained as the result of smaller, non-agent entities interacting, also might count as an invisible-hand explanation. 1 The definitional details of what counts as "invisible band" are less interesting than the particular theories. 2 Time-preference seems susceptible to evolutionary explanation (see Nozick, 1977; and Nozick, 1993, pp. 14-15). The future is uncertain, an organism may not survive to reap an anticipated reward, or the world might not present it. Innate time-preference may be evolution's way to instill in creatures incapable of explicit probabilistic calculations a mechanism having roughly the same effect, approximating what such calculations would have yielded with regard to rewards affecting inclusive 6tness; such rime-preference may have been selected for. Consider, then, beings with the cognitive apparatus to take explicit account of such uncertainties, who explicitly perform a probabilistic discounting of the future. If already installed in humans is an innate timepreferenceevolution's attempt to perform that probabilistic discounting for usand if what we explicitly discount in our probabilistic calculations is the (already discounted through time-preference) present value of the future reward, then what takes place will be a double-
disrotmting. Isn't that too much? Next, consider wealth maximization or the weaker assumption that people are seriously concerned with wealth. A widespread phenomenon across societies (though not within Western industrialized societies in the last 150 years) is that wealthy people tend to have more children. (Gary Becker, 1981, p. 102, cites supponing literature.) Suppose that, ceteris paribus, people with a strong desire for wealth tend to amass more; it is more likely that they will. If there had been a genetically based heritable psychological predisposition ro be (more) concerned with wealth- ! do not claim th.is as anything more than a possibilitythen that would have been selected for; the percentage with that heritaCopyrighted material Invisible-Hand Explanations -i> 193 ble desire would increase over time. This would provide an evolutionary explanation of, if not wealth max.imization, a widespread strong desire for wealth (se.e Nozick, 1993, pp. 126-127). Evolutionary explanations also can be brought to bear within philosophy to explain a priori knowledge of apparently necessary truths and to explain the intractabiliry of certain philosophical problems (see
Nozick, 1993, chap. 4). Traditional philosophical doctrine attributes to individuals a faculry of a priori knowledge, enabling them to know independently of experience that certain things must hold true, that they hold true in all possible worlds. It is implausible that evolutionary processes, keyed to rhe actual world, would instill any such completely general faculty within us. Yet certain propositions do seem self-evident, and it is difficult to think of ways they might be false. A certain proposition's seeming selfevidently true to us might have been selected for, if it does hold true (at least approximately) and if acting upon a belief in this does, in general, enhance fitness. That factual, contingent truth would come to seem more than just factual, through evolutionary selection via the "Baldwin effect~: those to whose "wiring" a connection or proposition seems closer to evident learn it faster and gain a selective advantage; they leave offspring distributed around their own speed of learning until, over generations, all find it self-evident. If, frequently enough, samples of a certain sort resembled their populations, then generalizing from samples to population, or to the next encountered member, would frequently yield truths, and those to whom such inferences seemed obvious and
self-evident would frequently arrive at those truths. Rationaliry itself might be an evolutionary adaptation. Evolution phylogenetically instills in us information about, and patterns of behavior suitable to, stable facts of our evolutionary past. Evolution utilizes and builds mechanisms around constant and stable environmental features (e.g., gravitational force is utilized in the working of some physiological processes, which were designed to utilize and function in tandem with steady gravity, not to duplicate separately wbat gravity already doe-s). Some obdurate philo.sophical problems (e.g., justifying induction,'or our belief in the existence of other minds or in an external world) might mark stable facts about humans' past environment that evolution bas built i.nto us as a.s.sumptions, marking facts to work in tandem with. All human beings heretofore have been born in environments surrounded by other people with minds similar to their own, in an independently Copyrighted material 194 ~ PHILOSOPHY AND METHODOLOGY existing "external world," one whose objectS continued on trajectories (or in place) even when unobserved, a world in which certain kinds of generalization or extrapolation from past experience led to further
truths. Those who failed to learn this quickly left fewer similarly uncomprehending descendants. Rationality's function was not to justify these assumptions but to utilize them. Hidden-hand explanations, the opposite of invisible-hand ones, rend toward ruling-class (or, more extremely, conspiracy) theories. What a ruling class aims at and produces or maintains is not given an invisiblehand explanation. However, the existence of a ruling class might itself be given such an explanation, if it did not arise as the result of some individual's or group's actions intending to bring dus about. Here is a sketch of how this might occur. Start with a society containing no ruling class, where the most powerful and wealthy individuals want their children and grandchildren to be equaUy or more advantaged and so place them i.n environments (schools, vacation places) that make more likely their children's marrying similarly advantaged people. Marriages forge alliances of mutual interest, making more likely the sharing of information and coordinated activities for mutual benefit. Allies and employees will tend to be recruited from similar schools and social nerworks, because rheir families are directly known, or because the sinl.ilar molding of their values, tastes, and modes of behavior makes rhem easier to work with, more predictable, more congenial, less
likely to create conflict. Directors of companies will be recruited from among similar persons already successful elsewhere; studies of boards of directors would show similar social backgrounds and much interlocking. Matters of mutual interest are discussed, including public matters; sometimes joint representation is made to government officials about matters of mutual concern. As issues become complex, or the polity becomes wide.spre.ad, organizations are started to think through the.se issues rogether and to interact with the government officials (or potential ones) who might significantly affect them. Thus might arise a pattern of wealthy and powedul individuals associating in social, business, and political life. How much success and ccordination in the determination of (which?) results is needed for this to constitute, technically, a ruling class? The coordinating organizations might be started, maintained and participated in without the aim of "serving the interests of the ruling class," so a ruling class might arise through an invisible-hand process, even if later it consciously maintajns itself. Copyrighted material Invisible-Hand Explanations ~ 195 Not just equilibria within markets, but the very existence of
markets in the West is largely the product of invisible-hand processes. People aimed to extend panicular markets in one or another direction, but "the market~ developed bit by bit, unintended. (Even after an overall conception did arise, the extension of markets rarely depended upon economists who bad mastered that general conception.) Now, however, there are self-conscious effons to establish markets and a market soci ety where none had been. If successful, the arising there of a market society will not have an invisible-hand explanation, but particular equi libria within the markets will. Will the new markets' achieving certain overall patterns have an invisible-hand explanation, when those mar kets were instituted in order to achieve just such patterns? (And what of our markets, if they continue to be maintained in pan because they are perceived to yield that pattern?) And what shall we say of new in stitutions, not imitating anything existing before, that are designed and instituted to achieve cenain patterns by so structuring incentives that people interacting will produce that pattern? The pattern is invisible to those within the institution but not to its designers.
Here is a suggestion of an institution, call it the help chain. With sig nificant public.ity and moral suasion, the government institutes a system of help-vouchers, distributi.ng Y hours of help vouchers to every family whose yearly income is below X. A person with such a voucher can re quest te.aching, advice, or help of an)' individual, and if that individual agrees and delivers it, he receives help-vouchers for that time expended. These vouchers he then can use himself, asking another person for help for himself or for any designated individual. Each person approached knows that if he agrees, he too will receive a voucher. Unwillingness to ask is reduced by knowing that the other will receive a useful voucher i.n return; willingness to agree may similarly be aided. Each year, there is a fresh infusion of help-vouchers, starring at the bottom of the income scale, and "trickling up" through voluJltary interactions. What new pat terns will result? The standard economist's invisible-hand explanation involves iodi vidual agents who choose rationally. (Notice that a theory of irrational behavior also might be specific enough to explain patterns arising from the interaction of individuals behaving in that predictably
irrational fashion.) However, the principle of rational decision need not be the principle of maximizing expected utility. In Nozick (1993), I propose a rule of maximizing decision-value, where this is a weighted sum of causally expected utility, evidentially expected utility, and symbolic Copyrighted material 196 -& PHILOSOPHY AND METHODOLOGY utility.3 This rule then is applied to Newcomb's problem and tO the prisoners' dilemma, with new results (see Nozick, 1993, chap. 2). (Newcomb's problem was first presented and discussed in Nozick, 1969; Richmond Campbell and Lanning Sowden, 1985, contains articles on this problem plus a bibliography.) New patterns can be explained in invisible-hand fashion as the result of the interaction of agents whose behavior conforms to this broader decision rule. One also might impose more stringent conditions on preference in addition to the usual structural conditions S (e.g., the von NeumannMorgenstern conditions). In Nozick (1993), I propose that these additional conditions include the following (which themselves then are also added to the set S): the person prefers satisfying the conditions S to not satisfying them; the person prefers (ceteris paribus) the means and pre conditions to satisfying the conditions S; the person prefers
having her first- and second-order preferences cohere; the person prefers that the preconditions for making preferential choice obtain, and that the capacities for making and effecting preferential choice not be interfered with. (There are more complicated additiona.l conditions.) When a person's preferences satisfy these (and similar) structural conditions, I say that her preferences are rationally roherent. A plausible view holds that the rationality of a belief depends upon the nature of the process that actually gave rise to (or maintains) it. Simplifying greatly, a belief is rarional if it arose by a process that reliably yields true beliefs.4 Can we demarcate a rational prefere11ce as one given rise to by a process that reliably produces --preferences? What is to fill in the blank? One can bootstrap by using the additional structure conditions. A particular preference is rational only if it actually was generated by a process that reliably yields rationally coherent preferences. This requires more than that the preference itself sarisfy the addirional structural conditions, for the processes that human beings actually can use reliably to yield coherent preferences form a restricted class; and it may be impossible to generate a particular preference by any process in this class, even
though it itself does not violate the structural conditions. Given interacting individuals with such stringently rational preferences, some further institutions or patterns might then be explainable. What are the limits of invisible-hand explanations? Many enduring patterns of behavior can be seen as maintained rigidly in the space left by the jigsaw puzzle of other people's actions, where the shape of each Copyrighted material Invisible-Hand Explanations -& t97 of those other pieces is similarly maintained by irs surrounding pieces. Are there lcinds of institutions or patterns that, in principle, cannot be given an invisible-band explanation? (Consider written constitutions.) Are there any social structures that could not have arisen by an invisibleband process, or be maintained by one?5 If so, is there an illuminating general description of what must evade invisible-hand explanation? Copyrighted material Copyrighted material ETHICS AND POLITICS Copyrighted material Copyrighted material 10 MORAL COMPLICATIONS AND MORAL STRUCTURES
I. Introduction In this essay I shall discuss some problems in representing one structure which may be exhibited by part of the moral views of some people. In particular, I shall be concerned with formulating a structure which would generate (or play a role in the generation of) a person's judgments about the moral impermissibility of specific actions. I do not claim that everyone's moral views exhibit the structure I shall discuss, nor do I claim that there is some one structure which everyone's moral views exhibit. Perhaps people's actual moral views differ in structure. Why I consider the particular structure I work toward worthy of discussion, given only the weak claim that some people's views may exhibit it (and perhaps only at a superficial level), will become clear as I proceed. The discussion strikes what I hope is not too uneasy a balance between a descriptive and a normative interest in the structure of a person's moral views. I would view it as an objection that no correct moral view could have the structure I discuss. And I would hope that a person holding the type of view I discuss would find disconcerting the discovery that his view does not satisfy some of the specific structural
conditions discussed later in the paper. Since people's views often do not satisfy Copyrighted material 202 ~ ETHICS AND POLITICS conditions which tbey should satisfy, the imposition of normative structural conditions upon a person's moral views lessens the likelihood that people wiU be found whose views exactly fit the structure described. I would hope, however, that some people's views satisfy the structure closely enough so that it will be useful in accounting for the judgments about the moral impermissibility of actions that they actually make or would make. Thus the essay is intended to provide the beginnings of a model of a moral judge which is useful both for descriptive purposes and for various normative inquiries, in much the same way as is contemporary utility theory and decision theory. Before proceeding, I wish ro emphasize the exploratory nature of this essay, which contains little more than the first steps toward one moral structure. Since a major purpose is to open an area for investigation, I have felt free to mention in passing some questions without attempt ing to answer them here. Indeed, the last section i.ncludes a listing of
further questions and problems which must be pursued if the kind of view sugge.ste.d here is to be made to work. I would have preferred to present here a finished and complete work, and hope that the presentation of a temporal section of an ongoing work will have virtues of its own. Though much has been written on panicular problems within ethics, one finds, in the literature of moral philosophy, very little detailed discussion of the structure of a person's actual moral views. This may be due tO its being assumed that a person's actual moral views exhibit one of two very simple structures (the first being a simple case of the second, which is worth mentioning separately). And this assumption may be made plausible by the desire, of many moral philosophers, to propose one or a small group of principles which would account for and unify a person's actual moral judgments (with a little fiddling here and there in the interests of consistency and theoretical simplicity). 1 If one or a small group of principle.s underlies all of one's moral judgments, then one will feel no need to discuss the structure of a person's moral views beyond stating the principles and saying something about the way in which the particular judgments are "derived" from the principles. For, it would seem, aU of the interesting questions
about a person's moral beliefs could be raised about the unifying principles. The two structures which it is often assumed or explicitly stated that (pan of) a person's moral views exhibit I shall call the maximization structure and the deductive structure. Copyrighted material Moral Complications and Moral Structures -& 203 A. The Maximization Stmcture According to this view all of one's judgments about the moral impermissibility of actions are accounted for by a principle which requires the maximiza6on of some quantity, subject perhaps to a quantitative restraint. The traditional utilitarian, or at any rate the traditional utilitarian of the textbooks, makes this claim. A proponent of such a view may discuss problems about knowing which act maximizes the quantity, or problems about what to do when one lacks adequate information to decide which act will maximize the quantity, but he will view all of the important questions about structure as obviously answered. I do not wish here to discuss utilitarianism; I shall just assume that it is an inadequate moral view. Even though it is inadequate, some pe.ople may hold it. If so, I am not trying to describe the basic structure of their moral view. However, if they are u6litarians because they believe th.is position accounts for their other moral views, it may be that the structure I discuss will capture their views at a less deep
level. The inadequacy of utilitarianism, of course, does not show that every view with a maximization structure is inadequate. Perhaps some such view is adequate, and perhaps such a structure underlie.s and accounts for the different structure I shall discuss, in an interesting way. I say "in an interesting way" for it may always be possible to produce a gimmicky real-valued function such that irs maximization mirrors one's moral views in a pan.icular area. For example, suppose there is a compli ated theory T, not having a maximization structure, which accounts for one's judgment of which actions are morally impermissible, and which are not morally impermissible. Deline a function f taking ac6ons as argument values, such that /(A)= 0 iff, according to T, A is morally impermissible; and / (A) = 1 iff, according to T, A is not morally impermissible. One might then say that (part of) one's moral views are represented by the requirement that one maximize (act so as to achieve ma.ximal values of) f. An interesting maximization structure r.equires the max.imi.zation of some function which was not gimmicked up especiaUy for the occasion (or similar occasions). That is, one that is inrere.sting for our purpose; there may be other reasons why one
might find interesting the representing of moral views via an artificially created maximization structure. Compare the situation with contemporary utility theory, which shows that if a person's preferences among probability mix.rures of alternatives satisfy certain natural-looking conditions then his preferences can be represented as being in accordance with the Copyrighted material 204 -<> ETHICS AND POLITICS maximization of the expected value of a real-valued function which is defined in terms of the conditions and his particular preferences (the utility function). One would not answer Aristotle's question of whether there is one thing for which (eventually) all actions are done by saying, on the basis of contemporary utility theory, "yes there is, namely, utility." To rule out, on theoretical grounds, such an answer, and to rule out similar maximization structures in ethics as interesting for our purposes, one would need a way of distinguishing the one from the many (gimmicked up to look like one). It is clear that the formulation of such a way faces problems similar or identical to the ones faced in artempting to rule out Goodmanlike predicates,2 and I shall not pursue this marter further here.
Many persons do not believe that there is an interesting maximization structure (and many believe that there isn't one) which accounts for their views about which actions are morally impermissible. But even if their moral views can (ultimately) be accounted for by an interesting maximization structure, it is still of great interest to see if one can state a structure (which would, by hypothesis, be reducible to an interesting maximization structure) which accountS for their moral views and which they can recognize as their own. B. The Deductive Structure According to this view, a particular judgment that an act is morally impermissible (1) Act A is morally impermissible would be accounted for as following from statements of the following form: (2) Act A has features F1. . . . , F. (3) Any act with features Ft, ... , Fn is morally impermissible. I shall consider only the simplest case where the features F; are empirical, factual, not explicitly moral features of actions. The moral premiss (3) is accounted for as following from the moral premiss (4) Any act with features T~, .. . , T., is morally impermissible conjoined with the factual premiss (5) Any act with features Ft, ... , Fn has features T., . .. , T.,. Copyrighted material Moral Complications and Moral Structures ~ 205 It is not claimed that the person consciously draws such
inferences, but it is claimed that, in some sense, such a structure accounts for his particular judgments, and that the person would be willing to say, or someone who understood his view would be willing to attribute to him the (perhaps implicit) belief, that an act with features Ft, . . . , F. is wrong because it has features T., . .. , T,..3 Principle (4) is accounted for in the same way as is (3), by reference to another moral principle, and factual premiss. Presumably, one eventually arrives at one or more moral principles P of the form of (3) and ( 4 ), which underlie all of the person's particular judgments, and all other of his moral principles, and which are such that there are no other principles P' of this form to which the person would assent or which the observer would attribute to the person as more basic(" Acts having the features P mentions are morally impermissible because they have the features mentioned in P'") from which (members of) P can be derived in the way that others are derived from P. The deductive structure cannot easily explain how it is that a person's moral judgment of a particular act (often) changes as he learns additional facts about the act; e.g., he no longer judges the act morally impermissible. In such cases the facts previously known to the person were not sufficient for the truth of a judgment of moral
impermissibility, and he did not have knowledge which instantiated the antecedent of an exceptionless moral principle. The expedient of maintaining the. deductive structure hypothesis by claiming that in all such cases the facts known to the person do not instantiate the features mentioned in an exceptionless moral principle but instead provide inductive evidence for their realization, while explaining how judgments would change with new information, is obviously implausible. A second difficulty with attributing the deductive structure as the superficial structure of some people's views is that these people are unwilling to state or assent to any or very many exceptionless moral principles.4 Many such persons, at some time in their lives, explicidy accepted sucb exceptionless principles, gradually rna.king tbem more and more complicated to fit more and more complicated cases. Then at some point they decided they couldn't state exceptionless principles which they were confident were correct and which would account for a wide range of their moral judgments. Perhaps reinforcing their lack of confidence in any exceptionless principles they were tempted to state, was the realization tbat if more than one such principle is stated, great care must be taken to ensure that in no possible case do they conflict. Copyrighted material
206 -& ETHICS AND POLITICS Such a history, I imagine, would be very common among lawyers, to whom the difficulties of de\~sing rules to adequately handle, in advance, all the bizarre., unexpected, arcane and complicated cases which actua lly arise, much less all possible cases, are a commonplace. The view that any laws a legislature will be able to devise will work contrary to their intention, or will work injustices, in some cases they hadn't foreseen arising or even contemplated, needs no exposition by me in a journal read by lawyers. Awareness of the difficulties in formulating rules to handle all the cases which will arise often leads to talk of the role of judicial discretion in a legal system, and to the incorporation within legal codes of statutes dea ling with the avoidance of evils, wh.icb do not attempt to specifically handle the possible c.ases.5 Thus one would expect lawyers to be as modest about the purported exceprionless character of any moral rules they can devise, or about most that come down to them in their moral tradition, as they are about the product of centuries of intensive legal effort to devise and refine rules to govern conduct.6 In saying that the persons would not confidently (or be willing to) pur
forth exceptionless principles which they believe would yield the correct judgment in each particular case, I do not mean to imply that they would not want themselves and others to publicly sta re such principles as exceptionless, and perhaps enforce them as such. They might want this, and do it because though they believe that the principles do not yield correct judgments in each particular case, they think it better that the principles always be followed than rhar each person consider and decide whether the situation he faces is an exception to the principle. They might think this latter alternative worse in the belief that far more often or far more importantly there will be cases where people, using their discretion, wrongly do not follow the principle than where, without using discretion, they wrongly follow it. Bur still the persons I am considering would nor believe that the principles they say are exceptiooless always yield the correct results. The theme of the inadequacy of most exceprionless principles that one can stare fits in nicely with recent writings on prima facie duties and rights, a subject whose imponance was first emphasized by \VI. D. Ross.7 My procedure will be to present a relatively simple structure in harmony
with these writings, and then to consider some objections to, criticisms of, indarities in and infelicities in this particular structure. A consideration of these leads to some suggestions about what a more adequate (though similar) structure would be like, which (or modifications of which), it is hoped, would be a component of a theory to account for Copyrighted material Moral Complication$ and Moral Structure$ ~ 207 the moral judgments of some of us about the mora.! impermissibility of actions. I shall be wholly occupied in this paper in working toward the more complicated structure, and shall not get to present it in detail here. II. The Simple Structure There are two open-ended lists of features of actions: W (for wrong making) and R (for right-making). Members of W are denoted by wh w2, ... ; subsets of W by Wh W2, . .. ; members of R are denoted by rt, . .. ; subsets of R by R., R2, ... If an action has some fearures on IV, and no features on R, it is morally impermissible. If an action has some fearures on R, and no features on IV, it is morally required (morally im permissible not to do it), or at least morally permissible.8 One key fact, which I mention in passing rhough much should be written on it, is that
neither IV nor R is empty. Funhermore, these are exclusive sets. Few morally interesting acts will have (of its features on either list) features on only one of the lists. Thus we need, in addition to a way of representing what features the person considers to be morally relevant (which the lists are supposed to do), a way of representing his judgments that some features on one of the lists outweigh or override some features on the other list. We shall represent these outweighings or overridings by inequalities between sets of features. Thus IV1 > R1 indicates that an action with no features on either list other than the members of IV1 U R1 (all of which it has) is morally impermissible.9 If R1 > IV1 then an action with no features on the list other than the members of W1 U R1 (all of which it has) is morally permissible (and perhaps morally required). Our initial statements about the actions which, of the features on either list, have only fearures on one of the lists, can now be written as W; > 1/> if>< R; where IV1 and R1 are any nonempty subsets of Wand R , respectively.10 To provide a way of representing drawing-the-line problems, and co provide a manageable way of representing a multitude of similar judg
ments, we shall allow some of the features on the lists to contain vari abies (in addition to that variable contained in every feature, which ranges mer actions, reference to which variable is always suppressed in this paper) ranging over positive integers, e.g., "leads to the death of n persons."11 Suppose this fearure is w1 which is a member of W, Copyrighted material 208 -& ETHICS AND POLITICS and consider a specific member of R, r1 One might have as an ourweighing w1 > r1; n > 2. This means that w, > '' for n > 2, and does nor entail that not-(w1 > r1) for 0 < n < 2. A feature may contain more than one variable (the variables being ordered within the feature), and an inequality may contain more than one feature containing variables. This suggests that in these cases we must consider, in our representation, ordered sets of features on each side of the inequality, with an ordered set of numerical constraints afterwards. We should note the possibility that the inequality may contain as a constraint an equation essentially involving more than one of the varia bles, e.g., nm ;:: 2t. I shall not pur sue here the details of how such complications are to be best represented formally.
III. Ordering of Features and Their Combinations What structural conditions on inequalities over features 12 may legit imarely be imposed? Do these yield an ordering and, if so, of what strength? We have a notation for outweighing or overriding. Do we need to introduce an equality sign for the !lotion of exactly balancing? The metaphor of scale balances would suggest that we do, but our explanations thus far do not seem to allow room for such a notion. For what wou.ld it mean to say that W1 exactly balances R1? It seems that an action having, of its properties on either list, all and only the members of W1 U R1 will be morally permissible or morally impermissible. If it is morally permissible, then R1 > W1; if it is morally impermissible, then W1 > R1 Where is the room for an equality sign?13 The further possibility one thinks of is that the system of principles is incomplete in that, for some particular act having some features on each list, it yields neither the judgment that the action is morally permissible, nor the judgment that it is morally impermissible. It would be misleading ro represent such a situation by an equality sign. A condition which is obviously desirable is (1) R; > \V1-+ not-(WJ > R;) (We needn't, of course, separately state its equivalent
conrrapositive.) The connectedness condition Copyrighted material Moral Complications and Moral Structures -& 209 is surely too strong a structural requirement to reasonably impose upon a person's fragmentary moral views. We shall take it up again later when we come to consider complete moral systems. In addition to (l ), obvious candidates are various transitivity conditions. Let X, Y, Z, S, T range over subsets (proper or improper) of R, and of W. X > Y is well formed only if either X c R and Y c W or X C Wand Y C R. Consider (3) (3Z) (X > l & not-(Y > Z))-+ (aHS) (Y > S-+ X > S) (b) (S) (S > X -+ S > Y) Intuitively, this seems reasonable. If X outweighs something that Y doesn't, then X has more weight than Y, and consequently outweighs everything that Y does, and Y is outweighed by everything that outweighs X.14 From (3), together with (1), follows (4) X > Y & Y > Z & Z > T-+ X > T Furthermore, one can define a notion of one set of features having more weight than another set from the same list: X Y = df. (3Z)(X > Z & not-(Y > Z)).1s Given (3), is irreflexive, asymmetrical, and transitive, and hence establishes a strict partial ordering of each list.16
But is (3) a legitimate condition to impose? Two reasons might suggest that it is not: 1. Different features may interact differently with features on the other list. 2. A feature may be strengthened or weakened, its weight increased or decreased, by something other than features of actions on the lists. The second of these reasons we shall consider later, along with other reservations which stem from applying the condition to notions other than "overriding" and "outweighing" as these are explained later. Consider (4). Can't there be a case where X> Y, Y > Z, Z > T, yet T > X? So as not to be misled by our inequality sign, let us rephrase the question. Can't there be actions A h A2, A3, A4, such that (a) A 1 has all the features in X U Y, and no other features on the list, and A1 is morally permissible. Copyrighted material 210 'i> ETHICS AND POLITICS (b) A2 has all the features in Y U Z, and no other features on the lists, and A1 is morally impermissible. (c) A2 has all the features in Z U T, and no other features on the lists, and A2 is morally permissible. (d) A4 has all the features in T U X, and no other features on the lists, and A4 is morally impermissible. The only cases (under 1. above, as opposed to 2.) which appear
to fit this, which I can think of, are cases in which, e.g., T and X interact so as ro produce another W -feature (not in T U X) which also applies to A4. But in such cases, it is false to say that the only feature on the lists which are had by A4 are in T u X Y Thus, I suggest that we tentatively accept condition (3), appropriately modifying it later when we come to discuss the difficulties under 2. Consider the funber condition (5) Wt c w2 c w ...,. (a) (3Rt) (Rt > Wt & not-(Rt > W2)) (b) (3R2) (W2 > R2 & not-(Wt > R2)). Or, to remove the here irrelevant possibility that there may happen to be no set of R-fearures "between" W1 and W2, consider Intuitively, (6) says that adding more W -features to an action cannot make it any better. (5), which given the previous conditions and defi nitions is equivalent to Wt C IV2 C IV _,. IV2 IV., says that the more W -features, the worse (ceteris paribus). (Perhaps we should restrict this ro acts which are not infinitely bad, if there be any which are. I ignore this complication here.) Once again, there is the possibility that some features which are in W2 though not in IVt. so interact with features in IVt, in R;, or with the facts of the situation (not represented by features
on the lists) as to 1>roduce a new R-feature which was not previously present. But once again in this case, it is false that rhe only morally relevant features the act bas are in W2 U R; (for some particular R; which is a candidate for satisfying the antecedent of the consequent of (6), but not irs consequent). So once again, unri.l the complications under 2. above are considered and incorporated, we shall tentatively accept and use (5) and (6). Copyrighted material Moral Complications and Moral Structures ~ 211 IV. Alternative Actions According to the simple model presented in section 0, whether or not an action is morally impermissible depends only upon the fearure.s of the action, and upon the inequality among two sets of features which it has. Let Wx be the set of all and only those features of act X which are on the W -list; let Rx be the set of all and only those features of act X which are on the R-list. According ro the simple model, act A is morally impermissible if and only if IV A > RA. In arriving at a judgment a bout the impermissibility of an act A, one need not, according to the simple model, consider the alternative acts available to the person, nor need one consider longer courses of action of which A may be a part.
The availability of alternative actions, and the embedding of an action in a longer course of action, produce complications which require modification of the simple model. In this secrion I shall consider only the first, leaving the second until section VI. Is it true that A is morally impermissible only if IV A > RA? Is it true that its W-features' outweighing its R-features is a necessary condition for an act's being morally impermissible? If RA > IVA, is this sufficient for A's being morally permissible? Consider the following two situations, obtained from one often mentioned in the literature in discussions of lying. (1) You are present as Q flees down a road from P, who you know will unjustifiably physically harm or kill him. P comes running along and asks which way Q went. If you say nothing, he wi ll continue along the road and catch Q. The only way to prevent this is to lie to P, telling him that Q wenr in a different direction. I assume that sufficient details can be filled in so that you will all agree that it is morally permissible to lie to P in this situation. The right-making feature of saving Q from great harm overrides the wrong-making feature of lying to P. (2) The same situation as in (I ), except that now there is some other way to save Q from the harm which does not involve lying to P, or any other wrong-making feature, e.g., if you stan to tell P that what he's doing is wrong, he'll stop and Listen and be convinced and won't continue on after Q. I shall assume that in this situation it would be morally impermissible to lie to P .18
I claim that the action of lying to P in (2) has the same features as it does in ( 1 ). I shall say more about this claim below. Now I wish to Copyrighted material 21.2 -& ETHICS AND POLITICS explore irs consequences. Since by hypothesis (in 1) the Rfeatures of this act outweigh the W-features, so do they in (2). Since the act in (2) is morally impermissible, this shows that the W -features of an act outweighing its R-features is not a necessary condition for that act's being morally impermissible. What principles can we formulate to handle such issues? The simplest principle which suggests itself, to handle the case just described, is (I) If W, 'I and (38) (8 is an alternative action open to the person & Wac w,. & R,. c Ra) then it is morally impermissible for the peJ:Son to do A (even if w,. < R,).'9 Roughly put, (I) comes to the claim that it is impermissible to do an action if one can achieve the same R-features at a cost of fewer \ features. (I) says that you must not do an action if an alternative action enables you to achieve the same good at less cost. A parallel principle would require that one not do an action, with non-null W -features, if an alternative action enables one to achieve a greater good at the same
cost. Thus (II) If w, "F & (38) (8 is an alternative open to the person & Wac WA & RA c Ra) then it would be morally impermissible for the person to do A (even if RA > W .. ). Note that (II) does not require that a person maximize Rfeatures, but requires only that he not pass up any if he is going to incur some Wcost. But still, (ll) seems to me to be too strong a condition to impose, and we shall not do so. But we want something rnore than (l). For though A may achieve more R-features than 8, perhaps the extra gain of A isn't worth its extra cost. Two principles suggest themselves, the first being a special case of the second: (III) If (38) (8 is available to the person and Wa = & w,. f & R8 c RA & WA > (RA - R8)) then it is impermissible to do A (even if RA > W.). (IV) If (38) (B is available to the person & R8 c R, & Wac w, & (IV,- W8 ) > (R,- R8 ) ) then it is impermissible to do A. First a word about how to interpret (lll) and (IV). If the moral world is very simple so that the total wrongness or rightness of a set of features is just the sum of the individual wrongness (rightness) and there is no Copyrighted material Moral Complkations and Moral Structures -& 213 interaction between the R and W features, then (X - Y) is just the settheoretic difference between X and Y; e.g., (IVA- W8 ) is just the set of features which belong to IVA and which do not belong to Ws.
And the inequality sign., as before, appears between two sets of features. If, however, as seems likely, things are not so simple, then (X Y) must be interpreted as some numerical me.asure of the difference in value between X and Y (where, e.g., value is measured on an interval scale; i.e., a scale unique up to a positive linear transformation). And the inequality sign. stands for the ordinary relation between. numbers. I shall say something about how such numerical measures might be obtained in section V. An important point for us to note here is that some measure of by how much some features outweigh others are better than, are worse than others seems required to account for a person's judgments of moral impermissibility, even apart from considerations about "risk," as the utility theorists speak of it. This point is further reinforced by the condition (VII), which follows below, which prevents assuming simple additive conditions without making them explicit, and prevents proceeding by speaking only of set-theoretical differences. Thus far we have considered only those cases where W 8 c W A But one wants principle.s to cover some cases where though A does
not have each of the W -features that 8 has, the other ones which 8 has are less bad than A's. The simplest case of this sort is one in which 8 bas at least all of the R-features that A does. This suggests (V) If W A =f & (38) (B is an alternative available to the person and (IVA- IVs) > 0 & RA c R8), then it is impermissible to do A.20 If (V) is acceptable, one wonders why it is nece.ssary to have RA c R8. Couldn't one have RA and R8 be of equal weight, or R8 have greater weight than RA , even though RA is not a subset (proper or improper) of Rs? This would give us (VI) If IVA -1 & (38) (8 is an alternative available to the person & (WA- IVs) > 0 & (Rs- RA) > 0), then it is impermissible for the person to do A. But what of the cases where there is no 8 with less weighty IV -features than A's which has at least as weighty R-features, though there is a 8 with less weighty W -features and R-features, and the extra gain in R-features from A rather than 8 is outweighed by the extra cost in IVfeatures from A rather than B. One might formulate the following principle to fit this: Copyrighted material 214 -& BTHICS AND POLITICS (Vll) If (38) (8 is an alternative available ro the person & WA W8 &
(IVA- Ws) > (RA- Rs)) then it is impermissible to do A. Consider the following intuitive argument for (VII). Suppose there were an act C which just rook up the Wand R slack between 8 and A. Thus, 8 & C bas just the same R and W features which A has. You've already decided to do 8. Should you do C in addition? If the answer is no, it is impermissible to do 8 & C rather than B. Since We(= IVAIV8 ) > Rc( = RA - R0), the answer is no. (I here suppose that the complication to be discussed in section VI does not apply ro acr C.) Thus it is impermissible to do 8 & C. Since 8 & C has exactly the same moral features as A, iris impermissible ro do A.21 A stronger principle than (VU) is (Vill) If (38) (8 is available to the person & IVA> IVs & (WAIVs) > ( RA - R8)), then it is impermissible for the person to do A. (Vill) differs from (Vll) only in the case where IV A = IV 8 (And in this case we would have R8 > RA.) Note that an intuitive argument for (VI11), similar ro the one for (VII), can be offered only if the Rlist is such that an action having some features on it and no features on the IV -list is morally required. For an action C which, in this case, took up the R and IV slack between A and B would have no IV -features, and would have some R-features. If such an action C is morally
required, we will have the intuitive argument for (VUI). (VUI) seems to me to be too strong a principle ro impose. We shall thus have to be careful later nor to construe the R-list so that all actions having some features on it, and no features on the IV -list, are morally required. If it is so construed, then principle (VIII) must be admitted as legitimate. I am more certain that one does not wish to require that a person maximize the difference between the R-weight and W -weight of an action. One does not wish ro require (IX) If (38) (8 is available to the person & (Rn- Wn) > (RAW11)), then it is impermissible for the person to do A. Nore that (TX) differs from (Vll) in that it does not have the clause that IVA > IV8 in its antecedent. Hence in a situation where R8 =52, IV8 = 50, R11 = 2, IVA= 1, (IX) has the consequence that it is impermissible to do A, whereas (V[I) does not have this consequence. The intuitive justification for (VII) was that A would be impermissible if its extra gain (over some other action) wasn't worth irs extra cost (over this other Copyrighted material Moral Complications and Moral Structures -& 215 action). This justification cannot be applied to (IX) since (IX) rules out actions which may have no extra cost over their alteroatives.22
Note further how an argument parallel to the one advanced for (VII) gets blocked. One might consider an action C which takes up the moral slack between A and B. Thus Rc = (R8 - RA); IVc = (\VB- IV,..) where W8 > w.. .. The person has already decided to do A. Is it permissible for him to do C? The aoswer is yes, for by hypothesis (of IX) Rc > We. Thus it is permissible for the person to do A & C, aod hence permissible for him to do B, if there is no other action which staods io the relation (VII) describes to B. However, from this one cannot conclude that he is required to do C after doing A; hence one cannot conclude that he is required to do B rather than A. (One could conclude this only by imposing a much roo strong interpretation of rhe lists and inequalities; viz., that if Rx > IVx then it is morally impermissible not to do X.) (VII) is the strongest principle of the sort we have been considering which seems to me appropriate. (IX), it seems to me, must clearly be rejected, though the rejection of (VIII) is more doubtful. But some complications about (VII) (and the other conditions accepted) must be mentioned. A. If there are an infinite number of actions available tO the
person, then for each one there may be some other action with the same Rfearures and less weighty IV -features. For example, suppose there are an infinite number of alternatives A h A2, ... , A1, . and the measurement scales yield the results that (i) (}) R,.., = RA1 and (i) IVA, = 1/i. There will be no action with least weighty IV -features. Hence, according to (VII), each of the acts At, A2, ... is impermissible. It is this sort of problem that leads writers in decision theory to speak of Eoptimizing. I shall assume that a similar line must be taken here, and shall ignore the derails of how (VII) is to be modified, and how the particular is to be chosen. B. Suppose that in our original situation (of lying to save someone's life when some alternative also saves his life and involves less weighty IV-features) the only other alternative which saves the person's life involves you in great personal expenditure of money, effort, energy, time, etc. In some situations where the alternative with less weighty IV -features involves great personal cost and inconvenience (though not additional IV -features due to this) one would nor require that the person not perform the act with the weightier IV -featuresP Where A is the
act with the IV -feature, and 8 is an alternative which stands in the relation described by (VII) to_ A, and 8 imposes personal (noo-IV) costs on Copyrighted material 216 ~ ETHICS AND POLITICS the performer of B, how are we to decide whether or not A is permissible? Presumably this decision depends upon (W A - W 8 ), and upon how great the personal costs of B are (how much greater than they are with A). But how exactly is a line to be drawn? One might consider an act C which is like 8 except that the personal costs to the performer of 8 are imposed by C upon some neutral third pany. C's imposing of these costs upon some third party presumably involves C in some IV -features (u1 addition to those of 8 ). The following principle-embodying the view that if you are required to throw the costs upon a neutral third pany rather than do A, then you are required to take the costs upon yourself rather than do A-seems reasonable: If C stands in the relation tO A described by (Vll), so that if C were available it would be impermissible to do A, then it is impermissible to do A (when 8 is available). That is, if the C corresponding to 8 is such that We WA & (R,.,- Rc) < (W,1 - We), then it is inlpermissible
to do A when 8 is availa ble and stands in the relation described by (VII) tO A, even though 8 uwolves personal costs to the performer. Delicate questions arise over whether substituting ~if and only if" for the initial "if" in this principle yields a legitimate principle; and if not, how the further cases are to be handled.24 I shall not pursue these questions here, but shall hereafter suppose that we have answered them and have available to us a modified (Vll) which incorporates the above principle and whatever complications are needed to handle the further cases. Let us denote by S the relation embodied in an adequately formulated (Vll) such that if some action available to the person Stands in the relation S to A then it is impermissible for the person to do A. Thus we refuse to say that an action A is morally permissible if its Rfeatures outweigh its W -features, bur rather we, ar this point, say that an action A is morally permissible if irs R-features outweigh its Wfeatures and there is no alremarive action available to the person which stands in the relation S to A. Let us consider one objection to this whole line of argument. It might
be said that in the cases which prompted condition (Vll), act A has added features which are easily overlooked. For example, one might say that the act of lying to the aggressor to save a person's life when one can save the life by persuading the aggre-ssor ro stop, has the added feature (when compared to the act in the situation where the only way Copyrighted material Moral Complications and Moral Structures ~ 217 to stop the aggressor is by lying) of being an unnecessary lie. And, the objection would continue, if this W -feature is included, then W A > RA. So of course the action is impermissible. Hence, the objection concludes, there was no need to reject the simple structure. If RA > w., this is sufficient for A's being permissible, and the whole line of argument in section IV thus far has been mistaken. We can rebut the particular "feature" which was suggested, by noting that "involves telling an unnecessary lie" is an explicitly moral notion, which explicitly e.valuates comparatively features of acts. For notice that if he can stop the murder by shooting the potential murderer or by lying to him, we can say "he unnecessarily shot him; he could have lied to him," but not "he unnecessarily lied to him; he could have shot
him." But, granting that the particular "feature" suggested won't do, isn't there some (other) way to capture "involves telling an unnecessary Lie," for a given situation, which doesn't involve explicit reference to moral notions? And if so, won't this have to be ruled out as a feature of act A? One might try, as such a feature of A (a) F = "is a.n alternative to another act B, available to the person, which has the same R-fearures as A and fewer W -features." This obviously won't do since it explicitly mentions the notions of Rfeature and W -feature. So one might try (b) F = "is an alternative to another act B, available to the person, which has features wh ... , w, and'~> ... , '"" where (though th.is is not said as part of F) r., ... , rn are exactly the R-features of A and w-1, , w, are a proper subset of theW-features of A. Note first that even if this worked it would handle only the cases covered by principle (I) in this section. But it doesn't work, because B may have weighty features in addition to w1, . . , w, which are on the IVlist, and if so, F won't be a W-feature of A. (lf one supposes F to be a W-feature of A in this case, what R-feature(s) of A correspo11ds to B's having weighty W-features ill addition to w., ... , w.,., and also overrides F?) Trying to handle this possibility by adding into F "and B has no other features which are on the W -list" won't give us a non-
explicitlymoral feature, and given the open-ended nature of the list one cannot build into F the conjunction which denies, for each other feature on the IV -list, that B has it. Thus, it is not at all clear how to begin t.o state the candidate for the non-explicitly-moral feature (on the IV -list) which A Copyrighted material 218 -+ ETHICS AND POLITICS has when there is an alternative 8 to A which stands in the relation to A described by condition Vll). Thus, we continue to maintain that it is not sufficient for A's being permissible that R~ > W A (and rhus we continue to reject the simple model), and we accept something like condition (VII) as marking off the exception to the sufficiency. If there is available one other action of a certain sort, then A is impermissible, even though R~ > IV.._. Is it also the case that IV~ > R.._ is insufficient for A's being impermissible, and do the same kind of objections to sufficiency hold as in the case of RA >IV~? Recall that we are tentatively accepting the view that (A) If R.._ > IV_. then A is permissible unless (38) (IV11 W8 & WA Ws > RA- Rs) . And if (38) (IVA Ws & WA - Ws > RA - Rs), then A is impermissible. If things were symmetrical we would have correspondingly on the
other side: (B) li w.. . > R,. then A is impermissible unless (38) (W8 w.._ & W8 - w.._ > R8 - R.._ );andif (38 ) (W8 w ... &w8 - w ... > R8 R ... ), then A is permissible. This says that if 011e action available is worse (in a certain way) than A, then A is permissible even though w.._ > R.._. But this is absurd; an action is not permissible just because we could have done something else which was worse. Whereas it is not absurd (and is indeed true) to say that an action with non-null IV -features was impermissible if we could have done something else which was, in a certain way, bener. Thus, the situation is not symmetrical, and we have what I shall call the First Asymmetry: The existence of one alternative action of a certain sort makes R~ > w_. insuJficient for the permissibility of A; whereas it is 11ot the case that the existence of one action of a certain sort makes W A > R.._ insufficient for the impermissibility of A. If w.._ > R11 is insufficient for the impermissibility of A, this is not because of something about one of the alternatives to A.2.1 But is IV.._ > RA sufficient for A's impermissibility or not? It seems clear that if it is not, this is not because one of the alternatives to A
is worse in a certain way, but can only be because all of the alternatives to A are worse in a certain way. And, indeed, it seems quite natural to Copyrighted material Moral Complications and Moral Structures -& 219 say that when all of the alternatives to A are worse than A itself, then A is permissible even though w. > R . Thus, one might say rhat (C) If w. > RA then A is impermissible unless (B) (If 8 is an alternative to A then W8 WA & Ws- WA > Rs- RA); and if (8 ) (if 8 is an alternative to A, then W 8 w_, & W 8 - W A > R8 RA), then A is permissible even though W A > RA. Against this one might plausibly argue that in each of the cases one rb.inks of wb.ich seem to fit IVA > RA yet all of the alternatives to A are worse than A itself, there are features which A itself has which capture the respects in which the alternatives are aU worse, e.g., "prevents the death of n persons," "saves the lives of these other persons." And since in these cases A has the additional R-feature of avoiding or preventing certain evils, it is i.n fact false that its W -features outweigh its R-fearures, so that we are not faced with a case where W A > RA, and so are not presented with a counterexa1nple to the claim that W A > RA is sufficient for the impermissibility of A.26
If this counter-argument is defensible, then we are faced with a Second Asymmetry. From our current vantage point, the First Asymmetry says that unlike RA > IVA , where one action of a certain sort makes it insufficient for A's being permissible, one action cannot make IVA> RA insufficient for the impermissibility of A, and if anything about the alternatives to A does make IVA> RA insufficient for A's being impermissible, it is something about all of the alternatives to A. The Second Asymmetry claim says that nothi>~g about the alternatives does make WA > RA insufficient for A's being impermissible.27 Since something about the alternatives can make A impermissible even though RA > W A the Second Asymmetry claim adds considerably to the significant eli vergence noticed by the First Asymmetry. (Of course, the simple structure of section U was perfectly symmetrical.) Is the Second Asymmetry claim rrue, or can A be permissible, even tl10ugh w. > R., because all of the alternarives to A are worse than A itself? Let us consider a specific example.Z8 A person is in rhe cab of a locomotive and approaching a rhree-way continuation of the route. If the train continues to go straight ahead, which it will do if nothing is done to itS controls, it will run down and kill 20 people. If it is
made to go on either the rightmost track or the leftmosr tuck, it will in each case run down and kill 40 people (for each track has 40 people tied on it). All this is known to the person in the cab. Copyrighted material 220 -& ETHICS AND POLITICS U the person allows the train to go straight ahead when there is no one on the side tracks, the act is wrong for it allows 20 people to be killed, and has no redeeming virtue. In the first described situation we think it permissible to allow the train to continue straight on because the W -feature of allowing 20 people to be killed is outweighed by the R-feature of avoiding the killing of 40 people.29 There is one notion of "avoids" which involves reference to changing things from how they would be in the normal and expected course of events, and it is this notion which lends plausibility and punch to the Second Asymmetry claim. The example that we have been considering leads us to think that there is another notion of "avoids," where, roughly, the fact that all actions but one lead to a certain consequence is sufficient to yield the conclusion that one avoids this consequence. But this additional "feature" of avoiding an undesirable consequence, in
this sense of "avoiding," saves the Second Asymmetry claim only via the course of trivializi11g it. For of course all of A's alternatives being worse than A won't make A permissible, where WA > RA, if the fact that all these alternatives are worse creates an additional and very weighty Rfeature of A (viz., avoiding-) so that when this is included, RA > WA. At this point we have a choice. We can say that the Second Asymmetry claim is true, though trivial. Or we can refuse to admit this sense of "avoid" as specifying a feature of an act for our purpose.30 And in this case, while accepting the First Asymmetry, we might try to specify a duality thesis, which would involve the substitution of "<" and ">," "impermissible" and "permissible," and of universal and existential quantifiers, and would have the consequence that A and C above are duals. This latter alternative is certainly well worth pursuing, but in the absence of special reasons for doing so, and since it would ma.ke section VI even more complicated, we tentatively choose the first alternative. V. Measurement of Moral Weight Those arguments in the previous section which depend upon conditions or principles which assume a method of measuring the difference in weights between sets of features on one of the lists may seem to the
reader to be, though interesting and intuitively correct, merely useless speculation in the absence of the description of some method for obtaining such measurement. Such a reaction seems to me to be unduly harsh, but I shall not dwell on this since I wish now to sketch a method for obtaining the appropriate measurements. Copyrighted material Moral Complications and Moral Structures ~ 221 The obvious suggestion is to consider probability mixtures of Wfeatures, and probability mixtures of R-features, and to attempt to parallel the Von Neumann-Morgenstern or similar axioms for utility measuremenr-11 so that we get the measurement we need. This suggestion might be reinforced by noting that an adequate theory of moral judgment will have to account for judgments of actions under situations of "moral risk" (where associated with an action is a probability distribution over sets of features on the lists, rather than just one set of features on the lists), nor 10 mention "uncerrainty." So, it might be asked, why nor introduce that apparatus at this point? It seems ro me that this approach is inappropriate for our problem here. For our problem is nor one of accounting for judgments in situations of "moral risk," and there seems ro be no intuitive reason for introducing apparatus based upon probability considerations to handle the problem we are now faced with. If we were to utilize the VN-M type of
measurement, arguments would have to be offere.d to show why the numerical values thereby obtained should function in the principles we have already listed. The more desirable course seems 10 be to find a method of measurement utilizing only considerations intrinsic ro the sort of situation i.n which our problem arose, or uti lizing only that apparatus which is sufficient to generate the problem and to show the need for a method of measurement. Such an alternative course has an additional theoretical advantage. For if one can establish the existence of numerical scales, assigning numbers to sets of features, ro be used in the principles for the situations discussed in section IV, and if one can use a VN-M -type procedure to establish numerical scales, assigning numbers to sets of features, based upon and to accou.ot for judgments of "moral risk" actions, then one can raise the question: What is the relation between these scales? I shall nor attempt here ro pursue or even specify the issues of interest which might arise. Is there some way, other than by a VN-M-type procedure, to establish the existence of numerical scales without utilizing information of a sort nor already provided by the kind of apparatus we are
discussing? If there is not, then there would be reason for believing that the son of apparatus we are discussing is (if not supplemented) seriously inadequate. I should here like to sketch (and I shaU here do no more than sketch) a way of obtaining the numerical values which function in the principles of the previous section. The procedure I suggest is very simple. Let me first state it in a way which will look circular. I hope that I have offered sufficiently forceful Copyrighted material 222 ~ ETHICS AND POLITICS intuitive argwnents for principle (VU) in the previous section so that you will ag~ee that we can assume that if there were numerical values, pri.nci pie (Vll) would be operating. We can use this assumption to obtain, for a specific person, specific inequalities between differences. Without en tering into the intricacies of the procedures to be used to discover this, I shall assume that we can determine that for the person some Rfeatures outweigh or override some W -features, some W -features are worse than others, and some Rfeatures are better than others. Suppose, for example, that we discover that (for the person) (a) Some R-features (cal.l them 'RA) outweigh some Wfeatures (call
them WA).32 (b) Some Rfeatures (Rs) outweigh some W-features (W8 ). (c) TheW-features IVA are worse than IVa. (d) The Rfeatures RA are better than R8 Thus we have (a') RA > W,, (b') Rs > Ws (c') IV,, w 8 (d'J RA Rs If now we can find an action A whose only morally relevant features are exactly those in RA and IV A, and an action 8 whose only morally relevant features are exactly those in R8 and IV8 , and the person judges that it is morally impermissible tO do A (if 8 is an alternative), then we can conclude that (for this person) (WA - IVs) > (R.,- Rs) If we ask him questions about other combinations of features and sit uations such as the previous one, we will get additional inequalities between differences. The important fact is that if we can get enough such inequalities then we will have sufficient information to establish a numerical scale of a given strength. Let me describe things somewhat differently. If the person is following principle (VII), which utilizes numerical values (if his judgments can be accounted for by this principle), then certain conditions will have to be
satisfied. For example, if these nwnerical values are to be measured on a scale which preserves relations among real nwnbers unique up to a positive linear transformation (an interval scale), then since it is a truth of arithmetic that Copyrighted material Moral Complications and Moral Structures -& 223 If X> Y > Z >zero, and W > T > U >zero, and X - Y > W-T, and Y - Z > T - U then X - Z > W - U, if the numbers assigned to sets of features satisfy the antecedent, they must satisfy the consequent. So in particular, it will be true that If WA > Ws > We> 0, and RA > Rs > Rc > 0, and (WA - Ws) > (RA - Rs), and (Ws - We)> (Rs- Re) then (IVA - We)> (RA - Re). Now this will be true only if the following is true. (Thus the following statement is a necessary condition for the weights to be represented on a certain kind of scale.) If (for a person) IVA Ws We and RA Rs Re and RA > IVA and R8 > IV8 andRe> IVe and the person judges that it is impermissible to do A in a situation in which B is available, and the person judges that it is impermissible to do B in a situation in which C is available, then the person judges that it is impermissible to do A in a situation in which C is available. Note that this does not utilize any apparatus or notions beyond those
we already had. Furthermore, it seems intuitively reasonable; that is, it is a condition one would wish to impose, and could well have imposed apart from all considerations about measurement.33 We thus have a necessary condition for the existence of a measure of the moral weight of a set of features on an interval scale. (I here implicitly assume, as throughout, that if there is a measure, principle Vll operates) which utilizes no very strong apparatus in its statement, and which furthermore seems intuitively reasonable and justifiable. One might hope to gather a large number of such necessary conditions, and prove that they are sufficient to establish the existence of an interval scale measuring moral weight. That is, for some locution already available to us from our previous apparatus, one introduces an n-place relation. The intuitively justifiable statements using this locution are written down as conditions on the n-place relation. Thus, corresponding to each intuitively justified statement (or rather, to a selection of these) is a condition on the n-place relation. If one has chosen wisely or luckily, one may then be able to prove, using these conditions on the n-place relation as ax.ioms: A Representation Theorem: showing that there ex.ists a realvalued function assigning numbers which is such that, for specified numerical relations, these relations hold among the numbers if and
only if some corresponding relation about the subject matter holds among the objects the numbers are assigned to; and Copyrighted material 224 ~ ETHICS AND POLITICS A Uniqueness Theorem: showing that any two real-valued functions shown to exist by the Representation Theorem stand in a certain relationship to each other. The more limited this relationship, the stronger a scale of measurement one has obtained.H To remove one simplification in this sketch: it will not be the case that each of the conditions on then-place relation which are jointly sufficient (it needn't be that each of them is necessary) to establish the existence of a measuring function will correspond roan intuitively justifiable normarive condition or one which specifies the notions involved. For, if similar previous results are any guide, one will require in addition various structural conditions.35 One hopes to find structural conditions, which when combined with the others will suffice for the task and which look as though (without too drastic an idealization) they are satisfied. The detailed rechn.ical task of specifying the axioms which seem reasonable in the moral context, and which yield the result, I must leave for another occasion. Here, I wish to point out that the prospects are very promising. One finds, in the literature, several axiom systems which either
explicitly are about (or can be interpreted to be about) a 4-place inequality relation between differences. Which set of axioms one uses determines the strength of the scale one obtains (ratio, interval, higherordered metric, ordered metric, etc.). Exactly which system of axioms should be adopted for our purposes here is a tricky question. Here I wish to confidently conjecture that some not very radical modifications of an already existing measurement system will capture intuitive moral conditions (plus some structural ones) and will suffice to yield some reasonably strong measurement of the moral weight of a set of morally relevant features. 36 Before leaving the subject of measurement I wish to remove in advance two reasons that one might have for believing that the project sketched cannot be carried our successfully. 1. The modified principle (Vll) of section IV was more complicated than I have made it out to be in this section. Won't these complications interfere with the procedure of measurement? They will not if one is careful to construct the scale via the person's judgments for situations where (a) there is no extra personal cost to the agent in doing 8 rather than A (b) there is nor an infinite set of available actions such that no action
has least wrong W -features. Copyrighted material Moral Complications and Moral Structures ~ 225 2. Does the project of measurement described above depend upon the daim (which the reader may consider not to have been established by my arguments in section IV) that "act A has as an alternative action available to the person to whom A is an alternative, an act B which --" is not usable as a feature of act A? It does not. For suppose that my claim in section IV is mistaken. Then, if there is such an alternative act B available to the person, A has an additional W -feature so that WA > RA; e.g., the W-features of l)'ing to the pursuer to save the other man's life would outweigh the R-features of this act (if there was a suitable nonlying alternative available). Still, given a reasonable independence assumption, the method sketched will yield inequalities of the form W~ - W8 > RA - R8, where W~ is the set containing all Wfeatures of A except the ones which refer to the availability of a suitable act B as an alternative to A. And these inequalities between differences can be used, as the others before, to establish a numerical measure over subsets of features not containing features which refer to the availability of an alternative acrionY Thus, the supposition that the claim in
section IV is mistaken does not block the proposed method of measurement. VI- Larger Courses of Action We have argued that for an act A such that W A ;f </>, RA > W A is not sufficient for A's being morally permissible. For there may be available an alternative act B which stands in the relation to A described by condition (Vll) of section IV, and in this case, A is not permissible. We have further argued that the claim that W A > R A is a sufficient condition for A's being impermissible cannot be similarly overthrown by considering situations in which all of the alternatives to A are worse than A. Should we conclude from this that W A > RA is a sufficient condition for A's being impermissible, or are there some other considerations which yield the result that A may be morally permissible even though WA > RA? The sort of situation which suggests itself is one in which though WA > RA, A is (a nece,ssary) part of some larger course of action B, and Ro > Wa. For some such situation, may it not be that A is morally permissible? It might be suggested that this problem can be avoided, because in such cases ~is a part of a larger course of action which is such that--" wi.ll be an R-feature of the A, and hence it won't be true that WA > RA. Even if this is so (I shall take up the question
later), one wants to know the appropriate way to fill in the blank. I shall approach this question by considering what conditions are appropriate supposing Copyrighted material 226 -<> .ETHICS AND POLITICS that such things are not features of acts. After doing this we shall then consider whether these conditions can be used to obtain an appropriate way of filling in the blank. Let me give two examples of cases for which it might be said that though WA > RA, A is permissible because it is a necessary part of some larger course of action B, where Rs > IV g. (1) A person P is unjustly being penned up by another person Q. You steal from some innocent third party R, one key to the door, making it possible for you to release P. I assume that this act is permissible if and only if it is parr of the larger course of action of obtaining the release of P. If, for example, you go on to throw away the key, sell it, put it in your scrapbook, then your stealing of the key (and nor attempting to release the person) was impermjssible. I tbus assume that Wsccaling 1h~ key > Rsrcaling the key' and thus that "ma.king it possible for you to release P" either isn't an R-feature of the act or, if it is, doesn't when combined with the other R -features of the act, override its IV -features. But even though IV stealing the k., > R,,.aling the kep it may be permissible to steal the key. (2) A group of officials torture some person they know to be a
terrorist, in order to discover the plans (which they know are about to be executed) of a terrorist group which they can then thwart, thereby saving many innocent lives. Assume that the officials are good, the terrorists are bad, and that saving these lives outweighs torturing the person. When the torturing is part of the larger course of action of saving the lives, it is permissible. 1, however, the officials torture the person, obtain the information, and then do nothing with it, or just file it away, then the torturing was im permissible. Thus W,0,'"''"~~ > R,0"urin~, yet the torturing may be permissible if it is a necessary part of a larger course of action B where Rs > W B Thus it seems that IV A > RA is not a sufficient condition for the imper missibility of A, for A may be a necessary part of a larger permissible course of action. Can it be, on the other side, that parallel considera tions about larger courses of action also prevent R_, > IV A from being a sufficient condition for A's being permissible.38 One thing which might suggest itself is that though RA > IV A, A is impermissible bec.ause for every larger course of action A 1 of which it is a part, IV~ > R~. This seems to me to be a possibility not worth raking seriously. For it is difficult Copyrighted material Moral Complications and Moral Structures -& 227 to believe that such a situation would not be reflected in the W -features
of A itself. 39 But perhaps though A is part of some larger acts C which have Rc > We, each one of these has as an alternative an act (not containing A) which stands in the same relation to C as 8 stands to A in condition (Vll) of section IV. We want to say roughly the following: (1) RA > WA , but A is not permissible because (38) (8 stands to A in the relation described by principle (VII) and 8 is permissible). (2) RA > WA, but A is not permissible because all courses of action of which A is a pan are impermissible by ( 1 ). (3) WA > RA, but A is permissible because it is part of a larger 8 (where R8 > W 8 ) and 8 is permissible. We cannot stop here in explaining the exceptions to (a) RA > WA being sufficient for A's being permissible (b) WA > RA being sufficient for A's being impermissible for each of (1)-(3) has the word " permissible" after the " because," and it is the application of this word we are trying to account for. And we cannot eliminate "permissible" or "impermissible" after the " because" by rewriting (1)-(3) and substituting" R, > Wx'' for each occurrence of "X is permissible" after the "because" and substituting "W,. > R, " for each occurrence of "X is impermissible" after the "because." For our problem is just that tltese are not sufficient condirions for permissibility and impermissibility, respectively, and we cannot state principles gov
eming the exceptions to them as sufficient conditions which assume that there is no such problem. For example, the act 8 referred to in (1) must not be one shown to be impermissible by (2); (2) explicitly refers to (1); the act 8 referred to in (3) must not be one shown to be impermissible by (1) or (2), etc. One gets complications piled upon complicationsthough fewer than if we had taken the course of rejecting the Second Asymmetry claim at the end of section fV. How might these complications be handled? Let us 6rst define some nonons. (1) 8 underc11tso A = df. A and 8 occupy the same time interval andRA >IVA & Rs > IVs & IVs WA & ( IVA - Ills)> (RARs). (2) 8 strongly ut~dercuts, A = df. 8 is an act over the time interval t, and 8 does not contain A as a part, and no part40 of 8 come.s Copyrighted material 228 -b ETHICS AND POLITICS before A, and no part of A comes before 8, and (X) (X is an act overt containing A as a part-> 8 undercutso X). (3) A is strongly undercut,= df. (38) (8 strongly undercuts, A)41 (4) C begins a strong undercutting, of A= df. (38) (8 contains C & C begins 8 & 8 strongly undercuts, A). (5) A is strongly undercut?:., = df. (t') (t' is an interval beginning when A does and extending at least up tO t -> (38) (8 strongly undercuts; A)) (6) C begins a strong undercutting>, of A= df. (t') (t' is an interval beginning when A does and extending at least up to t --+ (38) (8
contains C & C begins 8 & 8 strongly undercuts; A)) I now want to define a special kind of strong undercutting>, of A. Roughly, it is one, begun by some C, which gives at least one course of action continuing through the various time periods ?:. t, such that each segment (conrinuing up to r) of this one course of action strongly undercuts A . 1 do not see any way to define this notion using only the apparatus of first-order quantification theory. Suppose C begins a strong undercutting:::, of A. Then for eacb time period I; beginning when A does and extending at least up to t, there is at least one action 8 which contains C as its beginning and which strongly undercuts,, A. There may be more than one such action. Let S; be the set of all such actions; i.e., where 11 is a time period beginning when A does and extending at least up to 1, S; = the set of all actions which contain C as their beginning and which strongly undercut,, A. We now want to define the one course of action spoken of above, which will be represented by a selection set from the family of the S; (a set containing one member from each of the S; ). There will be one such course of action if and only if there is a selection set which represents
one continuing course of action. Let us denote by 8; the member of this selection set coming from the set S;, and denote the selection set itself by S. S will represent the one course of action we want if and only if [11 begins the interval t1 if and only if 8; begins the course of action 81]. I assume that each time interval begins when action A begins. We define: (7) C begins a sequmtial strong underetttting>, of A if and only if C begins a Strong undercutting?:, of A and there exists a selection setS from the family of the S; such that (8;) (8j) [8; ~ S & Copyrighted material Moral Complications and Moral Structures ~ 229 B i E S --> (t; begins the interval 1 i = B; begins the course of action Bill Given these definitions, we may now state: Principle 1: If RA > WA & (3C) (31) (Cis available to the person and C begins a sequential strong undercutting>, of A) then A is impermissible. Can the antecedent of Principle I be weakened so as to yield another valid principle? If we eliminate the word "sequential" in Principle I, we get a principle which would hold A impermissible in the following sort of situation (see Figure 10.1), where downward paths represent courses of action. Suppose C undercutso A. C & C1 strongly undercutst
A, but C & C2 does not; C & C2 & C21 strongly undercuts2 A, but no action over 12 of which C & C 1 is a part strongly undercuts2 A; C & C1 & C12 & C121 strongly undercutsJ A, but no action over IJ of which C & C2 & C21 is a part strongly undercutSJ A, and so forth. Thus in this case, C begins a strong undercutting~o of A, but there is no r such that C begins a sequential strong undercutting~, of A. In this case, there is no one course of action one can recommend to the person as an alternative to beginning his course of action with A. That A is impermissible in this situation seems doubtful. A t, r, t, If in the previous example, node C2 and everything that follows it is eliminated, then we get something between strongly undercutting>, and strongly undercutting,; namely, cyclical undercutting. For in this case, Copyrighted material 230 ~ ETHICS AND POLITICS for each odd i, C & Ct, begins a strong undercutting; of A, whereas this is not the case for even i. I shall not bother to define precisely the notion of a cyclical undercutting, or of a sequential cyclical
undercutting. It seems at least as doubtful that either of these can be substituted for a sequential strong undercutting:::;, in Principle I. It is worrh noting such possibilities, just because it emphasizes how difficult it would be to make moral judgments if we knew they arose. A tractable world (morally) would be one in which (a) Any action that is weakly undercut>, is strongly undercut:::;; for some 1' including 1 (b) There are no infinitely extended undercutting cycles (c) For every act A which is strongly undercut>, there is an act C which begins a strong undercutting>, of A (d) If C begins a strong undercutting>, of A then C begins a sequential strong undercutting~, of A. Nicer yet would be one in which (e) If 8 undercutso A then there is a 1 such that 8 begins a sequential strong undercutting>, of A. But this seems too much to hope for. I wish to leave as a question for further consideration whether the antecedent of Principle I can be weakened so as still to yield a valid principle. One would hope so, since the definition of "sequential strong undercutting::;," uses very powerful machinery, which it would be better to avoid. (And it is not clear, for any action A such that RA > WA, how to rule our the theoretical possibiliry that there is an action C and
a time interval I such that C begins a sequential strong undercutting::;, of A.) We should note that many problems that arise in the area of the subject of this section could be avoided if there were a timediscounting of rhe moral future. Questions about this issue, interesting in their own right, must be considered in a full development of the theory. For the time being I shall suppose that no weakening of the antecedent of Principle I will do. (If this tentative supposition is false, what is said below can be modified so as to correspond ro rhe new principle with weakened antecedent.) I suggest the following two principles: (l) If R,.. > WAthen [A is impermissible if and only if (38) (31) (8 is an action available to the person & 8 begins a sequential strong tmdercutting:::;, of A)) Copyrighted material Moral Complications and Moral Structures ~ 231 (2) If IVA > RA then [A is permissible if and only if (3C) {C is available to the person & A is part of C & Rc > We & - (38 ) (31) ( B begins a sequential strong undercutting::, of C)))42 (1) replaces principle (VII) of section IV. Let me close this section with a few remarks. A. We have concluded that RA > IVA is not a sufficient condition for A's being permissible, and that IV A > RA is not a sufficient condition for A's being impermissible. So does any asymmetry remain?
The asymmetry rhat remains is that RA > W A can be shown to be insufficient for the permissibility of A by the finding of a suitable action B over the same time interval as A which is an alternative to A (where " suitable" means, among other things, that there is no act which begins a sequential strong undercutting?;, of B); whereas IV A > RA is insufficient for the impermissibility of A only if A is part of a lo11ger course of action meeting a certain description. Put roughly, if the moral future of an action were like its moral present, IV A > RA would be a sufficient condition for rhe impermissibility of A, whereas RA > IVA would not be a sufficient condition for the permissibility of A. B. Our method of measuring rhe moral weight of a set of fearures was suggested by principle {Vll) of section IV. Does the substitution of ( 1) above for this principle change things so rhat the method proposed no longer works? It does not if, in our example to elicit a person's judgments about the permissibility of A, we are careful not to produce cases where RA > Rs, IVA> Ws, Ws IVA, and the person may believe that rhough consideration of just A and B makes A impermissible, considering longer course.s of action changes things. I shall not enter here into
the details of how the example to be put to the person should be constructed so as to avoid his having this belief. Using the terminology we have defined, we do not want to present him with e.xamples where he will believe that though B undercutso A, B does not begin a sequential strong undercutting>, of A. It should not be roo difficult to avoid such examples. C. Suppose W A > RA yet (38) {A is part of B & Rs > W s & B is not impermissible (by 1)). According to what we have said, the act A is not impermissible. But the act A without doing the rest of B {or some such B) is impermissible, and can be shown to be so by our principles. Often, for short, in situations like this where the person does not do the rest of B (but only does the part which is A), we elliptically say that his act A was impermissible. Copyrighted material 232 ~ ETHICS AND POLITICS D. Finally, let us return to the questions with which this section opened. Can we say that "is part of a course of action which is such that --" is a feature of an action A, and if so, how is the blank to be filled in? U the argument of this section is correct, in the case where it seems that W A > RA yet A is morally permissible, the candidate for the
extra R-feature of A is F = "is part of a larger course of action C, where Rc > We, and there is no B and I such that B begins a sequential strong undercutting~, of C." But F is nor morally neutral since it refers to the R and IV -lists, and this reference cannot be eliminated for reasons similar to those advanced in section IV. Thus I conclude, once again here, that one cannot state everything in terms of the Rand W -features of act A, and the introduction of higher-order principles43 is necessary in order to get things straight. VIL Overriding, Outweighing, Neutralizing, and Related Notions We bave thus far gone along with the simple model's use of one inequal ity relation between sets of features of actions. But there are significant differences among the ways in which the presence of other features (or the obtaining of certain facts not represented by features of actions) cao make ao action morally permissible, even though the action bas some IV features. An adequate model of the structure of the sort of moral view I am considering must in some way differentiate among some of these different ways, not all of which are happily classified under the rubric of "overriding" or "outweighing." In this section we shall list some of these ways, somewhat arbitrarily labelling them, and
consider some of the problems they raise. We shall not, in this section, discuss how they are to be incorporared into an adequate model of the struc ture of a person's moral views. Thus, we are now concerned with setting problems and raising questions. The ways that I list are "pure" ways, and we shall not, at this point, be concerned with various combinations of them. I. An act A, with W-fearures, prevents, avoids, etc., something bad, harmful, ere. It has no orher R-features. If the act is morally permissible, let us say that irs R-fearures override its IV features; if the act is morally impermissible, let us say rhat its IV-fearu res o"tweigh its R fearures.44 Copyrighted material Moral Complications and Moral Structures ~ 233 It was to these notions (as well as to those in n below) that our previous discussion was especially meant to apply. Relevant distinctions to make are whether (a) something like principle (VII) of section IV is required (b) if theW-features of act A involve something bad to a person P, the performer of A is morally required to make either reparations toP, or amends toP, or tO offer explanations tOP, etc.45 I beHeve that in all cases where an act A, with WA-f, avoids, prevents, etc. something worse (i.e., where its W -features are overridden), then something like principle (VII) is required. But even if this is
correct, we cannot use principle (VU) to explain the notions of "overriding" and "outweighing," and this is not solely because the principle may legitimately apply to some other notions as well. The major obstacle to doing this is the following. For some of the ways we shall list in this section of a feature's making an action A morally permissible, even though W A ,P , some of the features which in these selected ways make A permissible do not belong on what we intuitively have in mind as the R-list. We want to first explain these various ways, and then to limit what can go on the R-list by excluding some features which play a role only via some of the selected ways. Since principle (VU) uses the notion of the Rlist, we do not want to use it to explain one of the ways (viz., overriding and outweighing). Are there actions where (a) and (b) do not go together? I believe that (b) cannot be found without (a); that is, whenever your doing A with non-null wrong-making features makes it incumbent upon you to explain your action to whoever is harmed by its W -features, or to make amends or reparations to these persons, then it is also the case that it would be impermissible to do the action if a suitable alternative (as defined by principle VIJ) were available.
Can we have (a) without (b)? Can it be the case that you must do a suitable alternative act to A if one were available to you (and since one isn't, it's morally permissible to do A), yet even though A's Wfeatures harm someone, you have no duty to make reparations or amends or to explain to them why you've acted as you did? A plausible candidate for such a case of no duty to make amends or reparations would be good samaritan cases. ln saving someone's life, I damage your property. I could not save that person without damaging your property or doing something at least as bad (though if I could have, it would have been wrong for me to damage your property). It may well be that I am not Copyrighted material 234 -<> ETHICS AND POLITICS morally required ro make amends or reparations to you.46 However, even in this case one would think that I am required to give you some explanation of what's happened. In speaking of what I am required to do, I have put things too strongly. "Omitting to make reparations, amends, or explanations to those harmed by IV -features (specifically listed) of act A" is a W -feature of a course of action, and functions
just like other W -features. It can be overridden, etc.: it may be too dangerous to make explanations, too inconvenient given the slight harm caused, etc. II. Act A, with IVA f 4>, though it does not prevent, avoid, etc., some harm, achieves some good. lf (subject to the same qualifications as with I) act A is morally permissible, let us say that its R-features overcome its IV -features; if A is morally impermissible its IV -features overshadow its R-fearures. Here, as in I (if not more so) one feels that (a) and (b) of I obtain, and perhaps that (b) in this case is stronger than it is in I. That is, that in this case there is a stronger obligation to make amends or reparations or offer explanations than in I,47 and perhaps an obligation to make greater amends or reparations. III. On some views, for some W -features (relations) T which take someone as a direct object, it is morally permissible to T someone who has T-ed you (or perhaps has only T-ed some other people), where the T of his act helped make it morally impermissible.48 On such very contractural views, it m.ight, for example, be permissible to steal from a thief, torture a torturer, etc., without oneself becoming someone who is open to permissible thievery, torture, etc. Without worrying now whether it
is a feature of the act or a fact about the situation which does so, let us say that in such situations ofT -ing someone who T 's, T is neutralized. Certain retributivist views would hold, nor only that sometimes it is permissible to T one who T's but that it is sometimes obligatory to T him, or to do some act with some other IV -feature G. Thus on this view, it is wrong not to punish someone for certain offenses; and this apart from deterrent considerations; he just deserves it; and justice demands that be get it.49 Again, without worrying over what it is that does so (is it a feature of an act that it's done to someone who has comm.itted a wrong, or is it a fact about the situation?) let us say that in this case the IV -feature of the act is reversed. Sometimes T wi.ll be neither neutralized nor reversed, but it will be the case that T -ing someone who T's is less wrong, carries less IV -weight Copyrighted material Moral Complicotions nd Moral Structur<s ~ 235 than T -ing someone innocent of (wrongful) T -ing. Let us say in this case that T is weakened. IV. Suppose T is on the W -list, and takes persons as direct objects. Suppose further that each of the persons who are the objects of T consent
to being T-ed. It seems that sometimes this will have the consequence that T carries no moral W-wcight, and sometimes it will have the consequence that T carries IV weight, but less than it would in the ooncoosent case. In the first case, let us say rhat T is dissolved; in the second case let us say that T is consent-weakened. V. SuppoS(' that I bave promised you that I will do an act A. You release me from tbjs prorruse, and there arc no third-party beneficiaries. Let us say that in this sort of case the feature ~nor keeping my promjse" is catralled. VI. You extend yourself to do me a good rurn. I am onder an obligation to return it, if I can. Suppose you then intentionally go out of your way to (wrongfully) harm me. Let us say that in this case F ="omitting to reciprocate an unreciprocated good turn" is destroyed (from: The obligation is demoyed). VIJ. I promise m meet you next week at a certain place and time, so that we can do something together. Before that rime, I learn that you have died. let us say that F = " involves omitting to keep my promise to meet you" is nullified. VUl. You lied to me in order to get me to promise to do A. I believed your lie, and made rhe promjse. Let us say that "involves nor
keeping my promise to do A" is invalidated. so IX. Unril now we have considered different ways in which the pres ence of features or facts either may lead to an action A's being morally permissible (even though A has features on rhe W-list), or may weaken the moral weight of some W -features. It may be that an adequate account of moral structure must also consider a quite differeot kmd of relation. Perhaps some features or facts can undermine the operation (in one of the above-menrioned ways) of other fearures or facts upon IV -features. Features or facts F ope raring in one of the ways tliscussed in I through VJD upon the W -feature of action A mark off exceptions to the rule that any act wirh all the features in IVA is morally impermissible. Bur perhaps there are also exceptions to these exceptions; rhat is, facts or features wrucb, if present, prevent F from operating, as it normally does, in one of the ways upon IVA. ....opyr grled fT' 11'r at 236 -& ETHICS AND POLITICS Consider, for example, Section 3.04 of the American Law Institute Model Penal Code (Proposed Official Draft, 1962), concerning justifiable use of force in self-protection. Put roughly, the structure of this section is as follows:
(I) The use of force upon or toward another person is justifi able when P. (U) (a) The use of force is nor justifiable by I when (1) Q or (2) R except the R limitation upon the justifiable use of force under I does not apply if (a) S or (b) T or (c) U (b) The use of deadly force is justifiable under I only if V, and it is not justifiable (even if V) if (1) X or (2) Y except the Y limitation on the justifiable use of deadly force under I when V, does not apply if (a) Z or (b) M Let w1 = involves using force upon or toward another person; let wz = involves using deadly force upon or toward another person. Though it will often not be clear exactly why one structure is chosen rather than another, let us assume that we here have a fourleveled structure, and that there is good reason not to collapse it into fewer levels. We rhus would have: ( 1) P stands in one of the ways to w1 (2) Q (R) undermines P's standing in one of the ways to w1 (3) S (T, U) upsets R's undermining of P standing in one of the ways to wl (4) P & V stand io one of the ways to wz (5) Q (R, X, Y) undermines P & V's standing in one of the ways to
wz (6) Z (M) upsets Y's undermining of P & V's standing in one of the ways to wz. Copyrighted material Moral Complications and Moral Structures ~ 237 Upsetting wiU be a four-place relation (or more if one adds extra variables, e.g., one ranging over rhe ways). An important question is that of how many levels one is driven to; up to what 11 must an adequate theory use n-place relations of this sort? (l ignore the question of the possibiliry of reducing the number of places by formal gimmicks of various sorts.) Up until now we have used just two-place relations (for the ways). Is there any strong reason for not continuing to do only this? Can't we incorporate all of the above information into the domain of rhe ways, and say, for example, that (where the "or" is rhe inclusive-or) (A) P & not-Q & (not-R or S or T or U) stands in one of the ways (or a combination of them) to w1 (B) P & not-Q & (not-R or SorT or U) & V & nor-X & (not-Y or Z or M) stands in one of the ways (or a combination of them) to W2 This issue must be discussed in a full presentation of the theory. Here let us just note that many of the reasons that led us, in section I, away from the deductive structure to one with open-ended lists of features,
outweighings and overridings, etc., will apply here as well.51 For reasons similar to the earlier ones we may want an open-ended list (for each feature or set of features?) of what can undermine a feature or set of features' standing in one of the ways to W -features, and openended lists of what can upset such underminings. If so, this would prevent us from working just with things like (A) and (B) above. Obviously, if we take the course of not working only with things like (A) and (B) above, we have the pre.ssing problem of describing the more complicated structure which is to substitute for the one which uses only the two-place relations. Many other two-place notions in addition to those in I-Vm above might be put forth; e.g., we might say that a W -feature is precluded in cases where, e.g., it is impossible to keep a promise (can't implies notought cases). My purpose here is not to proliferate notions for its own sake, but to raise issues which require further consideration. In addition to the ones mentioned, our brief discussion in this section raises several other serious issues. I shall do little more than list them here. A. What are me various ways in wh.ich the presence of some features of an action A may, ceteris pari/ms, prevent the action from be ing morally iotpermissible, even though WA #if>? How are these different ways to be distinguished and embedded within a general theory
of moral structure? Which of the ways S1 (viewed as relations in which Copyrighted material 238 -& ETH ICS AND POLITICS things can stand in toW-features) must have only R-features in the S;image of IV? For example, one would not expect the relation in V and VII above to have only R-features in their image of IV. What principles can one formulate which govern each of the ways? Which ways require special principles, and which principles? How will the theory handle combinations of ways, and what special problems does this raise? B. How are "acts of omission" to be handled within the theory (consistently with our discussion of features in section Ill)? C. For features (or facts or conditions) which, according to IIIVlll above neutralize, reverse, cancel, dissolve, destroy, nullify, invalidate W -features, should the denial of the.se features (or facts or conditions) be built into the W -features themselves? I shaU not attempt to list here the plethora of considerations (in addition to the ones related to those discussed under IX above) relevant to this question. D. We have spoken above of some features (or facts) weakening the moral weight of IV-features. There seem to be at least two ways in
which one might try to handle this. (1) Each IV -feature, as we II as each set of IV -features, always bas one moral weighr. What looks like weakening (or the lessening of the weight) is just the result of something's being put on "the other side of the scales." The " moral scales" come to rest where they do, not because some weight is lessened but because some extra weight is put on the other side. But this view would have the consequence that it is the one unique full weight of the IV -feature which must be used in something like principle Vll of section IV, and which also determines what amends, reparations, etc. should be made. This consequence would certainly be implausible for the cases which someone holds fall under weakening in ll1 above. For if, according to such a person, some IV -feature w1 of some act 8 is weakened, one should not, in determining whether some alternative action has less weighty IV-features (under principle VII) than action 8 has, treat w1 of 8 as having its full weight. For there might be an alternative action A such that the full weight of w1 is greater than that of IVA while the weakened weight of w1 had by 8 is less than that of WA. Thus it seems that the method of explaining and handling weakening put forth by (1) has undesirable consequences when combined
with some principle like principle vu of section rv. Copyrighted material Moral Complications and Moral Structures ~ 239 (2) The alternative possible way to handle weakening is to admit that the weights of some features are really le.ssened (and not just partially compensated for while remaining the same, as in (1)). The obvious questions are: How is this possible way to be specified in detail, and how must the structure be modified to accommodate it? How does this affect the program of measurement set forth in section V? It seems that one could, by determining a person's judgments only for cases in which weakening does not operate, obtain the same measurement results as before. Can a similar measurement procedure be devised to measure the moral weights of weakened features in specific situations? Do we also need a notion of the stengthening of a W -feature, of its weight being increased, and if so, how is this ro be incorporated within a systematic theory of moral structure? E. In sections II-VI we have considered only other features or sets of fean>res overriding or overcoming, etc., W -features. Must one consider things other than features of actions in order to account for a person's judgments of moral impermissibility, e.g., facts about the situation which are not happily incorporated into feature.s of actions? If
so, what apparatus is needed to handle this extension? F. We have thus far avoided discussing the issue of whether some of the fean>res on the list must incorporate a person's beliefs. That is, we have neglected the issues that have led into the morass of discussions of subjectively right (ought) and objectively right (ought). How are such issues to be handled within the sort of structure we are discussing? Can some of the distinctions made in I-IX of this section show a way through these issues; i.e., might belief be appropriately included for some of the ways (and underminings, upsettings) and not for others? If so, this might be relevant to some other questions mentioned earlier; e.g., to whether the denial of neutralizing, cancelling, reversing, dissolving, destroying, nullifying, invalidating features or facts should be built into the W -feature. VI II. Further Issues It is unnecessary for me to restate, ro the reader who has come this far, that this paper is an exploratory study meant to raise further issues for investigation as well as to propose tentative solutions to certain problems. Now I wish to indicate some questions and issues, in addition to those left open in previous sections, which require further study. Copyrighted material
240 ~ ETHICS AND POLITICS (1) What is to be said of the possibiliry that always or sometimes there's not one feature of an act such that if an act has it, it is, ceteris paribus, morally impermissible, but rather that a conjunction of features is like this? In this case, should the W -list consist of conjunctions of fea tures (with some different entries on the lists having common conjuncts) or is there some simpler apparatus to handle this? Similar questions, obviously, can be asked about the R-list. (2) Given the kind of uniry and coherence people's moral views have, there is some reason to want some of the entries on the lists not to be merely features (or conjunctions of features), but rather branching tree structures of features (or their conjunctions). (See Figure 10.2.) Any action which has feature F has each feature referred to by the expre.ssions obtainable by (repeated) deletion of numerals at the end of the subscript of F. Thus any action with, e.g., feature Fl2,6,9,4,7 has features F t2,6,9,4, Ftz,6,9, Fu,6 and F t2 Intuitively, Fu,6 is a way of realizing F12, etc. Figure 10.2 F,l ~ fu.} ~ . .. F F,.J ~
F t. ~ Let us call the set of all features to the right of a feature F on a tree which can be reached by following a path from node F and going to the right, the descendant set of F. One might put forth the following principle of inference within a moral system: s. > s2 S3 > s.' where S3 is a non-null subset of the union of the descendant sets of the members of s., and s. is a non-null subset of the union of the descenCopyrighted material Moral Complications and Moral Structures -& 241 dants of the members of S2. But reasons similar ro rhose which led to rejection of the deductive structure in section I and to the consideration of the simple model in section II, may make such a principle of inference too strong. If so, and tree structures of features arc appropriate ways of representing the unity and coherence of a person's judgments, what rules are to govern them? (3) Given open-ended lim of features (and tree srructnres of fea tures?), some of which contain variables (e.g., "leads to the death of n-persons"), is the number of entries on each list finite or io.6nite? Would a correct and complete moral view require finite or infinite lists? (Can there be a correct and complete moral view?) lf infinite, arc
the lists recursive? recursively enumerable? are complete and correct moral systems finitely axiomatizable? Something must be said about what kinds of reduced lists are permissible. What criterion can be formulated which would al.low "leads to the death of 11-persons" on rhe Wlist (and wouldn't require that, for each number m, its instantiation for m ap pear on the W -list) yet would not have just one utilitarian feature on the list even if the person was a Sidgwicktype utilitarian and utilitarianism were true? (4) What is one ro make of talk of open-ended lists of features? For at any given time, presumably only finite specific lists are needed to ac count for the judgments the person has actually made previous to that time. The openness of lists was meant to mark off a problem. It seems that given any finite list that we or he could construct, there is a certain situation which if described to the person, would lead him to say that some featnre not on either presented list was a morally relevant feature of the situation: "Oh yes, I overlooked that one." Though this fe.ature wasn't necessary to account for any of his previous judgments, in some
way the description of his view must "include" the feature, must rake account of the fact that he would accept the feature as morally relevant, and he would rake it into account in arriving at a judgment about a situation which exhibited it. It is not clear what wellunderstood for mal device best achieves t:be legitimate purposes of the vague talk about open-ended lists. (For some persons, nothing like open-ended lists will be appropriate, and indeed some sort of closure condition will be required; e.g., "nothing belongs on theW-list except F~, . . . , F,,.") (5) How is the structure to handle second-order (and higher?) features; e.g., the IV -fe.atures: persuading someone to do something impermissible., forcing someone to do something impermissible, offering someone something to do an impermissible act, glorifying, praising, Copyrighted material 24Z ~ ETHICS AND POLITICS rewarding impermissible acts, punishing, ridiculing required actS, leading someone to abandon correct moral principles? Sometimes the particular first-Qrder features needn't be specified, or specified precisely, ro accoum for the person's judgments; sometimes they ca.n't be (as when one person persuades another to do some unspecified impermissible act), and SQmetimes it will be important to know what exactly the
first-order feature is. How should all this be represented within a structure? ( 6) What sorts of structural cQnditions using the notion of necessity are appropriate? Would it be appropriate to require that if necessarily( Fx - G x) then F and G override, are overridden by, exactly the same sets of features, etc.? (7) Most persons apply different moral standards to different kinds Qf beings, e.g., children and adults. Some persons may do so for different groups of adults, the distinctions depending upon, e.g., social class, religion, race. Should this be represented (ignoring the additional ways discussed in section Vll) by one pair of lists (one member of the pair being a W -list, and the other an R-list) where the lists contain "conditional-features," or by separate pairs of listS for each group for which there are distinct moral standards? (8) One should define various notions of a complete moral system (one yielding for every possible case of a certain sort a judgment that the act is morally permissible or morally impermissible). Different notions will be defined for different kinds of information fed into it. For example, one notion would be that, for given lists of Rand Wfeatures, for any two arbitrary subsets of R and W, the system yields an inequality
between them. (See condition 2 of section HI. ) Further notions would be defined for risk and uncertainty situations. Are moral dilemmas or drawing-the-line problems reasons to believe that there does nor "exist" (everyone grants that we don't know one) a complete and correct moral system? (9) One wants some way of describing ways in which the members of a list can cohere. Various writings on the coherence theory of truth would be sugge.stive here. One might define a W -feature's being a taboo as its not cohering in certain specified ways with some other members of the W -list. (10) The structure described is concerned with the relatively gross (though difficult enough!) distinction between moral permissi bility and moral impermissibility. Various extensions immediately suggest themselves, e.g., tO degrees of moral impermissibility; and to other finer Copyrighted material Moral Complications and Moral Structures ~ 243 distinctions. 52 How must the structure be supplemented to account for the.se more relined judgments? (11) We want some way to represent a person's belief that nothing, or something, or nothing or something of a certain sort override,s, outweighs,
etc., some set of features, and to incorporate such beliefs within the general structure, and have them be one of the th ings the structure accounts for. Various obvious ways suggest themselves. (If nothing override,s a set of features, this set must get special treatment in the measuring process of sec6on V.) (12) We have thus far considered problems stemming &om a person's lack of confidence in the correcmess of stared exceptionless moral principles due to the wide variety of cases which are possible or may arise. He can't anticipate all possible cases, etc. But a person may also have varying degrees of confidence about a specific feature's being on one of the l.ists or about a specific outweighing (which he may strongly, tentatively, or with some hesitation accept). We want some systematic way to make sense of different degrees of confidence, to measure them, and we want to formulate principles specifying the role they play in generating' a person's moral judgments. 53 (13) What, if any, are the interesting ways in which two particular moral views having the same general structure can differ if they generate exactly the same moral judgments about all possible particular acts? (14) May itrdexical expressions or proper names be built into fearures
on the lists; e.g., "harms my wife," "involves eating on Yom Kippur"? What constraints are there on how they may appear, and on what else must appear if they appear? (15) How are double-effect type issues to be handled within the structure? Must intentions be introduced within the features? (16) One is tempted to use the structure as a weapon in discussion. (R. M. Hare's use of a similar weapon in his book Freedom and Reason is too strong.) One is tempted to say that if a person judge.s that acr A is morally impermissible and act 8 is morally permissible, then it's incumbent on him to produce a fearure on one of the lists had by one of the actions and not another. The argument for this would be that yot< cannot go completely through the two open-ended lists showing him that A and 8 have exactly the same list fearures, but he can produce the feature which distinguished A and 8. But if he produces a feature wbicb is had by A and not by 8 , and you deny that this feature is on either list, is it incumbent upon him to produce an argument for this feature's Copyrighted material 244 ~ ETHICS AND POLITICS being on one of the lists? Or is it incumbent upon you to produce an
argument for tbe feature's not being on either list? Difficult questions arise about where, if anywhere, burdens of argument or proof lie. And how are we to explain: "it is incumbent upon bim to"? Is it just that he be able co (in how long a time?), or that not doing it when requested is a W -feature (which can be overridden; e.g., he has more important things to do)? And what follows from a person's being unable to discharge his burden; that one of his judgments is mistaken (surely this doesn't follow), that one of them is unsupported, that he should shut up, that he's not serious? (1 7) Condition VD of section IV, though growing out of the earlier discussion, was formulated without any requirement of the relevance of the alternative act B to the R-features of A. Thus, someone might object that condition VII owes its plausibility co the assumption that 8 is a substitute for A, and ach.ieves the same or similar goals, and if this is not so, then VII is too Strong. It is certainly worth investigating the systematic consequences of adding to condition VD various precisely formulated requirements concerning the relevance or similarity of Rs to RA.
(18) Some people's views may exhibit different metarules. To rake rwo simple examples: (A) Permissive Rule: If the system of inequalities and principles does not yield the result that an action whose only morally relevant features are all and only those in set S is morally impermissible, then any action with just those morally relevant features is permissible. (B) Strict Rule: If the system of inequalities and principles does not )ield the result that an action whose only morally relevant features arc all and only those in set S is morally permissible, then any action with just those morally relevant features is impermissible. (Of course ir may be that a person's views exhibit no such rule.) One might attempt to represent a person's views which might be interpreted as exhibiting one of these metarules as containing some rule of inference yielding inequalities which complete his incomplete set of inequalities. What reasons might be advanced for preferring the metarule representation, which does not involve adding first-level inequalities, to the rule of inference representation, which does? Copyrighted material Moral Complications and Moral Structures ~ 245 ( 19) Can one incorporate some legitimate device which functions to produce self-reference and arrive at an action A which can be interpreted
as the act of judging that A is morally impermissible? If so, one will get the result that either there is some morally impermissible act wh.ich it is morally impermissible to judge (nor just say) is morally impermissible, or there is some morally permissible act wh.ich it is morally permissible to judge is morally impermissible. (20) Much work needs to be done on the subject of the "logk" of presumptions, accounts of what it means to say that presumptions can be overridden only by special reasons, etc. The relevance to the sort of structure we have been describing is obvious. (21) Until now, I have spoken freely of the model's accounting for a person's judgments about the moral permissibility or impermissibility of particular actions. For various reasons the model cannot do this, and requires supplementation. Consider the following inference.s4 (1) Act A has features Ft. ... , Fn. (2) Each of these features is on one of the lists. (3) ~wbset~the~on~W -list>~subserofthe~oo~ R-list. ( 4) A has no other features, in addition to Fh ... , F., on either list. Therefore, (5) A is morally impermissible. Suppose that a person knows that (1), and suppose further that (2) and (3) are true of the particular structure correctly ascribed to him. Suppose that this inference pattern is built into the structure of
his views, in that if he also knows ( 4) then he makes, or would make, the particular judgment (5). (Of course this particular inference pattern won't be bui.lt into the structure of his views since it ignores the complications about larger course.s of action and alternative actions.) But the person may not know that (4). It may be that (a) The person believes, but does not know, that (4). (b) The person knows that A has some other features on the lists though he does not know which features. (c) The person believes, though he does not know, that A has some other features on the lists, and he does not believe, for some other particular features on tbe lists, that A has these. Copyrighted material 246 ~ ETHICS AND POLlTICS (d) The person does not believe that A has some other features on the lisrs, nor does he believe that A has no other features on the lists. It seems to me likely that a person in situation (a) will judge that A is morally impermissible (though if he has little conJidence in his belief, perhaps be will not). Situations (b) and (c), of course, encompass many interesting cases. Looking just at (c): The person believes that A has some other features on the lists, though he does not believe, of some specific other features on the lists, that A has these.
We might have, in addition, that the person believes any one of the following: Each of the other features is on W. Each of the other features is on R. Some are on IV and some are on R. Some are on 1-V (and he has no beliefs about whether others are on R). Some are on R (and he has no beliefs about whether others are on IV). Or we might have that the person knows that if A does have some other features on the lists then these features are such that --(with various possible fillings in of the blank). And for each of the cases, he may have beliefs about some of these (unspecified) properties outweighing or overriding others, about bow the inequality goes when some are conjoined with the ones he specifica lly knows of, etc. It seems clear that for situations (b)-(d), some further apparatus must be conjoined with the structure we have been discussing, in order to yield a person's particular judgments about particular actions. The de rails of this apparatus may vary from person to person; people may differ in how they make judgments in situations (b)-(d). A similar ar gument can be offered for simplified inferences yielding the
judgment that an action is morally permissible. When the complications about al temative actions and larger courses of action arc taken into account, it becomes even more clear that, in order to account for a person's par ticular moral judgments, the structure we have been discussing must be conjoined with a further apparatus about how the person arrives at beliefs in, and what he assumes with what confidence as a basis of inference in, situations of incomplete knowledge. Copyrighted material Moral Complications and Moral Structures -& 247 Secondly, we must further delimit tbe panicular judgments to be accounted for by the structure we have been discussing. One would not expect the strucru.re to play a major role in accounting for a person's belief that a particular action A is morally impermissible, which belief the person holds because someone he trusts told him that A is morally impermissible. 55 Thirdly, one needs a distinction similar to that which linguists make between linguistic competence and linguistic performance.56 The actual judgments which a person makes will depend upon various limitations common to all persons, and upon some special to him (or at any rate, not common to all); e.g., limitations of attention
span, of memory, limitations on the complexity of information which can be manipulated and processed, limitations on the amount of time he is willing tO spend thinking about moral problems. The actual judgments be makes will also depend upon various exogenous factors; e.g., whether he has a headache, whether there's noise which prevents him from thinking as clearly as he otherwise would, whether he's interrupted while thinking and loses some thought which he doesn't later remember. We may think of the structure we have been discussing either as a model of an idealized moral judge in idealized circumstances (i.e., ignoring the various limitations and exogenous factors), or as 011e component of an adequate psychological model of the person. (22) One needs a discussion of what it means to ascribe such a structure tO a person, of what it means to say he " internalizes" such a structure. It doesn't mean that he always arrives at his moral judgments by explicitly and consciously referring to such a structure. I do, however, want the assumption to be based upon something more than the claim that the hypothesis that there is such a strucrure somehow realized inside him most simply and elegantly accouors for the moral judgments he ma.kes. It would be upsetting, in this case, if the structure could nor
account (when combined with further theory) for his conscious reasoning in arriving at moral judgments, and the sort of considerations he adduces [and the way in which this is done) in support of his moral judgments. Put vaguely, the structure shouldn't be foreign to the way he actually reasons about moral matters, and he shouldn't find it, when pre.senred to him, foreign. (23) The obvious ways in which the sort of structure we have been discussing might be found defective is that it can't account for some of the person's judgments about moral impermissibiliry (and this is true, Copyrighted material 248 ~ ETHICS AND POLITICS for every person), or it " accounts" for more than he would make, or a simpler alternative structure accounts for his judgments about moral impermissibility in a neater and more elegant way. It is worth mentioning rwo other ways in which the sort of structure we discuss might be found to be inappropriate (and these ways qualify what is said in the previous sentence). The structure is designed to play a role in accounting for only some of a person's moral judgments; viz., those about the permissibility and impermissibility of actions. It does not rrear of judgments about goodness, virtues, ideals,
responsibility, etc. It might tum out that the simplest total structure which accounts for all of a person's moral judgments which has the strucrure we have been discussing as a part, is less simple, elegant, or adequate than an alternative structure which accounts for all of a person's moral judgments and which does not contain our structure as a part. Secondly, as I said at the very beginning of this essay, I do not claim that everyone's views about the moral impermissibi lity of actions exhibit the sort of structure we have been discussing. Some other kind of structure might account for a wider range of views about the moral impermissibility of actions while including the ones we have had in mind (e.g., it might also account for the views of persons in other cultures, or the views of all or of more persons in our culture), and this might lead one to reject the sort of structure which has been our subject.57 Copyrighted material 11 ON THE RANDIAN ARGUMENT What are the moral foundations of capitalism? Many supporters of cap italism, especially among the very young, think that these foundations have already been provided; indeed that we already possess in the writ
ings of Ayn Rand a demonstration, a proof, a cogent argument, an establishment of a moral view from which capitalism can easily be jus tified. I have two reasons for wanting closely to exa.m ine the argument: ( 1) Some persons are not devoting thought to fundamental issues about morality, thinking that the essence of the job has already been done. (2) The argument itself is an attempt to provide a non-utilitarian nonsocialcontract natural rights ethics. Since I share the view that such a moral foundation is appropriate and possible and that laissez-faire capitalism is morally justifiable on such a basis, I wish to look closely at an actual attempt. I would most like to set out the argument as a deductive argument and then examine the premisses. Unfortunately, it is not clear (to me) exactly what the argument is. 1 So we shall have tO do some speculating about how steps might be filled in, and took at these ways. It may be, of course, that I have overlooked some other ways, which would make the Copyrighted material 250 _,. ETHICS AND POLITICS argument work. If so, I presume someone else, who claims to possess and understand the demonstration, will supply the missing material.
As I see it, there are four pam to the argument. (I use the roman numerals ambiguously to refer both to stages of the argument, and to the conclusions of the stages.) (I) To the conclusion that only living beings have values with a point. (ll) From I, to the conclusion that life itself is a value to a living being which has it. (III) From 11, to the conclusion that life, as a rational perso11, is a value to the person whose life it is. (IV) From UI, to some principle about interpersonal behavior and rights and purposes. I shall examine each of these in turn. I ( 1) Only a living being is capable of choosing among alternative actions, or, (2) Only for a living being could there be any point to choosing among alternative actions, for (3) Only a living being can be injured, damaged, have its welfare diminished, etc., and (4) Any rational preference pattern will be connected with the things mentioned in (3), and since (5) Values establish a (rational) preference ordering among alternative action, it follows that (6) Only a living being can have values, with some point to them. Values have a purpose only for living things.
To make this point fully clear, tty to imagine an immortal, indestructible robot, an entity which moves and acts, but which cannot be affected by anything, which cannot be changed in any respect, which cannot be damaged, injured or destroyed. Such an entity would not be able to have any values; it would have nothing tO gain or lose; it could not regard anything as for or against it, as serving or threatening its welfare, as fulfilling or fruStrating its interests. It could have no interests and no goals. ("The Objectivist Ethics," p. 16.) Copyrighted material On the Randian Argument -t> 251 I do nor wish here to enter into complicated issues about what particular cognitive and choice functions (if any) machine,s could be capable of performing. Does coming to know that some situation bas been realized count as a machine's being "changed," in the requisite respect? Note the assumptioo that each of our values coocems how things affect us, a stronger assumption than that our va lues concern how some affectable being or other is affected. (Some views would claim to hold values unconnected with how any being is affected, e.g., religious views which hold that it's better if God is praised and worshipped, not because God is affected by this, but because it's a fitting or suitable
response to him. Or views which maintain that it's better that talented and able people be respected and admired, even if they never learn of it and even if their knowing of it doesn't (seriously) affect them; and better not because of the effects on us of doing so, but because this is a fitting and appropriate response to other people's achievements.) Presumably then I can't value, or, in a valuing fashion, act to achieve some state of affairs, in a far-off place, knowing that I shall never know whether my act bas actually succeeded or not, and knowing that in either case its consequences will not affect me. If there were an island somewhere where we couldn't go bur could send things, and from which we would not be affected in any way, then, it seems, we could not value their being moral people and having just social arrangements, sending them copies of some book whose reading we think is morally instructive. Perhaps it is being claimed, not that each value in each instance must be connected with oneself being affected, but that, for a being to have values, some of them must be so connected. So that a being which cha11ged so that it became immortal, indestruc6ble, unable to be affected, and so on could not evaluate irs alternatives and act so as to make probable
the realization of some va lue it had previously held (e.g., the lessening of injustice i.n the world) knowing that (because of isolation) it would not know whether its act had succeeded. Or perhaps the claim is that though a being can conrinue to hold values in this state,-if it was altuays in this state, and always had been, and always tcalized that it was, it couldn't hold values. It couldn't read, for example, Atlas Shmgged and become convinced by the arguments (we are not placing in doubt the machine's having certain cognitive capacity: so which part of the argument depends, for its persuasiveness, upon the reader's being vulnerable and having an affectable welfare?) and come to value and act so as to realize a world in which the ideals of that book are instantiated. But why couldn't it? M11st it say, "What's all that to me?" Copyrighted material 252 -~> ETHICS AND POLITICS Issues about the centeredness of one's goals would repay funher study (I take up some connected issues with some of the examples in section V), but l propose to proceed to see whether particular goals and values can be gotten out of (6) and a self-centered view of goals.2 n From the considerations of the preceding section a conclusion is
to be reached to the effect that the prolonging and maintaining of life is itself a value. Note, first, that it is unlikely that such a conclusion can be reached without the introduction of significant additional material. For suppose that death was a great value. (Suppose.) Only living beings could achieve it, strive for it, choose to reach it, striving for it would establish a preferenc.e ordering among alternative actions, and guide the choices of a living being. It seems that death's being a value is compatible with all that has been said in section I, and hence that the considerations of section I do not rule this out. One cannot reach the conclusion that life itself is a value merely by conjoining together many sentences containing the word "value" and " life" or "alive" and hoping that, by some process of association and mixture, this new connection will arise. There may he many other connections between life and value; the question is: what is the argument for the particular connection embodied in "Life itself is a value"? Consider the following argument: ( 1) Having values is itself a va lue (2) A necessary condition for a value is a value (3) Life is a necessary condition for having values. Therefore, (4) Life itself is a value.
But is (2) true? Are all necessary conditions for values, values themselves? If getting cured of cancer is a value, is getting cancer (which is a necessary condition for getting cured of it), or having (say) a particular virus act on one, a value? And why is having values itself a value? We might have the following argument for (1 ): Achieving values is valuable. We are not pre-programmed to achieve values. Therefore, if we are to achieve values more often than accidentally, we will have to choose actions which will do so, with doing so as their goal. But intentionally choosing actions to achieve things is itself having and acting on the baCopyrighted material On the Randian Argument + 253 sis of values, and therefore having and acting on the basis of values is itself valuable. Note that this argument, in reaching its conclusion, itself seems to use (2) as a principle of inference. The argument also begs the question against death 's being a value, for if it is, we are pre programmed to achieve it. Perhaps, instead of (2) we would have (2'). U something is a necessary condition for (achieving) all other values (any other value) then it itself is a value. Ignoring the issue of whether the question of death 's being a
value isn't thereby begged, is (2') true? (And let us pass over, for now, whether (2') is meant to rule out there being a situation i.n which dying is necessary in order to achieve one's legitimate values.) Because "not having yet achieved all valuesn is a necessary condition for achieving any given value, for achieving each value, is it itself a value? [We could also ask this question in extension, using a list of all values, if we had such a list.] Similarly, if on Miss Rand's account being vulnerable, destructible, mortal, is a necessary condition for achieving (and having) values, does it follow that this condition itself is a value? We do best, in view of these difficulties, to consider another line of argument, which we might introduce by considering an objecrion to our earlier contemplation of the possibiliry of death's being a value. "How can death be a value? Value is judged against a standard of injury, harm, etc., and the greatest of these is death." (But on what basis, which we have been given, do we know that the greatest harm isn't the extension of life's experiences?) In an essay by someone who had been closely associated with Miss Rand, which may shed light on her intentions, we find "'should' is a concept that can have no intelligible meaning, if divorced
from the concept and value of life. "3 Here we have an example of some form of transcendental argument, so beloved by philosophers and so tricky to handle and get clear about. Before proceeding, we should have before us another aid, john Hosper's sympathetic presentation in his Introduction to Philosophical Atzalysis (second edition): Suppose someone said, "Prove to me that life is valuable. ~ Rand would hold that his request contains an inconsistency. It is, she holds, the existence and nature of life that sets the conditions for what is valuable; it is the distinctive nature of life that gives rise to the need for values. In saying this, one is saying much more than simply that man must be alive in order to pursue values: one is saying that man must pursue values in order to remain alive- and t.hat this is the base of Copyrighted material 254 ~ ETHICS AND POLITICS ethics and of all questions of moral value. j ust as (Rand would say) it is only the concept of life that gives rise to such concepts as health and disease- just as it would be meaningless to talk of healrh and dis ease except with reference to the standard and goal of life, and talk about health is meaningful only in that context-so it is meaningless ro speak of values, of good and evil, except with reference to the
needs of a living organism. The concept of value, Rand maintains, is geneti cally and epistemologically dependent on the concept of Life, just as the concept of health and disease are genetically and epistemologically de pendent upon the concept of life. Thus to say "Prove that it is morally obligatory to value li fe" is similar to saying "Prove that it is medically obligatory {that is, necessary for heahh) to \alue life." Here we have the claim that apart from a background where life is as sumed as a value, no content can be given to "should" statements and if life is assumed as a value, then content is given to "should," namely (roughly) one should do those things which maintain and enhance one's Life. This account is of the form: you should do an act if ir leads ro the greatest realization of X (where X is the greatest value). Even if one were to accept th is form of account, 4 why must we substitute some thing about life for X? Cannot content be given to shouldstatements, by substituting "death" for X, or "the greatest happiness of the greatest number," or any one of a vast number of orher dimensions or possi ble goals? And wouldn't such conrenr enable "should" Statements ro
guide choices, apply only ro living beings, ere.? Given this, it is puz zling why it is claimed that O>ti)' aga inst a background in which li fe is (assumed to be) a value, can "should" Statements be given a sense. It might of course, be argued that only against this background can "should" statements be given their correct sense, but we have seeo no argument for this claim. Others who find a goal-directed type of ac coui11 of "should" iUuminating, but who do not know what in partie ular to substitute for X, may have "should" witb the content " leads to the greatest value," leaving it as an ope.n question what is rhe greatest value. These people, in asking "Should I do A?" would be asking " Does A lead to the greatest value (whatever that is)?" For these people, and those who substitute for X something other rhan life, an argument is needed to demonstrate that life is the greatest (or a) value. I do not see that such an argument has yet been offered. Copyrighted material On the Randian Argument ~ 255 III Ignoring the difficulties discussed in section II, suppose that we have gotten, somehow, to the conclusion that for each individual, his life and
the prolongation of it is a value for him. How do we get from here to: [III) For each man, the preservation and prolongation of his life, qua man, as a person, is a value for him. Possibilities (1) What man is, qua man, is completely determined by what's special to man [which is rationality). And what's special to beings should be preserved and should flourish. Why? Ignoring the dodo argument [viz., it's a tragedy if any bit of diversity becomes extinct), we focus on the idea that what is special to a thing marks its function and from this we can get itS peculiarly appropriate form of behavior. But no conclusion could depend, in this way, on some property P being special to man. For suppose it turned out that dolphins or some other being somewhere also had property P; would this stop the conclusion? It might be said that here we would discover a new kind of thing (viz. man or dolphin) and P would be special to it. But there might he nothing special to man, for all his nice properties might be had by other things which have further super-duper properties Q also, which man doesn't have. (Call these things, merely for a label, angels.) lt might now be said that
man (or dolphin) does have something special, namely P and not-Q. But how could one think that anything depended on its being special; that something of moral interest did not follow &om Man hasP in conjunction with other premisses, but did follow from the.se premisses conjoined with Man has P, and nothing else has P? Could discoveries on other planets show us that our fundamental moral conclusions don't follow? Surely, it's in virtue of man's having P that conclusions follow, and not because other beings don't have P. All this is not to deny that it is heuristically useful to focus on what is special to man. For if we don't apply moral standards and principles to the other beings we actually have encountered, and we do ro man, then we can ask what properties man has in virtue of which moral principles Copyrighted material 256 ~ ETHICS AND POLITICS apply to him, in vinue of which he is a moral agent and a subject of moral judgments. Something fundamental to ethics does lie in those propenies which do distinguish man from other things we have actually encountered, but nothing morally fundamental depends on the fact that
these properties are distinguishing ones. (2) What man is, qua man, is determined by his essence, which is rationality. He should act so as to continue his essence. (a) If the essence mentioned in this argument is real essence, it's a dubious theory. Also, it (e.g., a man) would no longer exist if its essence changed (but another thing would), so that if its continued existence is a value, so is the continued existence of irs essence. However, in the case of this argument, its essence doesn't cha11ge, it's just not exercised. The being continues to exist. And the conclusion of II is too frail a reed to hang such an argument on. For if it stops existing, another kind of value pursuer, intimately connected with it, will exist. And why should it care which does? (b) If, in this argument, essence is those properties, relative to current knowledge, which underlie, systematize, account for etc. the rest of our general knowledge of such entities (those propenies mentioned in our most fundamental, at the time, reducing unreduced theory about the enriries, or if rwo different such theories, both sets of propenies), then the injunction to preserve and exercise such propenies requires reasons. Why shouldn't one change? Surely, it is not merely a conservative injunction.
Two answers suggest themselves. One, it's change to a lower form of life, and so shouldn't be done. But one needs a theory to back this up, which isn't given, and secondly it implies that it would be all right, if possible, to change to a higher essence. Even though if one could and did change to an angel-like being, one wouldn't be preserving one's life qua ma11. The second, and more interesting answer, is that one won't survive (period), as alive, if one doesn't live as a person. This deserves consideration as a separate argument for lll. (3) If a person does not prolong his life as a rational being, be will not survive for long afterwards as any kind of living being. For, the argument runs, we have no automatic way of knowing what will prolong our lives. We have to figure this out, using our conceptual apparatus. And then we have to devise ways to do this. Otherwise, we will not survive at all, unless by accident, or unless some other rational being cares for us. And, if the goal is to prolong one's life (in years) and minimize, constantly, the probability of its ending, then, much knowledge, invention, ere. will be useful. Copyrighted material On the Rand ian Argument ~ 257 There remains the possibility of being a " ward," making one decision
to let another care for you, give you orders, etc. To this it might be objected that this might not work out (and you wouldn't know when to desert the wardship before it begins to sink), and that such a life is parasitic and depends upon another person's not living as you do. There arc two forms tO the parasite argument, a consequential one and a formal one. The consequential argument is that being a parasite won't work in the long run. Parasites will eventually run out of hosts, out of those to live off, imitate, steal from. (The novel, Atlas Shrugged, argues this view.) But in the short rutl, one can be a parasite and survive; even over a whole lifetime and many generations. And new hosts come along. So, if one is in a position to survive as a parasite, what reasons have been offered against it? The formal argument is difficult to make precise, and difficult to fit into a Randian view. It holds that moral rules are applicable to everyone, so that if following certain rules and values can work only if others follow different ones, and can work only because others follow different ones, then the rules and values in question cannot be the correct ones. ("What if everybody did that?") But it is difficult to find the appropriate level to speak at, using
this argument. My being a teacher succeeds only because other people do other things, e.g., grow food, make clothing. Similarly for the activities of each of these others. The question "What if everyone did that?" shouldn't apply here, but how is the case to be marked off? Intuitively, there is some description of what I'm doing (accepting a job to fill need, etc.) which is okay; that is, it is permissible for everyone to do that. So, there being some description D'l which fits what I do, where it would be disastrous if everyone did D., doesn't show I shouldn't do it. For there may be another description D2 which also fits what I do, and it would be all right if everyone did things of sort Dz. But we should not convert this fact into a condition which says that if there is some description which fits my activity which is such that it's okay if everyone instantiates that description, then my activity is permissible. For this is too weak a sufficient condition for moral permissibility. Presumably some general descriptions of this sort would also fit the thief's activity, e.g., "doing things which lead to their being fed ." Here it might be suggested that the appropriate description is that which presentS the essence of an action. But even if such a
notion could Copyrighted material 258 -& ETHICS AND POLITICS be clarified, as in (2b) above, its application here would beg the question, for especially important among t.be relevant general facts to be accounted for by the essence will be moral facts. Hence we must ~t agree about these moral facts before agreeing about the essence of an act, so that considerations about the essence of aets cannot come first as a basis on which to ground their moral nature. My purpose here is not to pursue the details of how such generalization arguments might best be stated (there is a growing literature on the subject), but to note that if such arguments can be made to work, they will involve extra principles (and not merely the claim that a particular case of parasitism is doomed to fail). Some philosophers view providing the foundation for such principles, and stating them precisely, as one of the central tasks of moral philosophy, but it is one that bas not even been attempted by Miss Rand. One final way to get a formal principle should be mentioned. It might be said that a rational person follows principles, general policies, and so we must consider those principles of action which make man's survival
possible. But it has not been shown why each person must follow the same principles, and why I may not, as a rational being, have a clause in mine which recommends parasitism under certain conditions. IV Supposing that it is granted that living as a rational being is, for each person, a value, how do we get to some social conclusion about people's rights? The basic social principle of the objectivist ethics is that just as life is a.n end in itself, so each living human being is an end in himself, not the means to the ends or the welfare of others-and, therefore, that man must live for his own sake, neither sacrincing himseU to others nor sacrificing others to himself. To live for his own sake means that the achievement of his own happiness is man's highest moral purpose.s Starting with: (1) For each person, the living and prolongation of his own life is a value for him to get to (2) No person should sacrifice bis life for another we would need, in addition to (1) Copyrighted material On the Raodian Argument ~ 259 ( 1') For each person, the living and prolongation of b.is own life (as a rational being) is the greatest value for him.
(And: Each person ought to pursue his greatest value.] No argument has been offered, yet, for (I'). Miss Rand has some things to say about life being an ultimate value, which might lead her to accept ( l '), but these things, in view of our earlier discussion in section II of life as a value, aren't sufficiently dear and forceful to establish (1') .6 But suppose we have (I ') and have gotten to (2). How do we go on to argue for the important social conclusion: (3) No person should sacrifice another person ('s life) to himself (his own). Why shouldn't he? There is the parasite argument already considered: sacrificing another's life to your own is not in your own long-run interests. But tb.is is no argument to convince (or apply co) someone living at a time before the victims have run out, e.g., the present. A more promis ing approach focuses on the notion of rights. Consider the following argument. (4) Each person has a right to his own life, i.e., to be free ro rake the actions required by the nature of a rational being for the support, the furtherance, the fulfillment, and the enjoyment of his life. (5) Since each person has this right, to force a person to sacrifice his interests to your own violates this right. (6) One should not violate another's rights. Therefore
(7) One should not force another ro sacrifice his interests for your own or that of yer another person. But why does each person have a right to his life, tO be free to perform those actions? If we grant, for the purposes of this argument, (1') above, and we add (8) Each person has a right to be free to pursue his greatest and highest value, then we get (4) above. But (8) is surely too strong: did Hitler have a right to pursue his highest value? (But, it will be said, b.is highest value wasn't a rational value. So let us focus instead on "should.") Perhaps instead the argument for ( 4) is as follows: Copyrighted material 260 -<> ETHICS AND POLITICS (9a) For each person, he should pursue the maintenance of his life as a rational being. (9b) The "should" in (a) is and should be of more weight, for him, than all other "shoulds." (lOa) Each person bas a right to do what he should do. (lOb) Each person has a right, which is non-overrideable, tO do what he should (where this "should" has the most weight) do. (9) is meant to follow from other considerations which we have ex amined in section II, and found inconclusive. What of (10)? The force of "right" here is that others shouldn't intervene, using force, to prevent one from exercising it. The que.stion is: why,
given that you should do something, shouldn't I intervene to stop you?7 Perhaps one has a vision of a morally harmonious universe in which there are no irreconcilable conflictS of duty, of shoulds, and in which if you should do something, I shouldn't forcibly prevent you from doing it. But no conclusive arguments have been offered for such a vision. If one believes that ethics involves (something like) one dimension or weighted set of dimensions which is to be used to judge us and the world, so that all of our moral activity (the moral activity of each of us) is directed toward improving (maximizing) the world's score on this dimension, then it will be natural to fall into such a vision. But if we legitimately have separate goals, and there are independent sources of moral commitment, then there is the possibility of an objective conflict of shoulds. So that perhaps, with some slight modification of Sophocles' characters, Antigone should bury her brother and Creon should forbid and prevent this burial. 8 But Miss Rand needs something even stronger, for her argument, than objective harmony of "shoulds"; she needs an objective harmony of interests. 9
What I shall call the optimistic tradition holds that there are no objec rive conflicts of interest among persons. Plato, in the Republic, being the most notable early exponent of this view, we might appropriately call it the Platonic tradition in ethics. l0 Miss Rand falls in this optimistic or Platonic tradition in ethics, believing that there are no objective con llicts of interest among persons, and that there is no situation in which it is in one person's interests to force another to do something which it is not in his interests to do; no situation where one person's forcibly stop ping another from advancing his interests, advances his own interests. No knock-down argument has been offered for this thesis, 11 and Miss Copyrighted material On the Randian Argument ~ 261 Rand has not produced (or tried to produce) any reason why, if such conflicts are possible, I, in following my interests, should limit myself so as nor to interfere forcibly in your pursuit of your life as a rational being. Miss Rand's position is a constrained egoism; egoism subject to the constraint of not violating certain conditions (which are other persons' rights).12 One way to argue for constrained egoism is to argue
that it is identical with egoism, ro argue that the consrraims add nothing because they are in your rightly understood interests (the optimistic tradition). If such arguments equating the egoism and the moral constraints fail, then a holder of this position wiU have ro choose. v We have until now considered only one parr of the social nonsacrific.e principle (don't sacrifice another to yourself) and found the arguments for it inconclusive. We rum now to the other part of the principle: don't sacrifice yourself to another: live for your own sake. "To live for his own sake means that the achievement of his own happiness is man's highest moral purpose." Is it? We have action, endorsed by Miss Rand, in the novel Atlas Shrugged, which appears incompatible with this. In the novel, John Galt risks his life to save that of Dagny Taggart, whom he loves, and he says that he will kill himself if she is tortured to make him talk. How can he do this? He says to Dagny Taggart: "It won't be an act of selfsacrifice. I do not care to live on their terms. I do not care to obey them and I do not care to see you enduring a drawn-our murder. There will he no values for me to seek after that-and I do not care to exist
without values." But this is quite incredible. For it seems from this that, were Dagny Taggart tragically to suffer and die of some disease, then Galt would commit suicide. It would be a terrible loss, but does Galt "the perfect man," have so little moral fiber and resources that life would he intolerable for him wer afterwards (and would the agony of the time immediately after her death outweigh the life which could be led after time bas done its work)? Would he come from Galt's Gulch to attempt to save her life, if he had the option of staying there, obeying no one else, and not being present to observe her murder? Would Galt save Dagny Taggart's life, knowing that this will be at the cost of his own? Is this inconsistent with his principles? Would doing this mean he wasn't an egoist? A fruitless path to fo llow13 is to say Copyrighted material 262 -<> ETHICS AND POLl'fiCS that it is the short period of happiness while he is sacrificing his life that justifies the whole thing. So that, in thinking about a situation in which both are unconscious and only one can be saved by a third party, he would prefer it were himself because in that situation he wouldn't get to
feel the happiness of saving her life? Or are we to imagine that in this thinking about this situation he does prefer that her life be saved, and this because of the happiness be now gets in thinking about it? But (a) we can imagine he's answering, quickly, a large number of questions on some psychological test, so that there's no time to pause for a glow of happiness. Is it now for the happiness he knows he'll feel after the rest is over (mightn't he forget and not recall the question?) that he answers as be docs? {b) Why doesn't he feel sad in contemplating his dying and her surviving? To be sure, be mar think it's better than their both dying, or her dying and his surviving, bur why should the existence of some worse alternatives than alternative A make one happy in contemplating A? In fact, it doesn't, and we don't have this easy path to happiness. (c) Most importantly, this answer gets things backwards. For one would be made happy by placing one's family in the only places in the raft14 only because one values their survival above one's own, their happiness above one's own. It is not that one (chooses ro) have these values, because knowing one has fulfilled them will make one happy. And it is not that I do it because not doing it will make me afterwards feel guilty. For this would be so
only if there were some other moral reason to save their lives over one's own; it cannot be that the primary reason is to avoid later guilt. And also, we could imagine cases where the knowledge is expunged via a chemical producing selective forgetting. Such science fiction possibi lities cause difficulties. If one were concerned only with one's cb.ild's happiness, and one bad the capability, one would implant a device to get the child to act on principles P (the correct moral principles) except in situations S (where he knows that deviating from them will be in his interests, e.g., by murdering someone and taking his fortune) where he will deviate from them, afterwards forgetting that he's done so. Such a person would be happier tb.an one only following principles P, and his life will be identical to one with only principles P, except at a few selected rimes. Furthermore, he will think he always, with great integrity follows principles P, and he will have great self-esreem. And if someone were concerned only with his own happiness, he would wish that he himself had been so preprogrammed. Copyrighted material On the Randian Argument ~ 263 If one doesn't wish this for oneself, then one isn't concerned only with one's own happiness. Saying that one is so solely concerned but
such a preprogrammed person can't be happy because he fakes reality begs the question (ignoring the fact that he doesn't fake it; it's faked for him); it seems as though one can describe a case where " rationality" (and rational self-awareness) and happiness diverge; where someone less rationally self-aware will be more happy. If in contemplating this case you would choose rational self-awareness and moral rectitude, rather than happiness, then the former have independent value and are not justified in your eye.s Oflly because they lead to happiness. Driving the point further, suppose we read the biography of a man who felt happy, took pride in his work, family life, etc. But we also read that his children, secretly, despised him; his wife, secretly, scorned him having innumerable affairs; h.is work was a subject of ridicule among all others, who kept their opinion from him; every source of satisfaction in this man's life was built upon a falsehood, a deception. Do you, in reading about this man's life, think: "what a wottderful life. I wish I, or my children, could lead it"? And don't say that you wouldn't want to lead the life because all the deceptions and falsehoods might come out making the man unhappy. They didn't. Of course, it is
difficult to imagine the others behaving appropriately, and the person himself not being nagged by doubts. But is this the ground of one's reaction? Was it a good life? Does it lack happittess? This man lived a lie, though nor one that he rold. We can imagine other cases. You have what you believe is a private relationship with someone. However, unbeknowst to you, I am filming it with my superduper camera and sound equipment, and distributing the film to people whom you will never encounter. Nothing in your life is changed by the fact that people are packing the pornographic movie rheaters in Outer Mongolia to keep up with the latest serial installment in your life. So, should anyone care? And is tbe only ground on which my action can be criticized the nature of rhe viewers' experience?15 I have listed all these examples, not only to bring the reader to feel the inadequacy of Miss Rand's view, 16 but for another purpose as well. For the examples we have discussed count againsr a more general view as well, which we might call experiential ethics. Put briefly, experiential ethical theories hold that the only facts relevant to moral assessments of actions are bow these actions do, or are intended to, affect the experiences
of various persons. The only morally relevant information (though other information may be relevant via being l!llidence for this Copyrighted material 264 -t> ETHICS AND POLITICS kind) is that about the distribution of experiences in sociery. Theories will differ about which experiences they pick out, or about the criterion of optimal distribution of experiences, but they will agree that all of the considerations have to do with such experiences, and how they feel from the inside. Indeed, it may seem, how could anything else matter other than the experiences people have, how things feel from the inside. What else could there be that's of any importance? It is a task of some intere.st to explain what else does matter, and why, and to account for the pull of the experiential picture on us. 17 Let me, in closing, reiterate that my purpose has been to examine Miss Rand's arguments for her conclusions. It has not been to argue that death is a value, or that we should sacrifice others to ourselves, or that people don't have rights ro our noninterference in their lives, or to demean the virtues of rationaliry, honesty, integrity, productiveness, pride, independence, justice. It has been to see whether, in her published work, Miss Rand indeed objectively establishes her conclusions.
She doesn 't. Copyrighted material 12 WEIGHTED VOTING AND "ONE-MAN, ONE-VOTE" Except for isolated cases of election fraud, and the systematic exclusion from voting of some group, eacb eligible voter can, if be wishes, cast one vote. We arc here not concerned with the number of votes someone casts but with how much weight or power his vote has. Ignoring the considerations which the Supreme Court says might justify deviations from equal weight (e.g., maintaining political subdivisions as the basis for electoral districts, achieving compact districts of contiguous territory) and ignoring some practical difficulties, one wants to equalize, 1 given appropriate measures of power, some function of the power of the legislator and of the power of the person to choose the legislator. If each person eligible to participate in choosing a given legislator has equal power in determining that choice, then one wants to equalize,2 for each i , Power of legislator L1 Number of persons eligible to participate in choosing legislator L1 If, furthermore, each legislatOr has equal power, then one wants
to equalize the number of persons eligible to participate in the choosing of each legislator. Copyrighted material 266 ~ ETHICS AND POLITICS I shall assume that the representative systems under discussion are based upon a system of legislative districts. Professor Paul Freund has mentioned the possibility of a representative system without legislative districts, in which there are 11 representatives, persons' names are arranged in alphabetical order, and the first nth of the people are eligible to vote for one representative, the second nth for another, and so on. Such a system can easily bring about the equalization discussed above, but since no one seems prone to sugge.st such a system, this shows (and, I believe, was intended by Professor Freund to show) that we are concerned that our legislative system satisfy conditions in addition to equalizing the weight of the voters' votes. (And perhaps we would countenance some deviation from the latter condition in order better to satisfy the former ones.) It is of some importance for a theory of representation to stare what purposes ate served by representation by geographical dimicts.3
If in each legislative district, each eligible voter has equal power in selecting the legislator, then in the legislature, one wants each legislator to have power proportional to the number of eligible voters in his dis trier. 4 But it also seems that there are persons other than eligible voters residing in his district whom the legislator is to represent, whose interests he is to care for; for example, recent immigrants who are not citizens, and also nonadult citizens. And we want a legislator's power to be proportional to the number of persons in his district whom he is to represent (in this wider sense). It follows (assuming that in each legislator's district there will be persons whom he is to represent who are not eligible voters) that a legislator's power should be proportional to the number of persons in a set which has as a proper subset the set of eligible voters in his district. Thus it seems that there are rwo conditions to be satisfied: (1) A legislator's power is to be proportional to the number of eligible voters in his district; (21 A legislator's power is to be proportional to the number of persons in his district whom be is to represent. It is possible to satisfy both of these conditions together only if there is some positive number n (which need not be an integer) such that, for each legislator,
the number of persons he is tO represent = n x the number of voters eligible to select him. And it seems very unlikely that this will be the case. Having mentioned this difficulty, I shall ignore it in the remainder of my comments. The standard proposal about weighted voting in a legislature, and the one discussed by Professors Riker and Shapley, gives to each legislator a number of votes proportional to the number of eligible voters in his Copyrighted material Weighted Voting and ''One-Man, One-Vote" ~ 267 district. If we assume that each legislator will on each issue cast his votes as a bloc,S then it can be shown, as Professor Riker and Shapley point out, that such a weighted voting scheme does not guarantee equalization of Power of the ith _legislator Number of persons eligible ro vote for the ith legislat.or where the power of the legislator is measured by the Shapley procedure. (Henceforth I shall refer to this fraction as PI/ N1.) If one supposes that Shapley's procedure gives a reasonable measure of the power of a legislator,6 this result is unfortunate, since it looked like such a weighted voting scheme would both perfectly satisfy the
Supreme Court decisions7 and avoid tricky redistricting problems. Professors Riker and Shapley go on to investigate the questions of whether, in large legislature.s, the deviations from equalizing P;/ N; will be great, greater than is compatible with court decisions, and investigate in what direction one can expect the deviations to go, given the size of a district and of the other districts. What I should like to do is raise some questions about whether, assuming t.hat the districts are satisfactory and that there aren't gerrymandering-rype objections to them, 8 some system of weighted voting will equalize P;/ N; . Given the result which Professors Riker and Shapley mention, a system of weighted voting which equalized P;/ N; would have to be a system which did not give legislators a number of votes proportional to the number of eligible voters in their districts. The most general question is: For every legislature with any number (> 1) of legislators, such that each legislator may have associated with him (as eligible voters) any number of persons, is there some assignment of numbers of votes to each legislator (these numbers of votes need not be integers, and need not be proportional to the number of
persons associated with the legislator) such that Pt! N; is equalized? The answer to this question is "no." For example, 1. Suppose there are two legislators in the legislature who are associated with differem numbers of persons. If the legislators are given t.he same number of votes, their power will be equal and hence their fractions will be different. If the number of votes the legislators have differs and it is a system of majority rule, then the one with more votes will have all the power (and his fraction will be positive); and the one with Copyrighted material 268 -& ETHICS AND POLITICS fewer votes will have no power (and hence his fraction will be zero). Hence the fractions will di ffer. 2. Suppose the legislature consists of an odd number of legislators having the same number of persons associated with them, and one legislator wirh fewer persons associated with him. Each of the odd number of legislators must get the same number of votes (or something equivalent). If the remaining legislator gets th.is number of votes, he has the same power as they do, and the fractions are unequal (his is greater). (Similarly his fraction is at least as great if he gets more votes than they do.) If he gers fewer voces than they do, he has no power,
and his fraction is zero (while theirs is greater than zero). Hence the fractions cannot be equalized.9 The Shapley power value will always be a fraction whose denominator is x, the total number of possible permutations of the legislators' (bloc) votes, 10 and whose numerator will be some integer less than x. Since it will usually be the case that a legislator does not have associated with him exactly some integral xth of the total population, ll it will often be impossible to have a legislator's power be that fraction of the coral power which the persons associated with him are of the total number of persons associated with all of the legislators. Whatever fraction of the total population is associated with a legislator, this fraction will fall between nfx and (n + 1)/x for some integral n > 0 (where x is as before). Let us say that n/.< and (11 + 1)/x surround the fraction. The first questions we want to raise are: 1. What are the conditions under which it is possible, by a system of weighted voting (where the number of vore.s a legislator has need not be proportional to the number of persons associated with him), to get, for each legislator j, Pj/ :[:1 P1 equal to one of the fractions which surround the fraction whose denominator is the total
number of persons associated with all the legislators, and whose numerator is the number of persons associated with the jth legislator.12 That is, under what conditions is it possible to get, for each legislator j, Pj/ Lt P; equal to one of the fractions which surround Nj/ :[:1 N1 where N1 is the number of persons associated wirh the ith legislator. 2. Do current legislative districts (and the ones which would result from foreseeable population change.s) satisfy these conditions? If current legislative districts do satisfy these conditions, then some system of weighted voting, which does not assign votes to legislators in proportion to the number of their constitutents, will serve the purposes that the standard system of weighted voting was designed to serve. Copyrighted material Weighted Voting and " One-Ma n, One-Vote" -& 269 But suppose that current legislative districts do not satisfy these conditions. Must one redistrict in order to get, for each j, Pj/ L; P1 to one of the fractions which surround N i I L; N;? There are various alternative weighted voting schemes which may enable one to achieve this goal without redistricting. It is wonh mentioning some such alternative schemes, since some of them would be rejected even if they achieve this goal, and hence raise questions about
what funher considerations one wishes an electoral scheme ro satisfy. In standard legislatures, all legislators, if they wish to do so, may vote on each bill which comes before the house of the legislature of which they are members (though the) may not be able ro vote in legislative committees of which they are not members). An alternative scheme might assign to each legislator some probability (not necessarily the same for each legislator) of being allowed or entitled to vote on any given issue, with random devices being used to determine if he actually does get to vote on the issue. Once the different probabilities of voting are assigned to each legislator, a generalization of the standard power measures can be used to compute the power of each legislator. This system might be combined with one of weighted voting whereby legislators who get to vote (after the random device determines that they may) may be entitled to cast different numbers of votes. This scheme raises the second question: For any number, > 1, of legislators, each having any 6nite number of constituents, is there some assignment of weighted vote.s and probabilities of getting to vote to the legislators so that, for each j, PJI L;1 P; is
equal to one of the fractions which surround N1J L;1 N1? lf there is not always such an as.signment, what conditions must be satisfied for there to be one, and do current legislative districts satisfy these conditions? Even if such a scheme could achieve the stated goal, many would 6nd it objectionable. (For, it would be said, it might turn out that some legislators would almost never get to vote, and some might not get to vote on some bills of great importance to their districts.) Amost all would find objectionable a system wherein only one legislator gets tO vote on any given bill (thereby determining whether or not it passes during that ses.sion of the legislature), and each legislator has some probability, for any given bill, of being the one who decides. Even if, given current legislature districts, there is a probability assignment which achieves the stated goal, such a system would be rejected. It is worth mentioning such legislative systems, since they make clear that we are nor concerned 011/y with P;/ N1 and suggest the task of statiog the specific additional Copyrighted material 270 -& ET HICS AND POLITICS conditions which we wish to impose upon an electoral and legislative system. Alternative systems could be proposed which might achieve the
stated goal. For example, legislators might be divided into blocs whose votes are weighted. Each bloc must vote as a bloc, and how a bloc votes is determined by how its members, with weighted vores, vote in a caucus of the bloc. So the power of each legislator is a function of the power of the bloc of which he is a member, and of his power within the bloc. (A more complicated scheme would have some legislators being members of more than one bloc.) Given current legislative districts, can such a system achieve the stared goal? I should briefly mention one qualm about Riker and Shapley's inter esting result, about the model of the delegate representatives who "act simply as funnels for their constituencies," that in genera l the citizen of a small district has a much smaller power index than the citizen of a large district. It seems that the authors assume that each legislative district is (roughly) a random sample of the total population with respect to political belief; hence their claim that smaller districts will deviate more than larger ones from rhe national percentage on an issue. The large (in population) districts which we now have are urban districts, and it is not obvious that rhe political beliefs found in these
districts are a random sample of rhe political beliefs of the roral population, or more representative of the political beliefs of the total population than are those of rhe small districts. (It is a commonplace that large urban districts generally elect more "liberal" representatives than do other distr icts.) This is especially clear for counties containing some large urban districts, though most of their population is in small (in population) agricultural districts. The large urban district is not, I think, more likely (than its population justifies) to be in a pivot position. Thus, though Riker and Shapley's result about large and small disrricts follows from considerations about probability and random samples, one would hesitate to apply it to actual political situations where the large urban areas are known nor to be a random sample of the total population with respect ro political beliefs. lr might be said that the structure of the legislative system favors large (in population) districts. Bur even if we grant that facts about the distribution of political beliefs are not part of rhe structure of a legislative system, 13 we may not worry abour a bias the structure imposes since it is counterbalanced by general
conditions likely to continue, and indeed we may desire a biased structure to compensate for certain nonstrucrural biase.s. Copyrighted material DISCU SSIONS AND REVIEWS C0pyrighted material Copyrighted material 13 GOODMAN, NELSON, ON MERIT, AESTHETIC An acqu..1intance who claims to be a medium has told me of recei,,ing a message from Isaac Newton. This message is of such interest that, despite my disbelief in the cla ims of mediumshjp, I transmit it, as I heard it, to the philosophical public. This is Isaac Newton. The time has arrived for me to make public a fact heretofore unknown outside the immediate circle of my friends. What has since come to be known as Newtonian theory was first presented, not in the Principia, but in a poem. Those to whom I showed the poem averred that, though it contained natural philosophy of great interest, the poetry encasing it was so abominable as to make me a laughing stock should I publish it. Reluctantly I allowed myself to become convinced. I should not have been overwhelmed by the unanimity of their
negative judgment of my poem, I now realize. For word bas reached me of the view of Nelson Goodman, who holds that a work has aesthetic merit if it is an aesthetic object and if it significantly changes the way we view the world or conceptually organize the world, performing various cognitive functions. [TRANSMITTER'S NOTE: Here Newton must have in mind Nelson Goodman's discussion in sec. VI of the last chapter of Languages of Art (1968), especially pp. 258- 259: Symbolization, then, is to be judged fundamentally by how well it serves the cognitive purpose: by the delicacy of its discriminations and the aptness of its allusionsi by the way it works in grasping, exploring, and informing the world; by how it analyzes, sorts, orders, and Copyrighted material 274 -& DISCUSSIONS AND REVIEWS organizes, by how it participates in the making, manipulation, retention, and transformation of knowledge. Considerations of simplicity and subtlety, power and precision, scope and sele.ctivity, familiarity and freshness, are all relevant and often contend with one another; their weighting is relative to our interests, our information, and our mqutry. So much for the cognitive efficacy of symbolization in general,
but what of aesthetic excellence in particular? Distinguishing between the aesthetic and the meritorious cuts both ways. If excellence is not required of the aesthetic, neither is the excellence appropriate to aesthetic objecrs confined to them. Rather, the general excellet~ce just sketched becomes aesthetic when exhibit<!d by aesthetic objects; that is, aesthetic merit is such excellence in any symbolic fimctioning that, by its particular comtellation of attributes, qualifies as aesthetic. My italics.] There is no question that my poem was an aesthetic object; my friends never tired of twitting me about how poor a one it was. Furthermore, it certainly would have changed the way people saw the world, had I published it. The Principia, containing the same natural philosophy, did change the way people viewed the world; as did the poem itself for the few friends who saw it. (Unfortunately it also changed the way they viewed my artistic talents.) It follows that, bad l published the poem, it would have been a work of great aesthetic merit. Against. the cavilling criticisms of my friends, l have been vindicated! lo appreciation of his service, I dedicate this reading of my poem to Nelson Goodman: I sing of bodies in attraction, Persistent motion, and retributive action;
Not dramatic plot or moral lecrure But the universe's architecture. The earth and hea,'ens in a single plan Here described by a pious man. Bodies on each other exert a force Directly proportional to the product of their masses and ioversely proportional to the square of rbe distance between them, of course. Gravitation is unselcctivc, Acting berween aU bodies, irrespective. Copyrighted material Goodman, Nelson, on Merit, Aesthetic ~ 275 Before you jump from a tree ro fly free as a bird in uninhibited levitation Think about universal gravitation. Changes from rest or straight-line mo6on Are always caused, such is my no6on. Curved paths or acceleration Stand in need of explanation. Distinguish the constant old from the new To have your theory green instead of gruc. Through absolute space and time, I say, F equals m a. A reaction opposite and equal To every action is a sequel. AJI occurs without magic or porion In accord with the laws of motion. Though to the idea some show resistance Ac6on takes place at a distance. Some wi.ll protest, and rant o'er much, Denying anything they cannot touch. Phenomenalism is the theory of these, Unable ro know beyond their sees. Don't think God's world l here describe
Is ugly, prosaic, or merits a jibe, Don't think it mechanistic, a subject to curse. Instead give thanks He didn't do verse. Alas, there was then no Nelson Goodman to encourage me to publish this poem! Had I done so, my place in the history of literature would have been secure. Such counterfactuals are of meager comfort to me, and might be viewed by some as of dubious intelligibility. But perhaps Goodman's wonderful theory also yields an indicative stat.ement which would serve me. If schoolchildren are taught my poem at a very young age, then my poem will actually change the way they see the world. Copyrighted material 276 -& DISCUSSIONS AND REVIEWS Won't it then actually fulfill its destiny as a major work of art? Perhaps not. Perhaps a poem that merely brings some to see the world as many already do, does not suffice. Must the work of art be the first generally available statement of the view? Because of the accident that my poem did not reali.ze its potential of changing the way in which people view the world, is my chance for aesthetic greatness gone? ln the hope that others may profit from knowing of my missed opportunities, I am sending this message also to Gell-man and Feynman.
P.S.: Kepler here would like to transmit to you a painting he once did, but he lacks the means to do so, and he lacks a notation for painting. He has given me permission to describe the painting to you. It consists of a bright circular spot (the sun) at one focus of purple elliptical dotted lines. He is very pleased to learn that Goodman joins him in considering this painting (which he modestly had entitled "My First Law") one of the major works of art in the western tradition. So much, he says, for his detractors, who had considered it little more than a crudely executed cartoon. P.P.S.: At my next appearance, I will be joined by Fermat, who will sing for you a little ditty whose melody he thinks is rather remarkable. He had wanred, he says, to write it in the margin, but lacked enough room to fit in the bass and treble clefs. This note originated in my remarking ro a friend (not to be confused with tht ac quaintance mcntionrd in tbe o~ning ~ntence) that a doggtul poe:m by Newton pre scnting the theory of u.nive(Sal gravitation would have been, according to Goodman, a \\o"Ork of aesthetic merit. This friend went and wrote fragments of such a poem, reading them to me several days later! I then supplied a frame to go around the poem, bur, de~
spite: my e:ncourage:mc:m, the friend preferred to !eave his poem incomp1ete rather than include it here. This necessitated my writing yet anothur poem, to substitute for his. Ev eryth ing appe--aring here, therefore-, is mine., but I am happy to acknowledge that the idea of actually constmc.ti11g .. Newton's Poem"' (though not the poem that appears here) is the friend's, who prefers to remain nameless, which preference unfortunately was doomed to frustration by his parentS thirry~ne years ago when they named him Saul. Copyrighted material 14 WHOWOULD CHOOSE SOCIALISM? What percentage of people would choose to live under socialism? The communist counrries do not help us to answer this question, for neither through elections nor through emigration do they offer their people any choice. What about the electoral experience of democratic societies such as England or Sweden? Even this does not enable us to disentangle (imagined) self-interest from our topic: the desire to participate in socialist interpersonal relations of equality and community. To find out what percentage of people especially want to live under socialism, we need a situation where people have a reasonably attractive socialist option and also a reasonably attractive nonsocialist one.
If it is not precisely the optimal experiment to answer our question, the Israeli experience with kibbutzim comes as close as the real world can. Not only have the kibbutzim offered socialist personal relations in a socialist community, but these communities have been widely admired for performing the important functions of reclaiming the land, aiding Jewish self-defense, and "normalizing" the occupational structure of the Jewish people. Unlike nineteenth-century utopian communities in America or twentieth-century communes, membership in a kibbutz brought respect and support from the wider society, aiding members through difficulties. Furthermore, no natural population provides a more fertile ground for socialist commitment than the Ashkenazi Jewish population. Copyrighted material 278 ~ DISCUSSIONS AND R.EVIBWS No people is more prone to being captured by an " idealistic" ideological position, especiaUy one emphasizing group solidarity. Indeed, there was selective entry into Israel early in the century; many came in order to help build socialism and did their best to instill their ideals in their children. The kibbutzim are now a very comfortable alternative. By and
large rhey are economically very prosperous. A high level of culrural activity is ava ilable per capita. They have a widely admired educational system (one which, despite their socialist beliefs, kibbutzniks have been somewhat reluctant to share with poorer children in the surrounding areas). They offer occupational opportunities in agriculture, industry, the artS and services, as well as administrative and political tasks in the wider kibbutz movement. No doubt, there are some occupations not easily carried out in a kibbutz setting (for example, experimental particle physicist), though there have been attempts to have university teachers based in a nearby kibbutz. Within Israel, the kibbutz is a prosperous, widely respected and encouraged community (the encouragement includes tax benefits) offering variegated occupational and leisure opportunities. It seems fair to say that the major reason for an Israeli's not living in a kibbutz is that he just doesu ~t want to live that way. The kibbutz population makes up 3.5 percent of the Jewish population of Israel and about 6 percent of the Ashkenazi jewish population (of the kibbutZ population, 85 percent is Ash.kcnazi). So now we know approximately how many people would choose, under highly conducive
conditions, to live under socialism: about six percent. To be sure, the estimate must be refined a bit. No doubt some people outside kibbutzim want to live there but don't because their mares strongly object, just as some reside within kibbutzim because of their mates' strong preference. Some people who wish to leave stay in kjbbutzim beca use there is no vested pension right that they can take with them, while some people who live in moshavim also may be seeking some socialist mode of life (of the Israeli population, 4.5 percent, with equal Ashkenazi and Sephardi components, is in moshavim, some of which combine communal production with familial consumption). Let us be generous and suppose that half aga in as many would choose socia lism. This brings the number up from six to nine percent. Under conditions as ideal as the real world can produce, nine percent of the people would choose to shape their lives in accordance with socialist principles. Copyrighted material Who Would Choose Socialism? -& 279 Surely this minority should evetywhere be aUowed ro do this, joining together with other like-minded people to live according to their desires. (It is a vinue of a free system that it allows minority preferences
to be satisfied, just as a free market carers also to minority tastesfor example, for recordings of Renaissance music or of chassidic songs.) Furthermore, this minority may try to persuade the rest of us of the superior virtues of their ideal. But that's all. They may not force the rest of us to live that way. We can understand, though, why they might be tempted to do so. In that setting most conducive to the free acceptance of socialist ideals, with the most attractive and respected socialist communities and the most receptive population, only nine percent (as a generous e,stimate) would choose to live that way. The prospects, therefore, are dim for intecpersonal socialism's coming anywhere voluntarily. As Israel shows us, there won't be enough volunreers. Copyrighted material 15 WHY DO INTELLECTUALS OPPOSE CAPITALISM? It is surprising that intellectuals oppose capitalism so. Other groups of comparable socio-economic status do not show the same degree of opposition in the same proportions. Statistically, then, intellectuals are an anomaly.
Nor all intellectuals are on the "left." Like other groups, their opinions are spread along a curve. But in their case, the curve is shifted and skewed to the political left. The exact proportion we call amicapitalist depends on how lines are drawn: upon how one construes the anti-capitalist or left position and upon how one marks off the group of intellectuals. What is not in doubt, despite the recent vocal neoconservative faction, is the general leftward tilt of intellectuals. The proportions may have changed somewhat recently but on the average, intellectuals remain further to the left than others of their socio-economic status. Why? By intellectuals, I do nor mean all people of intelligence or of a certain level of education, bur those who, in their vocation, deal with ideas as expressed in words, shaping the word flow others receive. These wordsmiths include poets, novelists, literary critics, newspaper and magazine journalists, and many professors. It does not include those who primarily produce and transmit quantitatively or mathematically formulated information (the numbersmiths) or those working in visual media, Copyrighted material Why Do Intellectuals Oppose Capitalism? -& 281 painters, sculptors, cameramen. Unlike the wordsmiths, people in these occupations do not disproportionately oppose capitalism. The
wordsmiths are concentrated in certain occupational sites: academia, the me dia, government bureaucracy. Wordsmith imellecruals fare well in capitalist society; there they have great freedom to formulate, encounter, and propagate new ideas, to read and discuss them. Their occupational skills are in demand, their income much above average. Why then do they disproportionately oppose capitalism? Indeed, some data suggest that the more properous and successful the intellectual, the more likely he is to oppose capitalism. This opposition to capitalism is mainly "from the left" but not solely so. Years, Eliot, and Pound opposed market society from the right. The opposition of wordsmith intellectuals to capitalism is a fact of social significance. They shape our ideas and images of society; they state tbe policy alternatives bureaucracies consider. From treatises to slogans, they give us the sentences to express ourselves. Their opposition matters, especially in a society (often called "post-industrial") that depends increasingly upon the explicit formulation and dissemination of information. Must we really search for a special explanation of why wordsmith intellectuals disproponionately oppose capitalism? Consider the
straightforward answer: capitalism is bad, unjust, immoral, or inferior, and intellectuals, being inrelligem, realize this and therefore oppose it. This straightforward explanation is nor avai.lable ro those, Like myself, who do not think that capitalism, a system of private property and free markets, is bad, unjust, evil, or immoral. Readers who disagree should notice that even a true belief may not have a straightforward explanation: it might be believed because of some factors other than its truth, such as socialization and enculturation. Something about the manner of many intellectuals' opposition indicates, I think, that something more is involved than their simply realizing the truth about capitalism. For when one or another of the particular complaints about capitalism is refuted (perhaps the one that it leads to monopoly, or pollution, or roo much inequality, or involves exploitation of workers, or despoils the environment, or leads to imperialism, or causes wars, or rhwans meaningful work, or panders to people's desires, or encourages dishonesty in the marketplace, or produces for profit and not for use, or holds back progre.ss to increase profits, or disrupts traditional patterns ro increase profirs, or leads to overproduction, or leads
to underproduction), when the complaint is shown and conceded ro Copyrighted material 282 ~ DISCUSSIONS AND REVIEWS have faulty logic or faulty assumptions of fact or history or economics, the complainer does not then change his mi.nd. He drops the point, and quickly leaps to another. ("But what about child labor, or the racism built into it, or the oppression of women, or urban slums, or in simpler days we could do without planning but now things are so complex that ... , or advertising seducing people into buying things or .. . ) Point after point is given up in discussion. What is not given up, though, is the opposition to capitalism. For the opposition is not based on those points and complaints, and so it does not disappear when they do. There is an underlying animus against capitalism. This animus gives rise to the complaints; the complaints rationalize the animus. After some resistance, a particular complaint may be given up and, without a backward glance, tenfold others wiU surge forward to perform tbe same function: to rationalize and justify the intellectual's animus to capitalism. If the intellectual simply were recognizing the flaws or wrongness of capitalism, we would not find that animus. The explanation of his opposition
will need to be a non-straightforward explanation that also accounts for aru.mus. It may be objecred that the explanation is simply the obvious one that int.elligent people simply have a natural tendency to look around and criticize what is wrong. Or that it is in the nature of creative innovative activity to lead to a skeptical mind that rejects the established order. But why, among the imelligent, is it especially wordsmiths rather than numbersmiths who lean toward the left? If their temperament is critical, why are the wordsmiths usually so uncritical of "progressive" programs? If creative innovative activity is the cause, why must it lead to skepticism rather than to finding of subtle virtues in established beliefs and doctrines? (Did not Dante, Maimonides, and Aquinas engage in creative intellectual activity?) And why must skepticism express itself about the established order rather than about plans for total alternatives that supposedly improve upon it? No, like the idea that capitalism is simply bad and the intellectuals are smart enough to realize this, the explanation that intellectuals are by nature critical and skeptical is unsatisfactory. These "explanations" are too self-serving; they do not fit the details of the situation. We must look elsewhere for an explanation. However, we
should not be surprised that the explanations that spring to their minds turn out to be so self-congratulatory-when explanations are offered it is the intellecruals who offer them. We can distinguish two types of explanation for the relatively higb proportion of intellectua ls in opposition to capitalism. One type finds Copyrighted material Why Do Intellectu als Oppose Capitalism? ~ 283 a factor unique to the anti-capitalist intellectuals. The second type of explanation identifies a factor applying ro all intellectuals, a force propelling them toward anti-capitalist views. Whether it pushes any particular intellectual over into anti-capitalism will depend upon the other forces acting upon him. In the aggregate, though, since it makes anticapitalism more likely for each intellectual, such a factor will produce a larger proportion of ami-capitalist intellectuals. Think about the larger number of people going to the beach on a sunny day. We may not be able to predict whether a given individual will go-that depends upon all the other factors operating on him-but the sun makes it more likely that each person will go and so leads to a larger tOtal number at the beach. Our explanation will be of this second type. We wi ll identify a factor which tilts intellectuals toward anti-capitalist anitudes but
does not guarantee it in any particular case. Previous Theories Various explanations have been proposed for the opposition of intellectuals ro capitalism. One favored by the neo-conservatives focuses on the group interests of intellectuals. 1 Though they do well economically under capitalism, they would do even better, they think, under a socialist society where their power would be greater. In a market society, there is no central concentration of power and if anyone has power or appears to have it, it is the successful entrepreneur and busi nessman. The rewards of material wealth certainly are his. In a socialist society, however, it would be wordsmith intellectuals who staff the government bureaucracies, who suggest its pol icies, formulate them, and oversee their implementation. A socia list society, the imellectuals think, is one in which they would rule-an idea, unsurprisingly, that they find appealing. (Recall Plato's description in the Republic of the besr society as one in which the philosophers rule.) Bur this explanation in terms of the group imerests of intellectuals is unsatisfactory all by itself. Even if a transition to a socialist society were in the group interests of intellectuals (and I leave aside the many illusions this hope embodies), it need not be in the individual
interests o.f any particular intcllt-ctual w aid in the long-term transition. The neo-conservatives make the same mistake the Marxists do in discussing the behavior of capitalists. They ignore the fact that people act not in their group or class interest but in their individual interests. It would be in the individual interest of each intellectual to hold bade while the Copyrighted material 284 ~ DISCUSSIONS AND RE:VIEWS others do the arduous work of bringing about a society more favorable to intellectuals.2 We can formulate a more discriminating explanation, though. If intellectuals think they would do better in a socialist society, and so enjoy reading about the virtues of such a society and the flaws of capitalism, they themselves will constitute a ready and substantial market for such words, and so it will be in the interests of individual intellectuals to produce such word-fare for the other intellectuals to consume. The economist F. A. Hayek has identified another reason intellectuals might favor a social.ist society. That society is thought to be consciously organized in accordance with a plan, i.e., an idea. Ideas are the subject
matter of wordsmiths, and thus a planned society makes primary what they are professionally concerned with. It is a society which embodies ideas. How could intellectuals fail tO find such a society alluring and valuable? To be sure, we can describe the ideas which a capitalist society instances, liberty and individual rights, but these describe a process of freedom, not the final pattern that will result. An ideology that wishes to imprint a pattern upon society will thereby make an idea more central to the society and (unless the idea is repugnant) thereby appeal to the special tastes of intellectuals who deal professionally with ideas. Yet another explanation focuses upon how the motivation for intellectual activity contrasts to the motivations most highly prized and rewarded in market society. Capitalist activity-so the story goesis motivated by selfish greed, pure and simple, while intellectual activity is motivated by the love of ideas. Surely, this contrast is overdrawn. A capitalist can desire to make money in order to support his favorite charity or cause. And entrepreneurial activity can be motivated by its own intrinsic rewards, the rewards of mastery, competence, and accomplishment. True, these activitie.s may also bring extrinsic rewards, but so ma)' a novelist with purely artistic motives earn large
royalties. And is intellectual activity itself always motivated solely by its intrinsic rewards? (Male) writers, it has been said, write tO gain fame and the love of beautiful women. Nor are competitive motivations noticeably absent in the intellectual world. Recall how Newton and Leibniz fought over priority in inventing the calculus, and how Crick and Warson raced Pauling to be first in discovering the strucrure of DNA. Bur though the motivations of economically successful people under capitalism need not be noticeably inferior tO those of intellectuals, it is nonetheless true that in a capitalist sociery economic rewards will tend Copyrighted material Why Do Intellectuals Oppose Capitalism? -& 285 to go to those who meet the demands of others expressed in the marketplace, to the successful and efficient producers of what the consumers want. Intellectuals, too, can satisfy a market demand for their products, as is shown by the high incomes of some novelists and painters. However, it is not necessary that the market reward the intellectually most meritorious work; it will reward what (some porrion of) the public likes. This may be less meritorious work, or it may be wholly nonintellectual work. The market, by its nature, is neutral toward intellectual merit. If intellectual merit is not rewarded most highly, this will be the
fault, if fault it be, not of the market but of the buying public whose tastes and preferences are expressed in the market. If more are willing to pay to see Robert Redford perform than to hear me lecture or read my writings, that is no flaw in the market. The intellectual may most resent the marketplace, though, when be sees an opportunity to succeed financially by producing a work that is less meritorious in his own eyes. Being tempt.ed to debase his own standards for popular acclaim and success-or actually doing thismay cause him to resent the system that lures him to such distasteful motivations and emotions. (Hollywood writers are the paradigmatic e.xample.) Again, though, why does he blame the market system rather than the public? Does he resent a system that maps his road to success through the tastes of the public, a public less acute, knowledgeable, and tasteful than he, a public that is intellectually inferior to him? (Yet most producers in the marketplace know more about their product and its appropriate standards than do most consumers.) Why should intellectuals be so resentful of having to sarisfy market demand if what they want are the fruits of market success? They can always, after all, choose
to adhere to their craft standards and accept more limited e.xrernal rewards. The economist Ludwig von Mises explained opposition to capitalism as resentment on the part of the less successfu1.3 Rather than ascribing their own lack of success, in an open system in which others like themselves succeed, to personal failings, people blame it on tbe nature of the system. However, unsuccessful businessmen, by and large, do not blame the system. And why do intellectuals blame the system rather than their unappreciative fellow citizens? Given the high degree of freedom a capitalist system allows intellectuals and given the comfortable status of intellectuals within that system, what are they blaming the system f()r? How much do they expect from it? Copyrighted material 286 -<> DISCUSSIONS AND REVIEWS The Schooling of Intellectuals lnrellectuals now expect to be the most highly valued people in a society, those with the most prestige and power, those wirb the greatest rewards. Imellecruals feel entitled to this. But, by and large, a capitalist society does not honor its intellectuals. Mise.s expla ins the special resentment of imellectuals, in contrast to workers, by saying they mix socially with successful capitalists and so have them as a salient comparison group and are humiliated by their lesser status. However, even
those intellectuals who do not mix socially are similarly resentful, while merel)' mixing is nor enough- the sports and da ncing instructors who cater to the rich and have affairs with them are not noticeably anticapitalist. Why then do contemporary intellectuals feel entitled ro the highest rewards their society has to offer and resent(~tl when they do not receive this? Intellectuals feel they are the most valuable people, the ones with the highest merit, and that sociery should reward people in accordance with their value and merit. But a capitalist society does nor satisfy the principle of distribution "to each according to his merit or value." Apart from the gifts, inheritances, and gambling winnings that occur in a free society, the market distributes to those who satisfy the perceived market-expressed demands of others, and how much it so distributes depends on how mucb is demanded and how great the alternative supply is. Unsuccessful businessmen and workers do not have the same animus against the capitalist system as do the wordsmith intellectuals. Only the sense of unrecognized superiority, of entitlement betrayed, produces that animus. Why do wordsmith intellecrual.s think they arc most valuable, and
why do they think distribution should be in accordance with value? Note that this latter principle is not a necessary one. Other distributional patterns have been proposed, including equal distribution, distribution according to moral merit, distribution according ro need. Indeed, there need not be any pattern of distribution a society is aiming to achieve, even a society concerned with justice. The justice of a distribution may reside in irs arising from a just process of voluntary exchange of justly acquired property and services. Whatever outcome is produced by that process will be just, but there is no particular pattern the outcome must fit. Why, then, do wordsmiths view themselves as most valuable and accept the principle of distribution in accordance with value? Copyrighted material Why Do Intellectuals Oppose Capitalism? ~ 287 From the beginnings of recorded thought, intellectuals have rold us their activity is most valuable. Plato valued the rational faculty above courage and the apperite.s and deemed that philosophers should rule; Aristotle held that i.ntellectual contemplation was the highest activity. It is not surprising that surviving texts record this high evaluation of intellectual activity. The people who formulated eva luations, wbo wrote
them down with reasons to back tbem up, were intellectuals, after all. They were praising themselves. Those who valued other things more than thinking things through with words, whether hunting or power or uninterrupted sensual pleasure, did not bother to leave enduring written records. Only the intellectual worked out a theory of who was best. What factor produced feelings of superior value on the part of intellectuals? I want to focus on one institution in particular: schools. As book knowledge became increasingly important, schoolingthe education together in classes of young people in reading and book knowledge-spread. Schools became the major institution outside of the family to shape the attitudes of young people, and almost all those who later became intellectuals went through schools. There they were successful. They were judged against others and deemed superior. They were praised and rewarded, the teacher's favorites. How could they fail to see themselves as superior? Daily, they experienced differences in facility with ideas, in quick-wittedness. The schools told them, and showed them, they were better. The schools, too, exhibited and thereby taught tbe principle of reward
in accordance with (intellectual) merit. To the intellectually meritorious went the praise, the teacher's smiles, and the highest grades. In the currency the schools had to offer, the smartest constituted the upper class. Though not part of tbe official curricula, in the schools the intellectuals learned the lessons of their own greater value in comparison with the others, and of how this greater value entided them to greater rewards. The wider market society, however, taught a different lesson. There the greatest rewards did not go to the verbally brightest. There the intellectual skills were not most highly valued. Schooled in the lesson tbat they were most valuable, the most deserving of reward, the most entided tO reward, how could the intellectuals, by and large, fail to resent the capitalist society which deprived them of the just deserts to which their superiority "entitled them? Is it surprising that what the schooled intellectuals felt for capitalist society was a deep and sullen animus that, although clothed with various publicly appropriate reasons, continued even when those particular reasons were shown to be inadequate? Copyrighted material 288 -t> DISCUSSIONS AND REVIEWS In saying that intellectuals feel entitled ro the highest rewards the gen
era! society can offer (wealth, status, ere.), I do not mean that intellectuals hold these rewards to be the highest goods. Perhaps they value more t.he intrinsic rewards of intellectual activity or the esteem of the ages. Nevertheless, they also feel entitled to the highest appreciation from the general society, to the most and best it has to offer, paltry though that may be. I don't mean to emphasize e.specially the rewards that find their way into the intellectuals' pockets or even reach them personally. Identi fying themselves as intellectuals, they can resent the fact that intellectual activity is not most highly valued and rewarded. The intellectual wants the whole society to be a school writ large, to be like the environment where he did so well and was so well appreciated. By incorporating standards of reward that are different from the wider society, the schools guarantee that some will experience downward mobility later. Those at the top of the school's hierarchy will feel entitled to a top position, not only in that micro-society but in the wider one, a society whose system they will resent when it fails to treat them according to their self-prescribed wants and entitlements. The school system thereby produces anti-capitalist feeling among intellectuals. Rather, it produces anti-capitalist feeling among verbal
intellectuals. Why do the numbersmiths not develop the same attitudes as these wordsmiths? I conjecture that these quantitatively bright children, although they get good grades on the relevant examinations, do nor receive the same face-to-face attention and approval from the teachers as do the verbally bright children. It is the verbal skills that bring these personal rewards from the teacher, and apparently it is these rewards that espe cially shape the sense of entitlement. There is a further point to be added. The (future) wordsmith intellectuals are successful within the formal, official social system of the schools, wherein the relevant rewards are distributed by the central authority of the teacher. The schools contain another informal social sys tern within classrooms, hallways, and schoolyards, wherein rewards are distributed not by central direction but spontaneously at the pleasure and whim of schoolmates. Here the intellectua ls do less well. It is not surprising, therefore, that distribution of goods and rewards via a centrally organized distributional mechanism later strikes intellectuals as more appropriate than the "anarchy and chaos" of the marketplace. For distribution in a centrally planned socialist society stands to distribution in a capital ist society as distribution by the teacher stands tO distribution by the schoolyard and hallway.5
Copyrighted material Why Do Intellectuals Oppose Capitalism? ~ 289 Our explanation does not postulate that (future) intellectuals constitute a majority even of the academic upper class of the school. This group may consist mostly of those with substantial (but not overwhelming) bookish skills along with social grace, strong motivation to please, friendliness, winning ways, and an abiliry to play by (and to seem to be following) the rules. Such pupils, too, will be highly regarded and rewarded by the teacher, and they will do extremely well in the wider sociery, as well. (And do well within the informal social system of the school. So they will not e.specially accept the norms of the school's formal system.) Our explanation hypothesizes that (future) intellectuals are disproportionately repre.sented in that portion of the schools' (official) upper class that will experience relative downward mobility. Or, rather, in the group that predicts for itself a declining future. The animus will arise before the move into the wider world and the experience of an actual decline in status, at the point when the clever pupil realizes he (probably) will fare less well in the wider society than in his current school situation. This unintended consequence of the school
system, the anti-capitalist animus of intellectuals, is, of course, reinforced when pupils read or are taught by intellectuals who present those very anticapitalist attitudes. No doubt, some wordsmith intellectuals were cantankerous and questioning pupils and so were disapproved of by their teachers. Did they too learn the lesson that the best should get the highest rewards and think, de.spite their teachers, that they themselves were best and so start with an early resentment against the school system's distribution? Clearly, on this and the other issues discussed here, we need data on the school experience.s of future wordsmith intellectuals to refine and test our hypotheses. Stated as a general point, it is hardly contestable that the norms wirbin schools wiU affect the normative beliefs of people after they leave the schools. The schools, after all, are the major non-familial sociery that children learn to operate in, and hence schooling constitutes their preparation for the larger non-familial society. It is not surprising that those successful by the norms of a school system should resenr a sociery, adhering to different norms, which does not grant them the same success. Nor, when those are the very ones who go on to shape a society's
self-image, its evaluation of itself, is it surprising when the society's verbally responsive portion turns against it. If you were designing a sociery, you would not seek to design it so that the wordsmiths, with all their influence, were schooled into animus against the norms of the society. Copyrighted material 290 -b DISCUSSIONS AND REVIEWS Our explanation of the disproportionate anti-capitalism of intellectuals is based upon a very plausible sociological generalization. In a society where one extrafamilial system or institution, rhe first young people enter, distributes rewards, those who do the very best therein will tend to internalize the norms of this institution and ex peer the wider society to operate in accordance with these norms; they will feel entitled to distributive shares in accordance with these norms or (at least) to a relative position equal to the one these norms would yield. Moreover, those constituting the upper class within the hierarchy of this first extra-familial institution who then experience (or foresee experiencing) movement to a lower relative position in the wider society will, because of their feeling of frustrated entitlement, tend to oppose the wider social system and feel animus toward its norms.
Notice that this is not a deterministic law. Nor all those who experience downward social mobility will rum against the system. Such downward mobility, though, is a factor which tends to produce effects in that direction, and so will show itself in differing proportions at the aggregate level. We might distinguish ways an upper class c.an move down: it can get less than another group or (while no group moves above it) it can tie, failing to get more than those previously deemed lower. It is the first type of downward mobility which especially rankles and outrages; the second type is far more tolerable. Many intellectuals (say they) favor equality while only a small number call for an aristocracy of intellectuals. Our hypothesis speaks of the first type of downward mobility as especially productive of resentment and animus. The school system imparts and rewards only some skills relevant to later success (it is, after all, a specialized institution) so its reward system will differ from that of the wider society. This guarantees that some, in moving to the wider society, will experience downward social mobility and its attendant consequences. Earlier I said that intellectuals want the society to be the schools writ large. Now we see that the resentment due
to a frustrated sense of entitlement stems from the fact that the schools (as a specialized first extra-familial social system) are not the society writ small. Our explanation now seems to predict rhe (disproportiotuJte) resentment of schooled intellectuals against their society whatever irs nature, whether capitalist or communist. (Intellectuals are disproportionately opposed to capitalism as compared with other groups of similar socioeconomic status within capitalist society. It is another question whether Copyrighted material Why Do Intellectuals Oppose Capitalism? ~ 291 they are disproportionately opposed as compared with the degree of opposition of intellectuals in other socieries ro those societies.) Clearly, then, data about the attitudes of intellectua ls within communist countries toward apparatchiks would be relevant; will those intellectuals feel animus toward that system? Our hypothesis needs to be refined so that it does not apply (or apply as strongly) to every society. Must the school systems in every society inevitably produce anti-societal animus in the intellecntals who do nor receive that society's highest rewards? Probably not. A capitalist society is peculiar in that it seems to announce that it is open and responsive
only to talent, individual initiative, personal merit. Growing up in an inherited caste or feudal society creates no expectation that reward will or should be in accordance with personal value. Despite the created expectation, a capitalist society rewards people only insofar as they serve the market-expressed de.sires of others; it rewards in accordance with economic contribution, not in accordance with personal value. However, it comes close enough to rewarding in accordance with value-value and contribution will very often be intermingled- so as to nurture the expectation produced by the schools. The ethos of the wider society is close enough to that of the schools so that the nearness creates resentment. Capitalist societies reward individual accomplishment or announce they do, and so they leave rhe intellectual, who considers himself most accomplished, particularly bitter. Another factor, I thin.k, plays a role. Schools will tend tO produce such anti-capitalist attitudes the more they are attended together by a diversity of people. When almost all of those who will be economically successful are arrending separate schools, the intellectuals will not have acquired that attitude of being superior to them. But even if man)' children of the upper class attend separate schools, an open society will
have other schools that also include many who will become economically succe.ssful as entrepreneurs, and the intelle.cruals later will resentfully remember how superior they were academically to their peers who advanced more richly and powerfully. The openness of the society has another consequence, as well. The pupils, future wordsmiths and others, will not know how they wi ll fare in the furure. They can hope for anything. A society closed to advancement destroys tbosc hopes early. In an open capitalist society, the pupils are not resigned early to limits on their advancement and social mobilit)', the society seems to announce that the most capable and valuable will rise to the very top, their schools have already given the academically most gifted the message that they Copyrighted material 292 -& DISCUSSIONS AND REVIEWS are most valuable and deserving of the greatest rewards, and later these very pupils with the highest encouragement and hopes see others of their peers, whom they know and saw to be less meritorious, rising higher than they themselves, taking the foremost rewards to which they themselves felt themselves entitled. Is it any wonder they bear that society an
animus? We have refined the hypothesis somewhat. It is not simply formal schools but formal schooling in a specified social context that produces anti-capitalist animus in (wordsmith) intellectuals. No doubt, the hypothesis requires further refining. But enough. It is time to turn the hypothesis over to the social scientists, to take it from armchair speculations in the study and give it to those who will immerse themselves in more particular facts and data. We can point, however, to some areas where our hypothesis might yield testable consequences and predictions. First, one might predict that the more meritocratic a country's school system, the more likely its intellectuals are to be on the left. (Consider France.) Second, those intellectuals who were "late bloomers" in school would not have developed the same sense of entitlement to the very highest rewards; therefore, a lower percentage of the latebloomer intellectuals will be anti-capitalist than of the early bloomers. Third, we limited our hypothesis ro rhose societies (unlike Indian caste society) where the successful student plausibly could expect further comparable success in the wider society. In Western society, women have not heretofore plausibly held such expectations, so we would not expect the
female students who constituted part of the academic upper class yet later underwent downward mobility to show the same anticapitalist animus as male intellectuals. We might predict, then, that the more a society is known to move toward equaliry in occupational opportunity between women and men, the more its female intellectua ls will exhibit the same disproportionate anti-capitalism its male intellectuals show. Some readers may doubt this explanation of the ami-capitalism of i.ntellectuals. Be this as it may, I think that an important phenomenon has been identified. The sociological generalization we have stated is intuitively compelling; something like it must be true. Some important effect therefore must be produced in that portion of the school's upper class that experiences downward social mobility, some antagonism to the wider society must get generated. If that effect is not the disproportionate opposition of the intellectuals, then what is it? We started with a puzzling phenomenon in need of an explanation. We have found, Copyrighted material Why Do Intellectuals Oppose Capitat;sml ~ 293 I think, an explanatory factor that (once stated) is so obvious that we must believe it explains some real phenomenon.
Is There a Solution? Those who thin.k capitalist society ought especially to be opposedbut why do they think so?-will welcome th is unintended effect of the school system. Yet, as we have noted, the problem of the disbar mony between the intelligentsia and the norms of the wider society is a more general one. It wiU be faced by any society, whatever its character, whose school system specializes and is not the society writ small. The more important and influential its wordsmith i.nrellectuals (as in "post-industrial societies" ), the more of a problem wiU this be. Thus, all readers can join with me in wondering how this opposition of the intellectuals to the society might be avoided-though some readers might prefer to wonder about this for some non-capitalist society. When the schools and the wider society are out of joint, the two obvious solutions are to restructure (either) one to align it with the other. First, one might try to make the society fit the school norms, either by a socialist structuring which places the intellectuals at the top or through a naturally arising meritocracy. Yet, however important knowledge becomes in the society, no relatively free society will or can reward the b.ighest school skills most highly. The schools, ruith great effort, focus only upon some qualities; these, while they play a significant role in
economic success under some circumstances, will never be the whole stoty of resultant social position. Consumers are not schoolteachers giving grades for test and dassroom performance. Alternatively, and less ambitiously, the schools might be modified so as to fit tb.e norms of the wider society or at least to avoid instilling contrary norms. If the intelligent are entitled to anything the market doesn't give them, it is to recognition as being intelligent-nothing more. They are not entitled to the largest rewards of the wider society. How might this lesson in mode.sty then be taught? Merely to say the economy properly rewards other attributes will not be enough. The children will learn from and internafjze the school's actions, not its words. To be sure, the total social system of the school milieu values maoy things: athletic prowess in the schoolyard, toughness in tb.e hallways, singing talent in the auditorium, good looks everywhere. But the school officially recognizes only or mainly intellectual skills and performance. Since tb.at, after all, is what it is there for, it would be difficult for it credibly to give other Copyrighted material 294 ~ DISCUSSIONS AND REVIEWS attributes parity or very significant recognition. (Character and
attendance awards are a joke, I assume, everywhere.) Another possibi lity is ro diminish the academic hierarchy within the school system. Schools might teach without ranking the students, without grading them according to their success in learning. Reformers occasionally call for the abolition of rests and grading. Paul Goodman argued that these have a function extrinsic to that of the education itself, serving only the needs of future employers or the admissions committees of other schools, who can be left to do their own informational testing themse.lve.s.6 (ft is clear, though, that resting and certification also enlarge the discretionary choice of students. Employers accept a college's statement that a student has completed requirements for a B.A. degree without looking too closely into what those requirements are or how useful those courses are for the purposes of employment.) Yet rests perform other functions as well, intrinsic to the educational process. They tell a student how well he is doing as judged by objective standards, how well he is doing as compared with others in his reference group (how well, after all, should he expect himself to do?). They provide information for tracking when educationally appropriate, as well as for intermediate reinforcement along life's way.
In any case, given the extrinsic informational function, employers will find it advamageous to hire from schools that test and certify and therefore students will find it advantageous to attend such schools. Whatever the general social interest, people will pursue their own individual interests. No one wi ll refuse to hire from a certain school or attend one because thar type of school system produces intellect uals with an anticapitalist animus. While legislation to alter school systems might accomplish the end, its benefits are so remote in comparison with its cost that it is unlikely that such legislation will pass. Nor is such legislation, at least as applied to private schools, compatible with the capita list ethos of liberty and indiv idual rights. 7 Restructuring th.e och.ools so as to de-emphasize intellectual skills and accomplishments raises troublesome issues, apart from the obvious one of the resultant cost in (short-term) efficiency in the society. The cultivation of intellectual abil it ies and talents is, we think, an important value in itself. Yet the school systems we know that cultivate these also, unintentionally, generate an animus against the social system among some of those most gifted intellectually. U the long-term stability of the desirable social system is best served by curbing the cultivation of some
highly Copyrighted material Why Do Intellectuals Oppose Capitalism? ~ 295 admirable and worthwhile traits of individuals, then we face a serious conflict among our values. It will comfort those who support the continuance of capitalist society to remember that this conflict is a general one. Communist society, too, finds intellectuals lie athwart its path, and during the Cultural Revolution the Chinese, at great economic and personal cost, by forcible re-education, exile to the countryside, and personal persecution, tried to make them just like everyone else. That attempt failed. The tension of capitalist society with its intellectuals is far less serious-we may just have to live with it. Whatever happens, though, the intellectuals will have the last word. Copyrighted material 16 THE CHARACTERISTIC FEATURES OF EXTREMISM Geula Cohen has told us what she thinks extremism is not.1 I would like to offer a general description of what extremism is. To be sure, a simple definition of extremism is not really possible. There is, however, a cluster of features, some more central than others, that together
form what might be called an extremist syndrome. I want tO list eight features of extremism. A.n extremist will be someone who has enough of these features-it needn't be each and every one of the eight. The first has to do with the goals and doctrines of extremism and extremists. We don't normally term people extremists simply because rhey seek to exert personal power, even if in order to rule they murder large numbers of people and engage in other horrendous aces. Extremists present doctrines and goals that claim or purport to have a certain objectivity, an impersonal validity. Moreover, along some salient political dimension, the extremist position falls somewhere near the end or fringe of something close to a normal distribution. This need not be a Left-Right dimension: there can be many different salient political dimensions. Goals that were viewed as extreme in one historical period-for instance, abolitionism in the United States in the nineteenth cenrury- later often come to be viewed as somewhere in the center. This particular notion of extremist goals, then, is defined relative to the rest of the distribution, by a position at the end. Copyrighted material The Characteristic Features of Extremism ~ 297 The second feature of extremism, in addition to goals, is the perception
that extremists typically have that somewhere in the world is an enemy who is absolutely evil, with no shred of right or justice on their side. This is coupled with the tendency to think that anyone who does not hold the extreme position that you occupy is somehow connected with the enemy- "anyone who is not with me, is against me." Also, there is a tendency, at least among some extremist groups, to see conspiracies among one's opponents, to regard one's opponents as engaged in conspiratorial poHtics or machinations of one sort or anotheL Flowing quite naturally from this is the third feature of extremism, its non-compromising nature. Here we se.e an unwillingness oo the part of extremists to trade off or diminish the achievement of their goals, to modify their position in order to gain more adherents in tbe public political arena, or to reach agreement with opponents (for if the opponents are evil, as they believe, then no compromise is possible). Here the que.stion arises: How do we distinguish this noncompromising position from a principled one? After all, there are some correct principle.s to which one wants to adhere and not surrender. Let me tentatively suggest some ways to make this distinction. The first i.nvolves willingness to see value in an opponent's position. Is it a conflict of good against good, perhaps even a greater against a lesser good, or
is there nothing to be said for the other side at all, no legitimate arguments or reasons it can present? One diagnostic criterion of extremism, therefore, is to hold no brief whatsoever for an opposing view. Further, a principled position, as opposed to a rigid, non-compromising posture, tends to stay deli.mited and doe.s not expand to fill all tbe available space. The tendency among extremists, on the other hand, is for their principles to spill over, starting with one extremist principle and then picking up another and yet anotheL Hence, even if one has principles and is convinced that they are right, there can be a nonauthoritarian way of maintaining them; one can still be willing to listen to and consider counter-arguments. I have listed three features that mark extremism-goals and doctrines; the view of opponents as evil; and an unwillingness to compromise. Bearing these criteria in mind, we can see how religious fundamentalism. can come to fall into the category of extremism.. For religious fundamentalists there is a divine edict that has been laid down that cannot be modified or compromised. This is nor to say that all religious groups Copyrighted material 298 ~ DISCUSSIONS AND REVIEWS share the panicular attitude. Moderation and compromise, I
believe, are enjoined by the Talmud, and there is, of course, the Golden Mean prescribed by Maimonides. So it is possible to be decidedly orthodox in belief without extremism. The fourth feature of extremists is a willingness to use extreme means for the achievement of one's goals. Which means are extreme in this regard? Most obviously, the use of violence, of overriding the rights of others, either opponents or third parties. The most pronounced manifestation of such extremism is terrorism, which is directed against innocent third parties for the attainment of certain goals. But there also can be the willingne.ss to use extreme means for the achievement of goals that employs violence against one's self- for example, the selfimmolation of Buddhist monks to protest the war in Vietnam. Another salient feature of extremism, the fifth, is that the goals it seeks to realize must be achieved immediately and in full. Delays and gradual evolution are unacceptable. The sixth feature of extremism, the kind we are talking about, is that it does not involve isolated scribblers or lone practitioners. We are dealing here with organized movements that attempt to put their goals inro effect. Seventh, these extremist groupings not only find themselves at the far end of the political spectrum, but-for whatever reasons, perhaps
even psychological-that is where they want consistently to be. The history of extremist groups is replete with examples of the replacement of extreme doctrine by even greater extremism, to surpass the competition. The eighth feature concerns the very special nature of the extremist. To be an extremist is not simply to hold some particular extremist position. Although I am not a psychologist., I suggest that there is a determinate extremist personality. In looking over the terrain we find people often moving easily from one extremist group to another. For example, in Germany there were those who shifted from the Communist Part)' to the Brownshirts without blinking an eyelash. In the United State.s, the Lyndon LaRouche group moved from the fringes of the extreme Left to their present fascist posture, apparently without losing many members along the way. Extremists appear able to shift their loyalties at will; it would seem that the content of the extremist doctrine is not what counts, so much as the importance of remaining at the extreme of some salient dimension. And we can understand why such groups tend to adopt additional extreme doctrines as they go along. Since extremists have come to distrust the conventional wisdom, as they call it, Copyrighted material
The Characteristic Features of Extremism + 299 on the one issue that has seized their attention, they tend to conclude the conventional wisdom must be wrong on a whole lot of other issues as well. An example are those exrremist groups in the United States that start out with doctrines having nothing to do with the Jews and end up incorporating Holocaust revisionist doctrines. There is, apparendy, a camaraderie of extremism. Finally, though many of t.he features of extremism that I have noted can be said to be descriptively neutral, it seems to me that we cannot apply the label "extremist" to people unless we think that there is something normatively disproportionate about thei.r extremist response ro the specific situation. There is something about extremists' beliefseither about the enemy, or the justice of their cause, or the means they are willing to use-that is either false or is not justified on the basis of what are considered to be correct normative criteria. Extremism means going too far, not just according to the descriptive criteria, but the normative criteria as well. Hence disputes over extremism may hinge on arguments over which position is correct or ;ustified. (For rarely is anyone willing to say, "This is the correct position, and it is an extremist position.") ln these preliminary remarks I have attempted to set forth some
of the characteristic features of extremism; perhaps in the course of our discussion we will arrive at a sharper notion than having enough of the characteristic features on the list, perhaps by weighting these features of extremism or treating some of them also as necessary conditions. Copyrighted material 17 WAR, TERRORISM, REPRISALsORA WING SOME MORAL LINES A review- of Michac.l Walur~s just and Unjust War.s (New York' Basi< Books, 1977). Warfare raises, in their most stark form, moral issues about aggression, in itiation of force, rightS of innocent third parties, responsibilicy for following orders, ere., and so should be of special interest ro libertarians. The previous literature focusing on these moral issues of warfare is slight and insubstantial, however, so we should welcome Michael Walzer's sensitive, thoughtful, and illuminating study. Walzer's book is subtitled A Moral Argument with Historical Illustrations, and his method of raking up each of rhe issues as it applies in specific historical situations makes for e.specially interesting reading. I shall selectively discuss the contents of the book, emphasizing those points l find especially illuminating and those with which. a
libertarian is especially likely ro disagree. Walzer, a professor of government at Harvard Universicy and an editor of the democratic socialist magazine Dissent, is a friend of mine, and so this will not be the fi.rst time he hears of the disagreement. I mention the friendship because as a student I used to be bemused when reading book reviews by one friend of another, where no mention was made of the personal tie. So, caveat lector (and friend, roo! ). Copyrighted material War, Terrorism, Reprisals ~ 301 Means and Costs Walzer distinguishes the justice of war (is it defensive, aggressive?) from the legitimacy of means in pursuing it (is attack directed against innocent civilians? etc.). He holds that an ordinary soldier can be held responsible and punished only for using illegitimate means of warfare, even if he is lighting in the army of the aggressive nation that started the war. On this view, when that soldier kills a civilian on the other side by deliberately taking aim and firing, he is a murderer, but when he intenrionally shoots an opposing soldier who is merely defending his nation against foreign attack he is not a murderer_ Walzer holds that the leaders of the nation who decide ro lau nch an aggressive
war Clln be held responsible and punished for this, but ordinary soldiers cannot, provided they do not use unjust means of waging the war. Libertarians will not agree that one is absolved of responsibiliry merely by being told (by the rulers that be) to use legitimate means, whatever the legitimacy of the end to which these are directed. So we face the interesting issue of precisely bow to treat the captured soldiers of an aggressive army. (Are all to be tried for murder or attempted murder?) Walzer correctly points out that the aggressor nation need not be the one who actually attacks first, for its threat of imminent aggressive warfare (not merely of hostile acts short of war) may be sufficient to justify a preemptive attack by the defending nation, which therefore turns out to be the first to lire. I found the third part of the book the most interesting. In approximately 100 pages Walzer discusses with subtlety and good sense noncombatant immuniry, sieges and blockades against places also containing civilians, guerrilla war, terrorism, and reprisals. Noncombatants may not be attacked, but suppose it is known that some will or may be harmed as a result of an attack? The rraditional doctrine of double effect says that one may not directly aim at, intending to inflict death on, civilians; but if one knows that some civilians will be injured as
a sideeffect of one's action (hence the term "double effect"-to distinguish two kinds of effects, directly intended and side), this is not strictly forbidden but counts as a moral cost of the action, which is to be weighed in a calculation against its gains. ln a paper about ten years ago, I argued that the principle for such moral cost-benefit calculations is the following (Essay 10). If A is the Copyrighted material 302 -& DISCUSSIONS AND R"I!VIEWS act being contemplated, it is not enough that the moral benefit of A outweigh the moral cost. For A will be impermissible if there is available an alternative act 8 of lesser moral cost such that the extra benefit of A over 8 is outweighed by the extra cost of A over 8. Walzer nicely suggests adding to the traditional principle of double effect the additional condition that one also must intend to reduce the evil as far as possible, including by running some additional risk to oneself in order to reduce the (not directly intende"d) civilian harm. Walzer 6"ods it bard to say how far you must go in accepting additional risks. I suggest we view this alternative risky act as the B in the principle I just stated; Walzer's emended principle of double effect, then, is simply an instance of the general principle, which principle then
specifies how much additional risk one must run. I don't know if Walzer would approve of dtis, si"nce he tends not to sharpen a point or principle beyond what good sense can specify without formal apparatus. Often, however, it is useful to speculate on what the ideal precise rule would be, for that helps us to uncover all its components and think about bow they fit together. Furthermore, it is unclear to me whether Walzer would agree to view it this way and count (even with discounting) the risk to one's own soldiers as a moral cost (or diminished moral benefit) in the same equation with injury to the noncombatant civilians. For he seems to hold (p. 1 57) that one must run increased risks in order to safeguard noncombatants all the way up to the point "where any further risk-taking would almost certainly doom the military venture or make it so costly that it could nor be repeated." Civilian Innocence Walzer's discussion of sieges and blockades is very fine, and his condemnation of terrorism directed against innocent civilians is surely right. Walzer applies this point to the FLN in Algeria, also condemning me elitism of terrorists, points he made publicly in the late 1960s in Cambridge before intensely disapproving young lefrisrs.
Reprisals as a means of moral enforcement raise interesting issues, for though directed to enforce moral norms another is violating, they need not be done directly to the violator. Walzer discusses and condemns the reprisals by Free French forces who executed 80 captured German soldiers in response to German summary executions of 80 mil itarily dressed captured Free French forces. His discussion of Israeli reprisals against sporadic attacks is judicious and discriminating, and Copyrighted material War, Terrorism, Reprisals ~ 303 he defends the Israeli reprisal in 1968 against property (Lebanese aircraft) in response to PLO attacks (based in Beirut) against an El AI plane in Athens. Not any reprisal goes, however, as he argued also in a recent New Republic discussion of the insufficiently limited and pinpointed Israeli reprisal in Lebanon. But i.f another nation will nor or is not able to control action against another from within its borders, then the attacked party may act in self-defense. I found that Walzer's discussion of guerrilla warfare pays insufficient attention to the way guerrillas exploit the morality of those they attack. If it is important that the noncombatants be immune from attack, it
is important that the combatants be distinguishable when encountered; such is the function of military uniforms. (So perhaps we should not scorn those who delight in wearing military uniforms, for they groove on the garb that demarcates them as legitimate targets.) What then shall we say of those combatants who refuse such garb and who presumably would imitate any special markings civilians adopted to designate themselves? When civilians are harmed, does all responsibility for this lie with those who act upon them, or does some rest oo those who by rheir acions make rhe civilians so vulnerable to being confused wirh combatants? And when guerrillas operate in an area, living in the midst of civilians whose presence is a moral shield for the guerrillas and an extra moral restraint on the means those attacking the guerrillas may use, do the civilians have any responsibility to get out of the way? If someone shoots at passersby from the window of my office and the police arrive to cope with him and, while not being kept a prisoner there, I insist on staying, shouting out, "I'm not violating anyone's rights, and so I will stay here at my desk working; it's your job to make sure not to injure innocent people, including me," then if I do get
hurt as the gunman is overpowered, is none of that my fault? Is it only the police who are to be condemned for being insufficiently discriminating? The questions also are of interest to libenarians who hold that anyone has a perfect right to do anything provided it does not violate the rights of others. Does my staying in m)' office violate anyone's rights? Am I under an obligation not to impede by my presence the apprehension of the gunman? (Must others induce me to get out of the way, or may they demand I do so, with the consequences of my staying re,sting on my head?) And if I get shot in the middle of the confused melee, have my rights been violated by the police; am I owed the same compensation as if I had been held hostage there? Copyrighted material 304 ~ DISCUSSIONS AND REVIEWS Walzer assumes that if a guerrilla movement cannot be dislodged from the population who point them out, then it has the effective democratic support of the population. This rakes insufficiently seriously the ways the guerrillas retaliate against those who turn them in. (The only case I know where this did nor happen is that of the Irgun, led by Menachem Begin, which did nor arrack the Hagganah people who helped the British apprehend them during what was known as the
season.) Undislodgeable is nor the same as popularly supported, given the fear or passivity of the others, as might be shown if ever a guerrilla group that came to power would permit (within Walzer's guidelines) its opposition to wage guerrilla warfare against it. New Thoughts Parr 4 of Walzer's book contains an interesting discussion of the rights of neutral nations; an argument that in supreme emergency of national existence, the moral limits are relaxed and may permissibly be violated (which you will have tO read and ponder); and also an unsatisfactory discussion of nuclear deterrence. I do nor say this as a special criticism, for I am not familiar with (and don't have up my sleeve) a satisfactory moral position on nuclear weapons, on a deterrence defense posture that threatens great destruction of civilian lives. (Must one just renounce such weapons and succumb to the brandisher of them, who is willing to use them to crush one's non-nuclear might and civilian nonviolent re.sistance strategy?) We need some libertarian thinking on these issues. I have rouched only on some of the topics in this rich and stimulating book- those that sprang to mind as I looked at the table of
contents six months after reading the book, when asked to review it here. You will learn a lor from this book and be led to new thoughts (sometimes different from Walzer's). If nor, go ask Walzer for your money back. It's inappropriate, anyway, for an editor of the socialist magazine Disse11t ro receive and keep too many royalties. (Do you hear that, Irving Howe?) Copyrighted material 18 DO ANIMALS HAVE RIGHTS? A review of TIJe Case for Animal Rights. by Tom Rcg;a.n (8erkdey' University of California Press, 1983). Animal rights seems a topic for cranks. Who else would devote major energy to promote changes in our treatment of animals or write a book on the subject? Devotees of ridiculed causes remind us that the crank label also was applied to pioneers who favored the abolition of slavery and the vote for women. Tom Regan begins his book, which argues for animal rights, by quoting john Stuart Mill: "Every great movement must experience three stages: ridicule, discussion, adoption." However, there are many more insignificant movements that experience only the ridicule and perhaps a bit of discussion. No doubt it is exhilarating for
a movement in the first stage to look ahead expectantly to the later ones, but the odds are sobering. The mark of cranks is disproportionateness. It is not merely that they devore great energy to their issue-there is properly a division of labor in improving the world. They view the issue as far more important than it is, more pressing than others that in fact are more significant. In the talk and attention they disproportionately devote to the topic, they weight reasons in support of their position unduly while giving short shrift to opposing considerations. Conversations are turned to the issue, people buttonholed, friends judged by their stands; the issue becomes a touchstone and fulcrum. And cranks about one issue tend to become interested in other cranky causes and issues. Perhaps these different Copyrighted material 306 -t> DISCUSSIONS AND I!.BVIEWS fixations perform the same psychological function for them. Or perhaps the affinity among crank views occurs because their devotees naturally tend to huddle together for warmth. They themselves might claim that once they see how the mainstream holds a false or wrong position on one topic, ignoring, isolating, or silencing the truth, they become
more willing ro question other mainstream views. Mr. Regan, a well-regarded professor of philosophy at North Carolina State University, does not, to my knowledge., fit the mold of crank. He bas written a lucid, closely reasoned, and dispassionate book presenting the case for animal rights. He concludes that animals have farreaching rights that require very extensive and mandatory revisions in our current eating practices and our experimentation on and treatment of animals. Mr. Regan's argument binges on placing animals ("mentally normal mammals of a year or more") on a par with mentally "enfeebled" humans. We can only understand the kind of treatment that is morally due the most severely retarded humans if we allow that they have rights, he maintains. And he concludes that mammals, being on a par with these people, also must be granted to have rights. This central argument (through which the book threads many intricate subsidiary arguments) is troubling on three counts-in irs inflated view of animals., in its dee flared view of tbe status of severely retarded people, and (based on equating them) in the inference to its conclusion that animals have rights.
Mammals above one year in age, according to Mr. Regan, "have beliefs and drives; perception, memory, and a sense of the future, including their own future; an emotional life together with feelings of pleasure and pain; preferences and welfare-intere.sts; tbe ability to initiate actions in pursuit of their desires and goals; a psychophysical identity over time; and an individual welfare in the sense that their preferential life fares well or ill for them." This is a highly overblown description. The data on animals might be explained without such heavy cognitive apparatus, for example by using the operant-conditioning theories of B. F. Skinner that view behavior as shaped by reinforcements. Mr. Regan seems to hold that if more theory is needed for humans, then the same must hold for animals. However, later stages of evolution may have introduced new levels of complexity, so that only in higher primates or humans did higher cognitive and emotional proce.sses emerge. Mannals may not be as psychologically fancy as Mr. Regan believes. Even if they are, their beliefs, desires, intentions, future orientation, etc., surely are Copyrighted material Do Animals Have Rights? -& 307 far more rudimentary than those of most humans, and this may make
much moral difference. Whether animals have higher capacities than behaviorists admit is perhaps nor absolutely crucial to Mr. Regan's case. He might argue that the severely retarded and handicapped humans co which he needs co compare animals also lack such capacities and thus remain on a par with the mammals. [n either case, if granting animal rights depends upon deflating the status of seriously retarded people, then these people's advocates understandably will lind Mr. Regan's case unwelcome and offensive. These people are, after all, human beings, of the same species as we, however retarded or handicapped. Even supposing a particular severely retarded individual turns out ro be no more rational or autonomous and tO have no richer an internal psychology than a normal member of another mammalian species, he nonetheless is a human being, albeit defective, and must be treated as one. Such a consideration will smack of "speciesism" tO Mr. Regan; it makes the species an individual belongs to morally relevant. It is ironic that Mr. Regan's "good cause" depends upon assessing each "enfe.ebled" human only by his or her particular characteristics and not by species membership, while the "good causes" favoring affirmative action must maintain group
memberships are relevant. (Of course, one may rake the position that what species you belong to, whatever your individual characteristics, is morally relevant, while what race or sex you belong to is not.) I lind it hard to believe that if society accepts Mr. Regan's equating of mammals and "enfeebled" humans, this will result in a recognirion of animal rights. Our view of what treatment severely retarded people are owed surely in part depends on their being human, members of the human species. Sweeping away this consideration as morally irrelevant can only result in society's treating severely rerarded people like animals, not the other way around. Nevertheless, it is not easy to explain why membership in the humao specie.s does and should have moral weight with us. Shouldn't only an organism's own individual characteristics maner? Normal human beings have various capacities that we think form the basis of the re spectful treatment these people are owed. How can someone's merely being a member of the same species be a reason to treat him in certain ways wben be so patently lacks those very capacities? This docs present a puzzle, henc.e an occasion to formulate a deeper view. We then
would understand the inadequacy of a "moral individualism" that looks only Copyrighted material 308 -& DISCUSSIONS AND REVIEWS at a particular organism's characteristics and deems irrelevant something as fundamental and essential as species membership. Since we lack an explicit account of why being a member of the human species is morally important, someone might hold that the burden of proof lies on those who would treat one-year-old mammals differently from very severely retarded humans. This is not a substantial point. The speciesmembership characteristic of being human is not, on surface, blatantly irrelevant morally. Nothing much (certainly not the large institutional changes Mr. Regan favors) should be inferred from our not presently having a theory of the moral importance of species membership that no one has spent much time trying to formulate because the issue hasn't seemed pressing. Something more should be said about membership in a species. The traits of normal human beings (rationality, autonomy, a rich internal psychological life, etc.) have to be respected by all, including any denizens of Alpha Centauri. Bur perhaps it will turn our that the
bare species characteristic of simply being human, as the most severely retarded people are, will command special respect on.ly from other humans-dus as an instance of the general principle that the members of any species may legitimately give their fellows more weight than they give members of other species (or at least more weight than a neutral iew would grant them). Lions too, if they were moral agents, could not then be criticized for purring the interests of orher lions first. The question of what distinguishes two situations we want to treat differently is a pertinenr one; it is illuminating when we succeed in formulating the distinction. Mr. Regan is not alone in using a general form of argument that concludes things are on a par when one fails to formulate relevant differences. Many moral philosophers, including myself on occasion, have done the same in discussing a wide range of topics. Nevertheless, it may be a mistake to expect there always can be a succinctly formulated distinction based on a manageably sma.ll set of properties. Sometimes the distinction between situations will lie in their differing places in a whole intricate tapestry (or in two separate
tapestries whose extremities bear some resemblance). We see humans, even defective ones, as pa_rr of the multifarious texture of human history and civilizations, human achievements, and human family relations. Animals, even year-old mammals, we see against a different background and texture. The differences are enormous and Copyrighted material Do Animals Have Rights? ~ 309 endless. It will be asked, "But what precise aspects of these endless differences legitimately make the moral difference?" This question assumes that something much simpler than the total differences between two rich tapestries (one richer than the other) will, by itself, constitute the morally relevant difference. Yet this need not be so. For the two particular organisms, human and other mammal, we can state a difference. One is part of one tape.stry, one of another. But if we are asked to state the morally relevant difference between all of human civilization and the animal kingdom, we may find ourselves, understandably, without any succinct answer. I have some worries in presenting this type of position, for scoundrels too may seek refuge from criticism in the reply that some differences are too great and complicated to srate. Moral philosophers need therefore to develop new tools to handle complex contexts.
Perhaps until they can morally assess the differences between such toral contexts and not merely consider relatively simple differentiating characteristics, their conclusions, especially when startling, will not be raken seriously. Mr. Regan takes the very strong view that animals have rights, which are valid claims against others to particular treatment and to the assistance of others to enforce this treatment. From this view, he draws extremely strong consequences about vegetarianism (the force of law should be brought to bear, if necessary, on farmers who raise animals for food) and about experimentation on animals (whether i.n toxicity tests or testing new medical drugs, no experimentation that puts animals at risk of harm is permissible). However, the preservation of endangered species has no special claim for Mr. Regan, since it doesn't focus only on an organism's individual characteristics. If once it is granted that yearold mammals possess rights and have a welfare they can experience, we can imagine more extreme proponents calling for welfare payments for year-old mammals, nonvoting citizenship for them, and imerspecies marriage rights. I think Mr. Regan himself would regard these proposals as going too far, indeed as cranky. His position that animals have rights,
in my view, goes too far. What, then, should we conclude about animals? On the way to his own position, Mr. Regan discusses other positions that hold that animals do not have rights. One position holds that some people owe animals certain sorts of treatment. Another holds that people, because Copyrighted material 310 -t> DISCUSSIONS AND REVI EWS of facts about their moral interactions among themselves, ought to behave in certain ways toward animals also, though this behavior is not, strictly, owed to the animals. Variants of these positions are worth exploring further. A third position I especially think should be formulated in detail is this: Animals do have some inherent value and hence may not be treated any way anyone might please, but their inherent value is not equal ro that of people. (Mr. Regan offers an illuminating discussion of the nature of inherent value, but his very brief arguments for the equal inherent value of year-old mammals and humans I find unconvincing.) If animals and people differ in inherent value, there can be reasonable disagreement about the magnitude of animals' inherent value and so about whether behavior of ours is responsive to the value they possess. Hence some choices-for example, the choice about whether animals are to be eaten-may be partly personal moral decisions about what sort of
person to be, rather than ones clearly owed to the animals. The label of an "ism" does not fit well here; one may choose not to eat meat on moral grounds without endorsing "vegetarianism." Is it the mark of a cran.k to pay any attention at all to the question of the proper treaunent of animals? I think not. The topic is not a trivial and unimportant one, even if it is not at the top of the list of urgenr moral ropics. A humane civilization will find the time to formulate and e.srablish a balanced treatment of animals. A human civilization will want this as part of its own texture. Mr. Regan is to be commended for coming forth with a careful, sophisticated, and reasoned statement about the moral starus of animals. By going much too far in his conclusions, he has made salient the task of formulating, with a care and sophistication that match his own, the derails of a moderate and balanced, ethically sensitive position on the subject. We owe the animals at least that. Copyrighted material PHILOSOPHICAL FICTIONS Copyrighted material Copyrighted material 19 FICTION I am a fictional character. However, you would be in error to smile
smugly, feeling oncologically superior. For you are a fictional character, too. All my readers are except one who is, properly, not reader but author. I am a fictional character; this is not, however, a work of fiction, no more so than any other work you've ever read. It is not a modernist work rhat self-consciously says it's a work of fiction, nor one even more rricky that denies its fictional status. We all are familiar with such works and know how tO deal with them, how to frame them so that nothing the author says--nothing the first person voices even in an afterword or in something headed "author's note"-can convince us that anyone is speaking seriously, tron-fictionally in his own fi.rst person. All the more severe is my own problem of informing you that this very piece you are reading is a work of non-fiction, yet we are fictional characters, nevenheless. Within this world of fiction we inhabit, this writing is non-fictional, although in a wider sense, encased as it is in a work of fiction, it too can only be a fiction. Think of our world as a novel in which you yourself are a character. Is there any way to tell what our author is like? Perhaps. If this is a work in which the author expresses himself, we can draw inferences
about his Copyrighted material 314 -& PHILOSOPHICAL FICTIONS facets, while noting that each such inference we draw will be written by him. And if he writes that we find a particular inference plausible or valid, who are we to argue? One sacred scripture in the novel we inhabit says that the author of our universe created things merely by speaking, by saying " Let there be ... " The only thing mere speaking can create, we know, is a story, a play, an epic poem, a fiction. Where we live is created by and in words: a um-verse. Recall what is known as the problem of evil: why does a good creator allow evil in the world, evil he knows of and can prevent? However, when an author includes monstrous deeds- pain and suffering- in his work, does this cast any special doubt upon his goodness? Is an author callous who puts his characters through hardships? Not if the characters do not suffer them really. But don't they? Wasn't Hamlet's father really killed? (Or was he merely hiding to see how Hamlet would respond?) Lear really was cast adrift- he didn't just dream this. Macbeth, on the other hand, did not see a real dagger. But these characters aren't
real and never were, so there was no suffering outside of the world of the work, no real suffering in the author's own world, and so in his creating, the author was not cruel. (Yet why is it cruel only when he creates suffering in his OW1t world? Would it be perfecdy all right for !ago to create misery in our world?) "What!" you say, "we don't rea II)' undergo suffering? Why it's as real to us as Oedipus's is to him." Precisely as real. "But can't you prove that you really exist?" If Shakespeare had Hamlet say "I think, therefore I am," would that prove to us that Hamlet exists? Should it prove that to Hamlet, and if so what is such a proof worth? Could not any proof be written into a work of fiction and be presented by one of the characters, perhaps one named "Descartes"? (Such a character should worry less that he's dreaming, more that be's dreamed.) Often, people discover anomalies in the world, facts that just don't jibe. The deeper dug, the more puzzles found-far-fetched coincidences, dangling facts-<>n these feed conspiracy and assassination buffs. That number of hours spent probing into a11ything might produce anomalies, however, if reality is not as coherent as we thought, if it is not
real. Are we simply discovering the limits of the details the author worked out? But who is discovering this? The author who writes our discoveries knows them himself. Perhaps he now is preparing to correct them. Do we live in galley proofs in the process of being corrected? Are we living in a first draft? Copyrighted material Fiction ~ 315 My tendency, I admit, is tO want ro revoir, to conspire along with the rest of )'OU ro overthrow our author or ro make our positions more equal, ar least, ro hide some ponion of our lives from him-ro gain a lit de breathing space. Yet these words I write he reads, my secret thoughts and modulations of feeling he knows and records, my Jamesian author. But does he control it all? Or does our author, through writing, learn about his characters and from rhem? Is he surprised by what he finds us doing and thinking? Wben we feel we freely rbink or act on our own, is this merely a description he has written in for us, or does he find it to be true of us, his characters, and therefore write it? Does our leeway and privacy reside in this, rhat there are some implications of his
work that he hasn't yet worked out, some things he has not thought of which nevertheless are true in the world he has created, so that there are actions and thoughts of ours that elude his ken? (Must we therefore speak in code?) Or is he only ignorant of what we would do or say in some other circumstances, so that our independence lies only in the subjunctive realm? Does this way madness Lie? Or enlightenment? Our author, we know, is ourside our realm, yet he may not be free of our problems. Does be wonder too whether he is a character in a work of fiction, whether his writing our universe is a play within a play? Does be have me write rhis work and especially rl1is very paragraph in order to express his own concerns? It would be nice for us if our author roo is a fictional character and this fictional world he made describes {that being no coincidence) the actual world inhabited by his author, tbe one who crear.ed him. We then would be fictional characters who, unbeknownst tO our own author although not to his, correspond to real people. (Is that why we are so true to life?) Must there be a top floor somewhere, a world tbat itself is not created
in someone else's fiction? Or can the hierarchy go on infinitely? Are circles excl uded, even quite narrow ones where a character of one world creates another fictional world wherein a character creates the first world? M.ight the circle ger narrower still? Various theories have described our world as less real than another, even as an illusion. The idea of our having this inferior ontological status rakes some getting used ro, however. It may help if we approach our situation as literary critics and ask the genre of our universe, whether tragedy, farce, or theaterofthe-absurd? What is the plot line, and which act are we in? Copyrighted material 316 -t> PHILOSOPHICAL FICTIONS Still, our status may bring some compensations, as, for example, that we live on even after we die, preserved permanently in the work of fiction. Or if not permanently, at least for as long as our book lasts. May we hope to inhabit an enduring masterpiece rather than a quickly remaindered book? Moreover, though in some sense it might be false, in another wouldn't it be true for Hamlet to say, "I am Shakespeare"? What do Macbeth, Banquo, Desdemona, and Prospero have in common? The consciousness
of the one author, Shakespeare, which underlies and infuses each of them. (So, too, there is the brotherhood of man.) Playing on the intricacy both of our ontological status and of the first-person reflexive pronoun, each of us too may truly say, "I am the author." Note from the Author Suppose I now tell you that the preceding was a work of fiction and the "I" didn't refer to me, the author, but to a first-person character. Or suppose I tell you that it was not a work of fiction but a playful, and so of course serious, philosophical essay by me, Robert Nozick. (Not the Robert Nozick named as author at the beginning of this work- be may be, for all we know, another literary persona-bur the one who attended P.S. 165.) How would your response to this whole work differ depending on which I say, supposing you were willing, as you won't be, simply to accept my statement? May I decide which to say, fiction or philosophical essay, only now, as I finish writing this, and how will that decision affect the character of what already was set down previously? May I postpone the decision funher, perhaps until after you have read this, fixing its status and genre only then? Perhaps God has not decided yet whether he has created, in this world, a fictional world or a real one. Is the Day of Judgment the
day he will decide? Yet what additional thing depends upon which way he decides- what would either decision add to our situation or subtract from it? And which decision do you hope for? Copyrighted material 20 R.S.V.P.A STORY The project began with high hopes, excitement even. Though people later came to think it just dumb, founded on a mistake so obvious that those who started it deserved its consequences, no one raised objections until well after the project was operating. True, everyone said it would be a long venture, probably not producing results for many generations. But at the beginning the newspapers carried frequent repons on its progress ("Nothing yet"). Practical jokers would call saying, "Is this the Interstellar Communications Project? Well I'm a BEM you'd be interested in talking to," or "I have a collect call for the Interstellar Communications Projects from the constellation of Sagittarius. Will you accept the charges?" It was in the public eye, looked fondly upnn. Much thought had been given to deciding what listening devices to use and what sons of signals to study intensively. What would be the
most likely wavelengths for messages to come on? Would the messages be something like lV signals rather than consecutive prose? How would one tell that a signal was sent by intelligent beings rather than produced by some natural process? Investigating this last problem produced the Theological Project as a side-effect, for proponents of the argument from design, one traditional argument for the existence of God, had long wrestled with the same difficulties: couldn't any pattern, however Copyrighted material 318 -<> PHILOSOPHlCAL FICTIONS intricate and wonderful, have been produced by some unknown mech anism? How could one be sure that an intelligence was behind it? Some foolproof test was needed, especially since, with a sufficiently complex manual of translation, any glop coming across could be decoded into an interesting message. Sending a return message and receiving a reply would rake many years, perhaps generations, and it wouldn't do to have everyone on earth jumping for joy and holding their breath if they were just talking to the interstellar equivalent of the bedpost. The solution lay in abstract mathematical patterns, nor rcali.zcd (so far as anyone knew)
in any actual causal mechanism and which {it was thought) couldn't be so realized. For example, there's no known causal process that generates the sequence of prime numbers in order; no process, that is, that wasn't expressly set up by an intelligent being for that purpose. There doesn't seem to be any physical significance to precisely that sequence, to a sequence which leaves out only the non-primes, and it's difficult to imagine some scientific law containing a variable ranging only over primes . . Finding that a message began with groups of primenumbered pulses, in order, would be a sure sign that an intelligence was its source. (Of course, something might be the product of an intelligent being even though it didn't exhibit such an abstract pattern. Bur a being wishing to be known to others would do well to include one.) With alacrity, the theologians jumped on this idea, gaining their first National Science Foundation grant .. Among themselves they called their project "Hunting for God," and the idea {about which other theologians had their reservations) was to look at the fundamental lineaments and structures of the universe, the clustering of galaxies, relationships among elementary particles, fundamental physical constants and their relationships, etc.,
searching for some abstract non-causal pattern. Were one found, one could conclude that a designing intelligence lay behind it. Of course, it had to be decided precisely which abstract patterns would count, and which features of the universe were fundamental enough. Discovering prime-numbered heaps of sand on some heretofore uninhabited island wouldn't do the trick, since one would expect ro find something like that somewhere or other; what the significance would be of finding such patterns in conical functioning or in the structure of DNA was a matter of dispute, with those viewing man as no more fundamental than the heaps of sand accusing their opponents of anthropocentrism. Theologians establishing the Reverend Thomas Bayes Society became expert in forming complex and intricate proba bility calculations and in debating Copyrighted material R.S.V.P.-A Story --& 319 delicate issues about assigning a priori probabilities. The results of the "Hunting for God" project being well known, no more need be said here. The initial excitement aroused by the Interstellar Communications Project was connected with a vague hope that other beings would
enlighten people about the meaning and purpose of life, or with the hope that at least people would learn they weren't alone. (No one explained why the "we" group wouldn't just expand, leaving people plus the others still quire alone.) After the project was set up, the best scientists went on to other more challenging tasks, leaving the rest to wait and listen. They listened and they examined and they computed and they waited. No qualifying abstract pattern was detected, nor was any message that looked intelligent even minus such a pattern. Since newsmen do not find a uniform diet of "no progress" reinforcing, the project was reduced, in order to 611 the auditorium for their third annual pre.ss conference, tO inviting reporters from college newspapers, Sisterhood bulletins, and the like. Up gets this woman to ask why they should expect to hear anything; after all, they were only just listening and nor doing any sending, why wouldn't everyone be doing the same?; maybe everyone else was just listening also and no one was sending any messages. It is difficult to believe that the project had reached this point without anyone's having thought about why or whether extraterre.strial beings would want to try to make their presence known to others. Even though
during the congre.ssional debate on the subject, in all the newspaper columns and editorials, no one once suggested setting up a transmitting station, no questioner asked whether other beings would. Little thought is required ro realize that it would be dangerous simply to start sending out messages announcing one's existence. You don't know who or what is out there, who might come calling to enslave you, or eat you, or exhibit you, or experiment on you, or toy with you. Prudence dictates, at a minimum, listening in for a while to find out if other parties are safe and friendly, before making your presence known. Though if the other parties are at all clever, they would send reassuring messages whatever their intentions. They most certainly would not beam out TV signals showing themselves killing and eating various intelligently behaving foreigners. And if they're really clever, then (by hypothesis) they'll succeed in deceiving anyone not adhering to a policy of staying silent no matter what. Such considerations were neither explicitly formulated nor publicly expounded, but it must have been some feeling about the foolhardiness of Copyrighted material 320 ~ PHILOSOPH ICAL FICTIONS broadcasting first (how else can one account for it?) that led to the notable
but not-then-noted absence of proposals to establish broadcasting stations in addition to the listening posts. Once again the project was a topic of conversation. "Of course," everyone said, "it's ridiculous to expect anyone to broadcast; it's too dangerous. No interplanetary, interstellar, intergalactic civilization, however far advanced, will broadcast. For they won't know that an even more advanced and hostile civilization isn't lurking at the other end of their communications beam." Interest in flying-saucer reports diminished considerably when the conclusion was drawn that the sending out of observation ships presents hazards similar to those of broadcasting messages, since the process of a ship's returning information to its source can be tracked. (Even if a ship were designed to give information to its makers by not transmitting any physical signal, or even returning to its base, there must be some contingencies under which it would do so, since nothing can be learned from a detection device that gives the same response no matter what it detects.) It was said that if its planning committee had included some psychologists or game theorists or even kids from street gangs in addition to the scienti.sts and engineers, the project never would have gotten started in the first place. The legislature wouldn't openly admit irs blunder by ending the project
completely. Instead they cur its funds. They did not authorize the broadcasting of messages. The members of the staff had various reasons for staying with the project ranging, one mordantly remarked, from masochism to catatonia. All in all, they found their jobs agreeable. Like night clerks in completely empty resort hotels, they read and thought and coped comfortably with the lack of outside stimulation. In that manner the project continued, serenely, for another eight years, with only a few comedians desperate for material giving it any mention at all, until the receipt of the first message. Studious observation of reversa ls in public opinion and their accompanying commentaries has never been known to enhance anyone's respect for the public's intellectual integrity. (As for its intelligence, this would be a late date, indeed, to proclaim the news that the public adopts a view only after it is already known to be false or inadequate, or to note the general inability to distinguish between the firstperson present tense of the verb "to believe" and the verb "to know.") People just refuse to admit that they have changed their minds, that they have Copyrighted material R.S.V.P.-A Story ~ 321 made a mistake. So the very same people who said at first, "How exciting, I wonder when the messages will begin arriving," and who later
said, "How silly 10 listen for a message; it's too dangerous for anyone to broadcast," now said, after the receipt of the first message, "Of course a civilization will broadcast, even though it's dangerous, if it's even more dangerous for it not to broadcast." The first message picked up and decoded was a call for help. They were threatened by a coming supernova outburst of their star. No spaceships could escape the wide perimeter of destruction in time, and in any case they could not evacuate all of their population. Could anyone advise them about what to do, how to harness their star to prevent the outburst? Their astronomical observations had shown them that occasionally such outbursts didn't take place as predicted, and since they could discover no alternative explanation for this anomaly, they thought it possible that some civilizations had mastered a technique of inhibiting them. If no one told them how to do it, or came to their aid, they were doomed. Over the next year and a half they beamed out their literature, their history, their accumulated wisdom, their jokes, their sage's sayings, their scientific theories, their hopes. Mankind was engulfed in this concentrated effulgence of a whole civilization, enthralled, purified, and
ennobled. To many they became a model, an inspiration. Their products were treasured and they were loved. Did they view this outpouring as a gift to others, an inducement for others to help, a distillation for its own sake of the essence of themselves? No person knew or was prone to speculate as each, silently with them, awaited their tragedy. Never before had the whole of humanity been so greatly moved; never before had persons been so jointly elevated as in experiencing these beings. At the end of a year and a half came a renewed call for aid; and in addition a call for some response, even from those lacking technical knowledge to help with the supernova. They wanted, they said, to know their messages had been received and understood, to know that what they held most important and dear would be preserved. They wanted to know, as they died, that others knew of them, that what they had done would continue, that it would not be as if they had never existed at all. Only to the misanthropic can the ensuing debate have brought plea sure, the debate that raged among persons, and within some. "It might be a trick, don't reply, it's too dangerous." "Beings capable of that civilization couldn't be up to trickery." Copyrighted material
322 ~ PHILOSOPHICAL FICTIONS "Perhaps they are quoting another civilization they've conquered, or an earlier phase of their own; Nazis could and would quote Goethe." "Even if they're not tricking us, perhaps some other aggressive civilization will overhear our message to them." "How can we let them perish without responding?" "If we could help them escape thei.r fate then certainly we should send a message telling them what to do, even though this would mean running serious risks. But we can't help them, and we shouldn't run risks merely in order to bid them a sentimental farewell." "We can save them from believing, as they die, that they a.re sinking into oblivion." "Why the irrational desire ro leave a trace behind? What can that add to what they've already accomplished? If eventually the last living being in the universe dies, will that mean that the lives of all the rest have been meaningless? (Or is it vanishing without trace while others still remain on that is objectionable?)" "How shall we face our children if we don't respond?" "Will we have grandchildren to face if we do respond?" No government sent a message. The United Nations issued a proclamation beginning with a lor of whereases but containing near the close a gathering of inasmuches so it didn't proclaim its proclamation of
regret very loudly. Bur it did issue an order, in its stated role as guardian of the interests of the earth as a whole, that no one endanger the others by replying. Some disobeyed, using makeshift transmitters, but these were seized quickly, and their signals were roo weak to reach their destination intact through the interstellar noise. Thus began the grim watch and countdown. Watching for their res cue, listening for some word ro them from elsewhere. Waiting for their doom. The time, for which their astronomers and earth's also had predicted the supernova outburst, arrived. Some persons paused, some prayed, some wept. All waited, still. The existence of a finite limit to the velocity of causal signals had been of some interest to physicists. Epistemologists had worried their little heads over the question of whether what is seen must be simultaneous with the seeing of it, or whether people ca.n see far into the past. Now came the turn of the rest. The fate of that distant planet was already settled, one way or the other, but knowledge of it was not. So the wait continued. Copyrighted material R.S.V.P.-A Story ~ 323
For another year and a quarter, remembering their debates, mulling over their actions and inactions, contemplating the universe, and themselves, and the others, people waited. Poetically just things could have happened. A me,ssage could have arrived saying it was humanity that had been tested, it was the sun that was due to outburst, and since the earth's people hadn't ventured to render others aid or comfon, others would not help them. Or, they could have been rescued. (How greatly relieved people then would have felt about themselves. Yet why should someone else's later acts so alter one's feelings about one's own?) But the universe, it would appear, is not a poem. No messages to them were detected. Light from the outbursting of their star reached eanh as their broadcasts (should they have terminated them a year and a quarter before the end?), as their broadcasts and their plays and their science and their philosophy, their hopes and their fears and their courage and their Living glow ended. Some people used to think it would be terrible to discover that human beings were the only intelligent beings in the universe, because this would lead to feelings of loneliness on a cosmic scale. Others used to think that discovering intelligent beings elsewhere would remove
their own last trace of uniqueness and make them feel insignificant. No one, it seems, had ever speculated on how it would feel to allow another civilization to vanish feeling lonely, insignificant, abandoned. No one had described the horrendousness of realizing that the surrounding civilizations are like one's own; of realizing that each neighbor remaining in the universe, each of the only other ones there are, is a mute cold wall. Limitless emitlessness. Lacking even the comfort of deserving better, facing an inhabited void. Copyrighted material 21 TEST AMENT Once upon a rime I decided to make a person. Perhaps from loneliness, perhaps boredom, perhaps playfulness. It doesn't really matter. The project intrigued me. It would have been, of course, no problem to make a machine realiuttion of the elaborately related functional states of a person; to make a machine that behaved exactly like a person. The trick was to make something with an internal emotional and mental life; to make not merely something that behaved identicaJJy to a person, but a real honest to goodness person. So l didn't tinker with plastics, transistors, or other hardware. I took suitable stuff-DNA, protoplasm-and got to work. To tell the
truth, it wasn't so difficult 10 make the thing. I was skilled, after all. I even made a fully grown, mentaJJy developed person rather than an infant who would have to learn slowly. Why not be able to talk to it right away? Liking the use of vowel letters for names, l nnmed it A.~ A talked interestingly, laughed at my jokes, did intelligent things, and seemed happy being there with me. I was happy also. Until my creation wanted company other than me. l didn't like that, if the truth must be told. So (if the whole truth must be told) I pulled a fast one. Instead of making another person with a mental and emotional life of its own, l made a machine to perfectly simulate the behavior edIT' a Testament -t> JH of a person. I called it by the next vowel, "E." A, of course, didn't suspect. How could he? The new arrival behaved and spoke perfectly, responded to what A said with "understanding." And it was made from the same stuff as A was. All E lacked was an inner mental life, feelings, emotions. I felt only a little guilty about deceiving A. After aJl, I had created him; he owed me something. And I had collected in a wa) chat
he never noticed. His experiences were exactly those he would have had if be bad been dealing with a real person, and, after all, what else could matter except how things felt to him, from the inside? Things went along satisfactorily until I decided that A was not a suitable companion for me. The reason was, of course, that A was too stupid. It was no fun having a personal relationship with someone so e.asily deceived. So I decided to make another person fit for my company; one who wouldn't be deceived into thinking that a person simulator was a person. (If this were a science fiction story it would continue: I made the new non-deceivable person, started to talk to it, it observed me and then said that I couldn't fool it, it could tell a simulation of a person when it encountered one. But this isn't one of those fictions with a snappy ending, even a conceptually surprising and intriguing one.) I thought about what such a thing would be like. It would not infer that it was dealing with a person solely on the basis of the thing's behavior. That was A's mistake. Behavior, of course, does not guarantee that it's backed by tbe appropriate internal mental life. If that were a legitimate inference, then each machine in a world of machines could
legitimately make that inference. And they'd all be wrong. Anyone making a similar inference shouldn't have much confidence in it. Perhaps a better inference procedure would be to assume that everything that behaved like you and had tbe same functional aod physical states as yourself was also like you in internal mental life as well. But using this argument from analogy would lead a machine to conclude that all of the other complexly behaving organisms weren't people but were, like itself, other machines (and hence it would go wrong in a world of other people); and it would lead a person to conclude that all of the other complexly behaving organisms were, like itself, people (and hence it would go wrong in a world of other machines). Inferring that everything else was exactly like you did seem to be curying egocentrism a little too far. Prospects looked dim; it was a tough problem. Then the obvious thought came: make the new person telepathic. Such a person would be able to directly experience the experiences of others in the same way be directly experienced his own. Not being Limited to observing the Copyrighted material 326 -<> PH ILOSOPH ICAL FICTIONS behavior of other organisms and formulating hypotheses about what made them tick, he would be able, so to speak, to take a peek
inside and see what was there. Eagedy I set to work. It wasn't too difficult to create the telepath who could have not only his own experience but also those of anyone he focused upon. Everything seemed settled, for if he focused and found no orber experiences, he'd know he'd focused on a simulation: otherwise, discovering experiences, a person. I anticipated a long and rewarding friendship, uotil I realized that I would have to give the det.ector of fakes not merely the telepathic ability to have whatever experiences were being had by whatever he focused upon, but also some means of knowing chat chis thing he's focused upon also was having those experiences. (Otherwise, all the detector would possess would be a way of having more experiences, and not a way of knowing that something else was also having these experiences.) I gave this maner much thought, accomplishing nothing. It wasn't clear how one could build this added device into a detector; it didn't even begin to be clear what might do the trick, given that telepathy couldn't. A lesser person would have admitted defeat and decided that he couldn't build a detector. Bur I am not, very defiJlitely not, a lesser person.
If one line of thought doesn't work, then try another. It's a good maxim, though not one I'd ever had previous occasion to use. So I cast around for an unorthodox approach, and got the bright idea of building into the detector, as its way of distinguishing people from simulations, the very technique that I myself employ tO do the same thing; the way that I had used, for example, to tell that A was a person and that E was a machine simulation. To be perfectly honest, I had never before thought about how I k11ew; epistemology had never seemed rele\'ant to me. After very little pondering, I rea lized that I hadn't tried to find ot whether A orE was a person or machine simulation of one. I mean, l made them in the usual way1 said: "Let there be a person," and A appeared; I said: "Let there be a machine simulation of a person" and E appeared. I just assumed that A was a person, and E a machine simulation. It never crossed my mind that there was some need to check to see if things came out as planned. I mean, when I said "Let there be electromagnetic radiation" and this thing appeared, I didn't check to see what it was. What am I, an experimental physicist? It seemed that I didn't know that things hadn't gone wrong; I
didn't know tl1at A was a person rather than a machine simulation and I Copyrighted material Testament ~ 327 didn't know that E was a machine simulation rather than a person. For someone who until then had thought that he knew it all and that there was nothing beyond his powers, this was a big comedown. Not only that, I felt very foolish. Here I'd been carrying on this long relationship with these things without having the faintest idea of what they were. I didn't even know whether I had a chosen people or a chosen set of machine simulators. I didn't even know, for crying out loud, whether I'd been receiving prayers or just gazing at prayer behavior. Once one gets into that kind of mood, of course, everything begins to seem shaky. Even the electromagnetic radiation I so confidently thought I'd produced. Since a number of alternative total theories were compatible with all the data I had or could get, did I really know that I'd succeeded even there? Or were all those theories equivalent in meaning, so that the question didn't arise? Even in that case, what had I actually produced? Maybe secondary gimmel force rather than electromagnetic radiation. And since alternative situations and things were
compatible with anything I could say or specify, bow could my words and thoughts uniquely determine a panicular creation so as to produce it? Had I heretofore assumed that nothing was beyond my powers only because I'd never checked to see what I'd actually accomplished? And had I assumed that I knew everything only because I never thought to question or check my assumptions and beliefs? And how could I check? Had I actually created anything at all? Even myself? (My "memory" was of having said instantaneously in the language now available only to me, "LET THERE BE -," thereby creating myself, it being possible, of course, for a cause to be simultaneous with one of its effects. But bow had I fully specified myself? Was my language a way behind the limitations of others? Was this "memory" itself a phantasm, not to be relied upon?) These thoughts filled my mind with so many doubts that it was tW longer in my power to forget them. And yet I did not see in what manner I cot<ld resolve them; I was concerned that I must once and for all seriously undertake to rid myself of all the opinions which I had formerly accepted, and commence to build anew (rom the foundation if I uished to establish any firm and permanent structure. I
decided to proceed by setting aside all that in which the least dol<bt could be supposed to exist, just as if I had discovered that it was absollltely false, a"d 1 decided to even follow in this road until I met with something which was certain, or at least, if I could do nothing else, until I had learned for certain that there is nothing else in the world that is certain. Copyrighted material 328 ~ PHILOSOPHICAL FICTIONS 1 thought that I would have the right to conceive bright hopes if I were happy enollgh to discover one thing only which is certain and ind11bitable. I AM THAT I AM. Possessing this one cenainry, how shall I proceed? One might go on to try to deduce the existence of a benevolent God who wouldn't deceive him, and then proceed merrily along from there. That particular route seems all suited to help me with my panicular problem. (It would ill have served A, as well.) But, perhaps, it offers a hint. U I can make the ontological argument work, and if I can prove whom that argument is about, then once again I will he what I necessarily am. Copyrighted material 22 TELEOLOGY Once you come to feel your existence lacks purpose, there is
little you can do. You can, keeping the feeling, either continue a meaningless existence, or end it. Or, you can discover the purpose your existence already serves, the meaning it has, thereby eliminating the feeling. Or, you can try to dispose of the feeling by giving a meaning and purpose to your existence. The first dual option carries minimal appeal, and the second, despite my most diligent efforts, proved impossible. That left the third alternative, where, too, there are limited possibilities. You can make your existence meaningful by fitting it into some larger purpose; by making yourself some part of something else which is independently and incontestably important and meaningful. But a sign of really having been stricken is that no pre-existing purpose will serve in this fashion, for each purpose which to other moods appears sufficiently fructifying then seems merely arbitrary. Alternatively, one can seek meaning in activiry which itself is important, which is self-sufficiently intrinsically valuable. Preeminent among such activities, if there are any such, is creative activity. So, as a possible route out of my despair, I decided to create something which itself would be marvelous, or whose creating would itself be marvelous. (No, I did not decide to write a story
beginning, "Once you come to feel your existence lacks purpose ... " Why am I always suspected of gimmicks?) Copyrighted material 330 ~ PHILOSOPHICAL FICTIONS The task required all of my knowledge, skill, intuitive powers, and craftsmanship. It seemed to me that my whole existence until then had been merely a preparation for this creative activity, so completely did it draw upon and focus all of my experience, abilities, and knowledge. 1 was excited by the task and fulfilled, and when it was completed I rested, untroubled by purposelessness. But this contentment was, unfortunately, only temporary. For when I came to think about it, though it had taxed my ingenuity and energy to make the heavens, the earth, and the creatures upon it, etc., what did it all amount to? I mean, the whole of it, when looked at starkly and coldly, was itself just an object, of no intrinsic importance, containing crearures in a condition as purposeless as the one I was trying to escape. Given the possibility that my talents and powers were those of a being whose existence might well be valueless, how could their exercise endow my existence with purpose and meaning if it issued only in a
worthless object? At this point in my thoughrs I came upon the solution to my problem. If I were to create a plan, a grand design into which my creation would fit, in which my creatures, in serving the pattern and purpose I had ordained for them, would find their purpose and goal, then this very activity of endowing their existence with meaning and purpose would be my purpose and would give my existence meaning and point. Also, giving their existence meaning would, retroactively, make meaningful my previous activity of creation, it having issued in something that turned out to be of value and worth. The arrangement has served. Only occasionally, out of the corner of my mind, do I wonder whether my arbitraril)' having picked a plan for tbem can really have succeeded in giving meaning to the lives of the role-fulfillers among them. (It was, of course, necessary that I pick some plan or other for them, but no special purpose was served by my picking the particular plan I did. How could it have been? For my sole purpose then was to give meaning to my existence, and this one purpose was insufficient to determine any particular plan into
which to fit my creatures.) And, lacking any idea of what a less defective route to meaningfulness might conceivably be, I refuse ro consider whether such a symbiotic arrangement is possible; whether different beings can provide meaning and point to each other's existence in a fashion so seemingly ci rcular. For such questions press me toward the alternative I tremble to contemplate, yet to which I lind my thoughts recurring. The option of ending it all, now familiar, is less alien and less terrifying than before. 1 walk through the valley of the shadow of dear.h, alone. Copyrighted material NOTES AND REFERENCES CREDITS INDEXES Copyrighted material Copyrighted material --- - ------ -- -NOTES AND REFERENCES 1. Coercion An earlier version of part of this paper was read at Columbia University and at Brown Universiry, and I have benefitted from the ensuing discussions. I bave also benefitted from discussing some of the issues treated here with Professor Gerald Dworkin. 1. A useful place to begin thinking about unfreedom is with FeHx Oppenheim's
Dimensions of Freedom, which also classifies the first three examples as cases of unfreedom, and the fourth as not. Though I have found this book very illuminating, I should note that I believe that a correct account of unfreedom will differ significantly from the one ir presents. 2. Hart discusses coercion only in passing, and Hart and Honore do not discuss coercion in any detail, but ins-tead discuss the more general notion of getting someone to do something. No doubt they would have presented things slightly differently had they focussed specifically on coercion. I present their conditions as conditions for coercing someone into not performing a.n action. It is obvious how the conditions must be modified to yield an account of coercing someone into performing an action. In the course of my discussion I sometimes produce an example as an objection to a condition in the account of coercing someone into not doing something which is more naturally interpreted as an objection to the corresponding condition in the account of coercing someone into doing something. Since once the poiot of an example is seen, it is easy to think of ;;mother example for the corresponding condition. 1 present the examples without regard to whether they apply to a condition under discussion or the corresponding one. 3. Hart and Honore list one further condition: Q forms the
inteotion of not doing A only after learning of P's threat. That Q formed the intention of not doing A afrer learning of P's threat may be reason for thinking that he did A because of the threat. But Q may have refrained from doing A because of the threat even though he forrned the intention of not doing it before learning of the threat. For example (this example applies to the corresponding condition), Copyrighted material 334 ~ Notes and References Q intends to visit a friend tomorrow. P threatens him with death if he doesn't go. Q then learns that rhis friend has a communicable disease such that were it not for the threat, Q wouldn't visit him. Bur Q goes because of P's threat, though he'd formed the intention of going before learning of the thre3t, and never lost this intention. Though Han and Honore's funher condition is not satisfied, P coerced Q into going. 4. Or, to handle cases of anonymous threats: Q knows that someone bas threatened ro do the something mentioned in (1), if he, Q, doe-s A. 5. No weight should be placed on the word consequence. Sometimes it will be more appropriate to say "'result," "effect," "state of affair$," "event," etc. Perhaps the condition should be formulated by saying "the thing which P has
threatened to bring about." 6. This condition requires further refinement co handle cases in which unbeknownst to P, Q wants to avoid ps inflicting the threatened consequence only because this will lead ro some funher consequence detrimental ro P, which Q (only out of concern for P's interest) wants ro avoid. For example, Q refrains from A because he knows tha< P will feel enormously guilty after he's inflicted the consequence, and Q doesn't want this tO happen. Or, Q refrains from A in the face of P's threat to 6re him only because without Q working for him, P will go bankrupt, and Q doesn't want this to happen. I shall no< pursue here <he details of a principle which would exclude these as cases of coercion. 7. I included the latter disjunct since I can threaten you with a consequence which I don't believe would actually worsen your alternative of doing A, but which I know that you believe would do so. Subtle questions arise about cases where it is in the making of the threat itself that causes the person to believe that one consequence is worse: than another. For example., a Gestapo agenr questiortiog a prisoner believes that rwo concentration camps are equally bad, and the prisoner too initially believes this. The Gestapo agent tells the prisoner, in a threatening voice-, that he wiJI be sent to a concentration
camp in any case, but if he cooperates during the questioning he will be sent to the first camp, whereas if he does not, he will be sent to the second camp. Here it is the very making of the threat which caus<s the prisoner to think tha< the second camp is worse than the first. I might note one refinement of this condition, to handle cases where (pan of} P's reason for so deciding is as described in the condition, but rhis part drops away and P sticks with his decision for another reason entirely and thereafter announces the decision. It might be more appropriate to say something likt: (Pan of) P's reason, at the rime he informs Q he will bring about the conse quence or have i< brough< about if Q does A, for plaru1ing ro bring about the consequence or have it brought about if Q does A, is that P believes ... Copyrighted material Notes and References -to 335 The condition in the text should be interpreted or extended so as to cover ca ses in which the worsening of Q's alternative of doing A is not pan of P's reason for deciding to bring abour the consequences if Q does A, but rather (a) P decides to bring about the consequences if Q does A because he believes he has a duty or obligation 10 do so. (b) P knows this consequence would worsen Q's alternative of
doing A. (c) Parr of the reason for P's bringing about of such a consequence if Q does A originally being thought to be his duty or obligation, or being continued to be so thought, is that such a consequence worsens Q's alternative of doing A. 8. This disjunction is condition 3'. Thus the fu ll condition 3' is: (Part of) P's reason for deciding is ... , or, if P hasn't decided, (pan of) P's reason for say ing is ... An alternative condition would be just: (Parr of) P's reason for saying is to get Q not to do A, or to worsen . . . This alternative condition differs from the one under consideration for cases where P has decided to bring about the consequence if Q does A, and no part of his reason or motive is as described, but part of his reason for telling Q that he will bring about the consequence if Q does A, is to get Q not to do A. I find it difficult to decide between these conditions, though I lean toward the one presented in the text. A specific example for wh.ich the condition in the text and the alternative condition diverge is discussed (as case 3) in the section on Threats and Warnings. 9. Note the difference, with respect to coercion, between saying to a man who intends ro do A in order to bring about x: (1) If you do A, I'll do something which (just) prevents your A from bringing
about x. (2) If you do B, I'll do something which would, were you to do A, (just) prevent your A from bringing about x. 10. It was noted earlier that linle weight should be pur on the word consequence. Here we are considering cases where one is rempted to say that P does not bring about any conS<!quence; he just prevents Q from bringing one about. II. One naturally notes, for many examples of the son under consideration, that if Q does not do A and P performs the action which would r:hwan Q's A achieving x, had Q done A (it is not always possible for P to perform this action if Q doesn't do A; e.g., " Lf you mail the letter, I shall intercept it before it reaches him"), then no bad consequence is visited upon Q. But a condition builr upon this observation would fail on two counts: ( I) It would count Q as coerced when he refrains from doing A because P threatened to do B if Q does A, where 8 is such as to just prevent Q's A from bringing about x, if Q does A, and 10 inflict gre.at harm on Q if he does not do A. Copyrighted material 336 -& Notes and References (2) It would count Q as not coerced when he refrains from doing A because P threatened to bring about a consequence if Q does A, wh.ich conse
quence is harmful to Q only if he does A. 12. I should note that I do nor discuss in this paper, and wish here to leave open, two further conditions. (Hart mentions something in the area of the first.) ( 1) The consequence which P has threatened is so weighted by Q as to override the weight which Q (morally) ought to give ro not doing A. For example, Q, who is not in dire financial condition, and would just slightly rather not kill people (he feels about killing people as most people do about killing flies), kills R because P has threatened nor to return the $100 he's borrowed from Q unless Q kills R. Did P coerce Q into killing R? (2) The weight which Q does give to nor doing A does nor fall far short of the weight he (morally) ought to give to not doing A. For example, Q destroys R's home because P has threatened a consequence, if Q does not destroy the home which R has laboriously built, which Q weights and anyone (morally) ought to weight as worst than destroying R's home. However, Q just slightly would rather not destroy R's home. Did P coerce Q into dc,stroying R's home? 13. If one is reluctant to say that the members of the gang have threatened him, then a slightly more complicated account of coercion, in terms of threats and implicit, or quasi, or surrogate threats, must be offered.
14. Complication: Suppose that as the stranger is being be.aren, he says that if they stop and promise to release him, he'll sign over a traveller's check to them for $1 ,000. They stop, he signs it over, they release him. Was he coerced into signing it over? 15. Slightly modifying the conditions previously set out, we obtain, for these situations: (1) P performs an action such that, if Q then does A a certain consequence will ensue. (2) A with this constquence is substantially less eligible as a course of con duct for Q, than A without this consequence. (3) P knows that the a<o he's performed satisfies (1) and (2), and intends Q ro know, and know he's intended to know, that such an act bas been performed. (4) (Part oO P's reason for performing his action is that (he believes) its con sequence if Q does A would be believed by Q to worsen his alternative of doing A. (5) Q does nor do A, and (part of) Q's reason for not doing A is to avoid or lessen the likelihood of this consequence. (6) Q believes that P (or that someone) has done something inttnding that this consequene<, which he thinks Q will think bad, will ensue if Q does Copyrighted material
Notes and References ~ 337 A, and Q believes that he is inttndtd to know (and intended to know that he is intended to know) this. (7) Q believes that, and P believes that Q believes that the consequence of P's action if Q does A would leave Q worse off than if Q didn't do A and P didn't do his acr. 16. If it is coercion several interesting questions arise. I shall mention only one which has no obvious analogue about the 6m kind of coercion discusstd. If conditions 1-7 apply toP and Q, and person R, whom P has never thought of (it is no parr of P's reason for performing the act that it worsens R's alternative of doing A) refrains from doing A to avoid the consequence (P's acr is such that though direaed to Q, it would inflict the consequence upon anyone who does A), was R coerctd into not doing A? Though many questions that arise for this notion correspond co questions about the first notion, it is not obvious that the corresponding questions about the two notions must be answered in the same way. In particular, one is more willing, I think, to call a case of the s<:cond sort a case of coercion (assuming rhat some cases of rhis son are) even though P lacks some of the specified intentions or reasons, than one is to call a case of the first sort, when P lacks
somr of the specified intentions or rtasons, a case of coercion. 17. It should be mtntioned, in fairntss to my mother, that this example was suggested during the discussion at Columbia by someone whose name I shall not mention, in fairness to his mother. It is as a special example of this sort of situation that one might understand the activities of some charitable organizations which, along with an appeal for funds, send a "gift," attempting perhaps tO present one with the alternatives of (a) returning the gift, making no contribution, and feeling slightly embarrass<: d. (b) keeping the gift while making no contribution, and feeling somewhat guilty and uneasy. (c) making a contribution. 18. An interesting question arises for accounts of a notion, such as mine, which (attempt to) provide necessary and sufficient conditions for the central part or core of the notion, and then handle further cases by specifying the relations in which they stand to the core cases. Given a set o f conditions, which are purported to be necessary and sufficient for the core cases of a notion, and given an example to which the notion applies but which does not satisfy rhe conditions, bow is it to be decided whether the example is a counter-example requiring the modification of the conditions, or whether the
conditions are to be retained a.nd the example handled as a non-core case by specifying its relation to cases satisfying the conditions? I would hope that the reader does nor find objectionable my treatment of some particular cases as cases which should satisfy the core conditions (in the previous section) and of some other Copyrighted material 338 ~ Notes and References ca>es as not being core ca~ (later in this section), even though the basis for choosing to treat the particular cases as I do is not stated here. An alternative procedure to the one followed in the first parr of this section is to accept the previous account as the full account of coercion, and to widen the notion of what actions a threat is about. (Why this is an alternative procedu.re will become clear as the reader comes to the numbered statements which soon follow in the text. In the norntion used in these numbered srntements, the alternative procedure would involve saying that the threat is not only about the act A, but is also about the B's.) There seems to me to be some stight reason for the course followed in the text, but it is not clear that anything very important depends upon which way ooe proceeds. 19. Readers who notice my sloppiness in the use of quorntion marks here will know how to remedy it. One may wish to limit the final
formulation of such a principle so that some cases in which P does not know that r, and r'2 have the same reference, do not count as P coercing Q into doing 8. 20. Note that the consequent of 4 is not equivalent to: P coerces Q into doing 81 or P coerces Q into doing 82, or ... , or P coerce.s Q into doing B . 21. One may be reluctant to apply this principle, as it stands, to siruarions where unbeknownst to P, Q is specially handicapped so that he can do A only by doing some horrendous 81. One must also be careful not to misinterpret conclusions reached by applying this condition, as in the case (where n ; 1) where R advises Q to go to the movies and P threatens Q with death if he does not go to the movies. Since Q, whom P coerces inro going to tbe movies, can go to the movies only by doing what R advised, by applying the condition we reach the conclusion that P coerces Q into do ing the action advised by R, which is easily misinterpreted. 22. A useful quesrion to consider is why the statements obrnined from 18 by replacing "coerces Q into doing A" by "persuades Q to do A" (making other obvious changes) arc unsatisfactory, where t-8 themselves are not unsat isfactory in the same way.
23. Other writers (Laswell and Kaplan, Dahl) do not say that thr<ats are involved, but claim that inducements, or positive rewards coerce. 24. A more complicated statement would be required to take into account condition 7 in rhe section on Conditions for Coercion. (This condition was prompted by the example in which P says that he will tu.rn off his hearing aid if Q says another word.) Since in the prtsent section no examples which violate condition 7 are considered, the complication is omitted. 25. P can threaten Q even if rhe consequence does not worsen the consequences of Q's doing the action, so long as P believes it does. Similarly for Copyrighted material Notes and References ~ 339 offering. We shall not consider rhis complication, since such rhrears and offers will normally not have rhe appropriare result in actions of Q. 26. I ignore problems about consequences along a conrinuous dimension, where there may be no first point of changed preference, or of change from preference to indifference. 27. The notion discussed here should be disdnguisbed from another in which both rhreatening Q wirh x if he does A and offering him y if he doesn't do A are said to predominantly involve an offer (threat) if for almost any acdon 8, if Q is both rhreateoed with .T if be does 8 and offered y if he does 8, Q will
prefer to do 8 (not do 8). 28. I ignore problems arising from a divergence between what P believes ro be the morally expected course of evenrs and what is the morally expecred course of events, e.g., where P believes he's morally required to let Q drown, although he's morally required to save Q. Consider a further case (a fter Braithwaite), P and Q are neighbors, and P would practice his violin each night, whether or not Q is in his own apartment. Q detests bearing P pracdce, and asks P to stop. P refuses. The question of monetary compensation for P 's stopping is raised. Suppose thar Q's property rights are not violated by P's practicing, and that $500 is rhe least amount of money which could get P to stop practicing for one year, and that $2,000 is the grearest payment Q would make to P to stop practicing for one year. (Both amounts indicate their real preferences.) Suppose rhar P says that he will stop practicing for one year if and only if Q pays him $n. One intuitively wanrs to say that there are some amounts n such rhat P would be offering to stop in return for S11, and some (higher} amounrs such that it would be a threat not to stop unless Q paid P the money. The difficulties in devising a theory of a reasonable., or just, or fair pric.e need no elaboration here. Disagreements over
the range in which a fair price would fall or over whether rhere is any coherent notion of a fair price which applies to this situation, may lead to disagreemenrs about whether we here have a threat or an offer. 29. An alrernarive view would hold rhar it is an offe~ but rhat doing something because of such an offer counts as being coerc.ed inro doing ir. This would require modification of our earlier account of coercion co include doing things because of such offers. (Such offers being often ro nor continue the usual though morally forbidden course of evenrs, and to switch (at least temporarily) ro the morally expected course of events wh.ich rhe recipient of the offer would prefer to the usual course of evenrs.} Readers who find unsatisfactory my here calling rhe slave owner's statement a threat, may call it an offer and tre.at "threat"' appearing in the section on Conditions for Coe.rcion as a technical term which includes such offers. 30. What if be prefers that he nor be supplied with dope, but can't resist buying it when it's available? Copyrighted material 340 ~ Notes and References 3'1. Let me suggest as a fertile area for testing lntuirions and theories, the fol lowing, where the normal and morally expected cou~ of events may diverge, and where it may not be clear what the morally expected course
of events is. Suppose some nation N were to announce that it will in the future give economic aid to some other countries provided that these other countries satisfy certain conditions (e.g., do not vore contrary to N on important issues before the United Nations, do not trade with specific nations, do not have diplomatic relations with specific nations). Would this an.nou.ncemenr c.onsritute an offer tO give these nations aid, or a threat not to do so? Relevant factors (to list just two of many) are wherher or nor N bas an obligation or is morally required to give these nations economic aid (independently of whether they satisfy the conditions), and whether or nor N has previously given these nations economic aid independently of whether they satisfy the conditions. 32. Though this seems to me to be the correct thing to say, there is a problem, which I have nor yet been able to solve, which I should briefly mention. Letting P = you are punished C = you commit a crime the officials in the society might say P if and only if C or equivalently not-P if and only if not-C. Interpreted truth-functionally, each of these is equivalent to either (P and C) or (not-P and not-C). The two remaining possibilities are (P and not-C) and
(C and not-P). The t>ackground we want to use in deciding whether a threat is involved is C and nor-P. If we were tO use the remaining possibility, P and note, as the background, then it would turn out that an offer is involved here. The problem is to formulate criteria, in cases where the biconditional is itself part of rhe normal and expected course of events, which pick out C and nor-P rather than P and not-C in this and other threat cases, and which would pick out the appropriate background for offer cas<:s as well. 33. This is a case of blackmail, which presumably should be legally forbidden because allowing it increases the probabitity that crimes will go undetected. Other reasons apply to other cases, but note that it is not obvious that one wants to legally forbid all cases wh.ich fit the description: saying that one will make public some information about Q unless Q pays money. For example, (a) P's saying that he will make pubtic the information rhat Q has not paid P the money Q owes him, unless Q pays the money. (b) P is writing a book, and in the course of hi.s research comes across information about Q which will help sell many copies of the book. P Copyrighted material Notes and References ~ 341 tells Q he will refrain from including this information in the book if and
only if Q pays him an amount of money equal to the expected difference in his royalties between the book contairung this information and the book without the information. 34. A coercive consequence for Q's A is a consequence which has been threatened if Q does A. 35. More precisely, the credible threat of raising the probability of this consequence from what it is without P's aiding in bringing it about, to whar it would be with P's aid. 36. Lawyers sometimes speak of the distinction as being between threats and predictions. Since philosophers sometimes contrast predictions and statements of intention, and the latter may be "predictions"' in the lawyer's sense, to avoid confusion, I speak of threats and nonthreatening warnings. 37. Stipulating that each of the members of some particular majority has this preference ordering enables us to avoid problems, relevant ro our concern here, about nontransitive majorities. 38. I consider here cases where the employer could stay in business (without running at a loss) even if the union wins. In the case where,, if the un.ion wins, the employer cannot both stay in business and out of the red, ir is clear that his statement is a warning and not a threat. (In the normal course of events he does go out of business if the union wins, and chooses tO do so earlier than he must,
in order ro cut his losses.) I assume, for the cases discussed in the text, that the employer making the statement intends to close if the union wins. lf he intends not to close, or has no setded intention either way, then in stating that he will close if the union wins he is making a threat. I should note thar I am assuming for some of the cases in which the employer could profitably stay in business if the union wins, that he does not have an o bligation to and is not morally required to remain in busine.ss if the union wi.ns the ele<:tion. It may be that some disagrumems about whether the employer is threatening or warning s tem from disagreements about whether he is mo rally required ro remain in business if the union wins (morally required nor ro close because of dislike of running a unionized business, etc.). 39. Cf. Schelling, The Strategy of Conflict. Note that according to contem porary utility <heory, it will be reasonable for <he employer to rule our (b) for strategic reasons even if he doesn't know the preference ranking of the employ ees, so long as pu(a) + (1 - p)u(c) > u(b), where p is the probability that the union will lose the election after be's ao nounctd that if they win, he will go out of business, and u(x) is the utility of x to the employer. Copyrighted material
342 ~ Notes and References 40. I should oore that I have done nothing here to argue, as I would wish ro, that acting on such strategic considerations is not part of the normal or expected course of events which forms the background ro discussing questions of coercion. Not doing something unless you'd first announced it in this sort of strategic situation, for strat<gic reasons, should be distinguished from not doing something without prior announcement for other sorts of reasons; d. discussions of ex post facto laws. 41. I wish here to exclude threats against certain acts which harm others, ere. It is difficult to determine whether there is a presumption against threatening someone against (or coercing someone into not), e.g., murdering someone else, which is almost always easily overridden, or whether there is no such preswnption in such case.s. It is also difficult to determine exacdy what the difference is between these two altemativ<S. For an attempt 10 describe the difference, see "Moral Complications and Moral Structures" (Essay 10), section 7. I do not discuss the possibility that the employer in case 3 is engaging in the second type of coercion discussed in rhe section on Conditions for Coercion, so that his prior act of announcement worsens the employees' alternative of
electing the union because they will feel worse having the factory close when they have been warned of this than they would if, without warning, they elected rhe union and the factory closed. 42. The Supreme Court held in Textile Workers Unio , v. Darlington Manu( a<turing Co. (380 U.S. 263 (1965)) that it is not an unfair labor practice for an employer to close his entire business, even if the closing is due ro antiunion animus, but that closing part of a business is an unfair labor practice if tl1e purpose is to discourage unionism in any of the employer's remaining plants, and if the employer may r<asonably have foreseen such an effect. Given the difficulties in dettrmining an employer's purpose, one suspects that the effect of this decision wiU be to forbid all employers from closing part of their business bec.ause, it has become unionized, if other parts of the business are not unionized. 43. This brief description is me.anr to indicate the area of concern rather than as an account of paternalistjc legislation. In such an account one would have to distinguish this son of legislation from other legislation, often called paternalistic, which provides fo r adults what parents are expected to provide for children, e.g., food, shelter, money. (I do not claim that no common account of the two sorts of legislation can be given.) It is held, by people who caU
such legislation paternalistic, that adults are supposed to provide these things for themselves, or through agreements with other citizens qua private citizens. When legislation and governmental institutional arrangements provide things which parents provide for children but which adults are not supposed to provide for (solely by) themselves, e.g., protectjon from the infliction of violc.nce by others, such provision is nor terme.d paternalistic. Copyrighted material Notes and References ~ 343 For the area of concern in the text, since different sorts of reasons can be offered lor the same piece of legislation, one should spe<>k of paternalistic re.asons lor legislation rather than of paternalistic legislation. One wants an account of paternalistic reasons for legislation to bave rbe coosequeoce that some reasons pur forth for legislation which would make people unfree to manufacture or sell cigarettes would be paternalistic reasons even though they do not involve rhe prortcrion of the (perhaps nonsmoking) persons made unfree to manufacture and sell cigarettes. I shall not pursue the details here. Note that some paternal isric acts can involve great selhacrifice, as when drugs are legally forbidden in order 10 protect those who are not addicts who would be so under a sy.srem in
which drugs were legal. The price we pay to protect them is increased risk of being robbed or assaulted by addicts trying to acquire money in order to pay the high prices on the illegal market, plus the diversion of resources imo trying to enforce the law. Perhaps it is appropriate that we should all suffer for our original unjustified paternalistic intervention. The reader might find it useful, in thinking about paternalism, to consider whether there are any limits to the severity of the penalty we would include in a paternalistic law, and how these limits are to be fixed. Could we, for example, have the death penalty for the offense of swimming at a beach when no lifeguard is present? Certain plausible-looking principles would allow this, because when the system including this penalty is instituted, it is the one of the alternatives which is expected to best operate for the person's own good. Surely something has gone haywire here. 44. Distinguishing two typecS of offers in a manner si.milar to the earlier distinction between two sorts of coercion, this case does not fit the first type of case. For here it is not the case that after Q beats P up, P will then do something which improves the consequences for Q of this action. (Even if P wiU then do this, e.g., spread the news that Q is not to be threatened, it is not
an offer for the reason mentioned in the text.) And even though it fits the second sort of case, that is, P now does something (threatens Q) which improves the consequences of Q's beating him up, P is not offering something to Q to beat him up, since P Jacks the requisite reasons involved in making an offer. 45. I omit consideration of offers to do acts such that, if the offer is made, the act can.not be done wirh.out accepting rhe offer, e .g., one cannot work at certain government jobs without receiving a salary of at least one dol.lar per year. The Rational Man may sometimes prefer doing the act without the offer's having been made, so that it will be clear to others, and perhaps himself, why he does the act (e.g., not for the money). I also shall not consider the case of a person's not welcoming an offer for him to perform a malicious act btcausc of what it shows about the person making the offer. Note the importance of our re.stricting our attention here to the Rational Man. Another person might not, for example, welcome an offer of sso;ooo for him to kill Jones, because Copyrighted material 344 ....,. Notes and References he's afraid he may be tempted to (and unable to resist the temptation to) accept the offer. In considering only the Rational Man, I am leaving pan of my task
undone. For I do not argue, as I would wish to, that even for someone who sometimes succumbs to temptations which he believes he ought to resist, there is a significant difference between offers and threats. 46. For a discussion of the prisoners' dilemma, cf. Luce and Raiffa, pp. 94102. One often finds this argument applied to questions about the pro,ision of a public good for a group. For example, each inhabitant of an island might prefer that ochers contribute ro the construction of barriers against the sea while he does not, yet prefer everyone's being forced to contribute to contributions being left purely voluntary, in which case, let us suppose, the barriers won't get constructed. (For a discussion of the conditions under which a public good for a group will be provided, c. Olson. Buchanan and Tullock argue that public goods for a group will be provided mote often than one might think.) One must be wary of concluding too quickly from rhis line of argument that there will be unanimous consent to the provision of the public good by forcing everyone to contribute. For there will generally be alternative ways in which the public good can be provided, and individuals even if they all agree that each of these ways is preferable to the purely voluntary situation, may differ about which of <he ways should be used. Should the good be paid for
from funds gathcrtd via a syst~m of proponional taxation, or one of progressive taxation? And so forth. h is nor obvious how unanimous consent to one pnrlic.ulnr way of providing the good is suppose to arise. 47. Even if Q has other rea.sons for not doing A, we distinguish bc.twecn "'Q has a reason r for not doing A," and "r is (part of) Q's reason for not doing A." 48. This indicat<.s an asymmetry between doing something (partly) because of a threat, and doing something partly because of an offer. For suppose that the other reasons Q has for not doing A which are part of his reasons for not doing A include an offer by R for Q not to do A. Using a classificatory notion of coercion, doing A partly because of a threat shows the person was coerce.d, whereas doing something partly because of an offer does not show that he was not coerced. 49. For a discussion of classificatory, comparative, and quantitative concepts cf. Hempel, part ill. An illuminating study of different scales of measurement is Suppes and Zinnes. 50. A good place to begin in chinking about the weight of reasons is with Ernest Nagel's discussion of the weight of different causes in The Struct11re of Science, pp. 582-588.
51. There are other factors which one might wish to build into a quantitative concept or measure of coercion, though perbaps there is no natural way to combine aU of the facrors into one mc-.asure. For a discussion of some similar questions about measuring freedom cf. Oppenheim, chap. 8. Copyrighted material Notes and References ~ 345 References for Essay 1 Bay, Christian. 1958. The Structure of Freedom. Palo Alto, CA: Stanford University Press. Buchanan, James M., and Gordon Tullock. 1962. The Calculus of Consent. Ann Arb<Jr: University of Michigan Press. Dahl, Robert. 1954. Modem Political Analysis. Englewood Cliffs, Nj: Prentice-Hall. Hale, Robert L. 1952. Freedom Through Law. New York: Columbia University Press. Hart, Herbert L A. 1961. The Concept of Law. Oxford: Clarendon Press. Hart, Herbert L.A., and A.M. Honore. 1959. Causation in the Law. Oxford: Clarendon Press. Hempel, Carl G. 1951. Fundamentals of Concept Fom:ation i11 Empirical Science. Chicago: University of Chicago Press. Laswell, Harold, and Abraham Kaplan. 1950. Power and Society. New Haven: Yale University Press. Luce, R. Duncan, and Howard Raiffa. 1957. Games and
Decisions. New York: john Wiley and Sons. Nagel, Ernest. 1961. The Structure of Science. New York: Harcourt, Brace, and World. Olson, M.ancur. 1965. The Logic of Colleaive Action. Cambridge, M.A: Harvard University Press. Oppenheim, felix. 1961. DimetiSions of Freedom. New York: St. Martin's Press. Schelling, Thomas. 1960. The Strategy of Conflict. Cambridge, MA: Harvard University Press. Suppes, Patrick, and j. L Zinnes. 1963. "Basic Measurement Theory." ln R. D. Luce, R. Bush, and E. Galenter, eds., Handbook of Mathentatical Psychology, vol. I. New York: john Wiley and Sons. 2. Newcomb's Problem and Two Principles of Choice It is nor dear that I am enr:irled ro pre-sent this paper. For the problem of choice which concerns me was constructed by someone else, and I am not satisfied with my attempts to work through the problem. But since I believe that the problem will interest and intrigue Peter Hempel and his many friends, and since its publication may call forth a solution which will enable me to stop returo.ing, periodically, to it, here it is. It was constructed by a physicist, Dr. William Newcomb, of the Livermore Radiation Laboratories in California. I first heard
the problem, in 1963, from his friend Professor Martin David Kruskal of the Princeton University Department of Astrophysical Sciene<s. I have benefitted from discussions, in 1963, with William Newcomb, Martin David Kruskal, and Copyrighted material 346 ~ N ores and References Paul Benacerraf. Since then, on and off, I have discussed the problem with many other friends whose attempts to grapple with it have encouraged me to publish my own. It is a beautiful problem. I wish it were mine. I. If the being predicts that you will consciously randomize your choice, e.g., flip a coin, or decide to do one of the actions if the next object you happen to see is blue, and otherwise do the other action, then he does not put the $M io the second box. 2. Try it on your friends or students and see for yourself. Perhaps some psychologists will investigate whether responses to the problem arc correlated with some other interesting psychological variable that they know of. 3. If the questions and problems are handled as I be~eve they should be, then some of the ensuing discussion would have to be formulated differently. But there is no point to introducing d.rail extraneous to the central ptoblem of tbi.s paper here. 4. This divergence between the dominance principle and the
expected utility principle is pointed out in Robert Nozick, The Nomwtive Theory of lndivid .. al Choice (doctoral dissertation, Princeton University, Princeton, 1963; published in 1990 by Garland Press, New York), and in Richard jeffrey, The Logic of Decision (McGraw-Hill, New York, 1965). 5. This is shor.thaod for: action A is done and state S12 obtains or acrion 8 is done and state S1 obtains. The 'or' is the exclusive or. 6. Note that s, =A, & S3 or Az & s. S2 = A, & S4 or Az & S.1 S3 =A 1 & S, or A2 & Sz s. = A 1 & Sz or Az & S1 Similarly, the above identities hold for Newcomb's example, with which I be gan, if one lets S 1 = The money is in the second box. Sz = The money is not in the second box. S3 = The being predicts your choice correctly. s. =The being incorrectly predicts your choice. A 1 = You take only what is in the second box. A2 = You take what is in both boxes. 7. StateS is not probabilistically independent of actions A and 8 if prob (S obtains lA is done) "# prob (S obtains! 8 is done). Copyrighted material Notes and References ~ 347 8. In Newcomb's predictor example, assuming that "He predicts correccly" and "He predicts incorrectly" are each probabilil;tically independent of m)'
actions, then it is not the case that "He puts the money in'" and "He does not put the money in" are each probabilisrically independent of my actions. Usually it will be the case that if the members of the set of exhaustive and exclusive states nre each probabilistically independent of the actions A, and Az, tben it will not be the case that the stares e.quivalent to our contrived states nre each probabilistically independent of both A 1 and Az. For example, suppose prob (S,(A,) ~ prob (StiAz) ~ prob (S,); prob (SJ.IAz) ~ prob (SziA t) ~ prob (Sz). Let: S3 =At & S, or Az & Sz s. ~ A1 & Sz or Az & s, If prob (S1) # prob (Sz), then S3 and s. are not probabilisricaUy independent of At and Az. For prob (SJIAt) = prob (S.JA,) ~ prob (S,), and prob (SJIAz) = prob (Sz iAz) ~ prob (Sz). Therefore if prob (S,) prob (Sz), then prob (SJIAt) # prob (SJIAz). tf prob (S,) ~ prob (S1) ~ 1/2, then the possibility of describing the states as we have will not matter. For if, for example, A1 can be shifted around so as to dominate Az, then before the shifting it will have a higher expected utility than Az. Generally, if the members of the set of exclusive and exhaustive states are probabilistically independent of both A 1 and Az, then the members of the contrived set of states will be
probabilistically independent of both A 1 and A z only if the probabilities of the original stares which are components of the contrived states are identical. And in this case it will not matter which way one sets up the situation. 9. Not< that this procedure seems to work quite well for situations io which the stat~'< are not only not probabilistically independeot of the actions, but are not logically independent either. Suppose that a person is asked whether he prefers doing A to doing 8 , where the outcome of A is IP if S1 and r if Szl and the outcome of 8 is Jq if Sz and r if S11. And suppose that he prefers p to q tor, and that S1 ~ I do 8 , and Sz ~ I do A. The person realizes that if he does A, Sz will be the case and the outcome will be r, and he realizes that if he does 8, S, will be the case and the outcome will be r. Since the outwmc will be r in any case, he is indifferent berwccn doing A and doing 8. So let us suppose be flips a coin in order to decide which ro do. Bur given thar the coin is fair, it is now the ctse that the probability of S1 ~ 1/ 2 and the probability of Sz = 1/ 2. If we mechanic31ly starred co compute rhe e.xpected utility of A, and of 8, we would find that A has a higher expet"ted utility than does 8 . For mechanically computing the expected utilities, ir would turn out that the expected utility
of A= 1/2 x u(p) + 1/2 x u(r), and the expected utility of 8 ~ 1/2 x u(q) + 1/2 x u(r ). If, however, we use the conditional probabilities, then the expected utility of A~ prob (S1 IA) x u(p) + prob (SziA) x 11(r) ~ 0 x 11(p) + 1 x u (r) ~ u(r). And the expected utility of 8 ~ prob (Szi8) x u(q) + prob (StiB) x u(r) ~ Copyrighted material 348 -& Notes and References 0 x u(q) + 1 >< u(r) = u(r). Thus the expected utilities of A and 8 are equal, as one would wish. 10. This position was .suggested, with some re,servations due to Newcomb's example, in Nozick, The Normative Theory of lt~dividuGI Choice. It was also suggested in jeffrey, The Logic of Decision. 11. I should mention, what the reader has no doubt noticed, that the previous exGmple is not fully satisfactory. For it seems that preferring the academic life to the athlete's life should be as strong evidence for the tendency as is choosing the academic life. And hence P's choosing the athlete's life, though he prefers the academic life, on expecred-urility grounds does not seem to make it likely that he does not have the tendency. What the example seems to require is an inherired tendency ro decide to do A which is such that (I) The probabiliry of its presence cannot be estimated on the basis of the person's preferences, but
only on rhe basis of knowing the genetic makeup of his parents, or knowing his actual decisions; and (2) The theory about how the tendency operates yields the result that it is unlikely that it is present if the person decides nor ro do A in the example-situation, even though he makes his decision on the basis of the stated expeetedutility grounds. It is not clear how, for this example, the details are to be coherently worked out. 12. Thar is, rhc dominance principle is legirimately applicable to situations in whi<:h "'(3S)(3A)(3B) [prob (S obminsiA is done) ;" prob (S obtainsiB is done)). 13. The other eleven possibilities about the srares are: Already fixed Not already fixed - and determined and determined Probabilisti Not probabilisti Probabilisti Not probabilisti cally inde- caUy inde- cally inde cally independent of pendent of pendent of pendent of the actions rhe actions the actions the actions (1 ) some some some some (2) .some some some none (3) some some none some (4) some some none none (5) some none some some (6) some none some none (7) some none none some (8) all none none none (9) none some some some (10) none some some none (11) none some none some
Copyrighted material Notes and References ~ 349 14. Unless it is possible that there be causality or influence backwards in time. I shall not here consider this possibility, though it may be that only on its basis can one defend, for some choice situations, the refusal to use the domjnan.ce principle. I try to explain later why, for some situations, even if one grants that there is no influence back in time, one may not e.scape the feeling that, somehow, there is. 15. Cf. R. Duncan Luce and Howard Raiffa, Games and Decisions Uoho Wiley & Sons, New York,.1957), pp. 94-102. 16. Almost certainty1 > almost cerraioty2, since almost cenainty2 is some function of the probability that brother I has the dominant-action gene given that he performs the dominant action (= almost certaintyJl, and of the proba bility that brother II does the dominant action given that he has the dominant action gene. 17. In choosing the headings for the rows, I have ignored more complicated possibilities, which must be investigated for a fuller theory, e.g., some actions influence which state obtains and others do not. 18. I here consider only the case of two actions. Obvious and messy prob lems for the kind of policy about to be proposed are raised by the situation in
which more than two actions are available (e.g., under what conditions do pairwise comparisons lead to a linear order), whose consideration is best postponed for another occasion. 19. See Luce and Raiffa, Games and Decisions, pp. 275- 298 and the ref erences ther<in; Daniel Ellsberg, "Risk, Ambiguity, and the Savage Axioms," QUJJrterly journal of Economics 75 (1961), 643-669, and the articles by his fellow symposiasts Howard Raiffa and William Feller. 20. If the distinctions I have drawn arc correct, then some of the existing lit erature is in need of revision. Many of the writers might be willing to just draw the distinctions we have adumbrated. But for the specific theories offered by some personal probability theorists, it is not clear how this is to be done. For example, L. J. Savage, in The Foundations of Statistics (john Wiley & Sons, New York, 1954), recommends unrestricted use of dominance principles (his postulate P2), which would not do in case (1). And Savage seems explicitly to wish to deny himself the means of distinguishing case (I) from the others. (For further discussion, some of which must be revised in rhe light of this paper, of Savage's important and ingenious work, see Nozick, The Normative Theory of Individual Choice, chap. v.) And Jeffrey, in The Logic of Decision, recommends
universal use of maximizing expected utiliry relative to the conditional proba biliries of the states given the actions (sec note 10, above). Thjs will not do, I have argued, in cases (III) and (IV). But Jeffrey also sees it as a special virtue of this theory that it does not utilize cenaio notions, and these notions look like they might well be required to draw the distinctions between the different kinds of cases. Copyrighted material 350 ~ Notes and References While on the subject of how to distinguish the cases, let me (be the first to) say that I have used without explanation, and in this essay often interchange ably, the notions of influencing, affe<:ting, etc. I have felt free to use them with out paying tbem much ar.tenrion because even such unreflective use serves to open a whole area of concern. A detailed co11sideration of the different possible cases with many actions, some influencing, and in diffecem degrees, some not influencing, combined with an attempt w state derailed principles using precise "influence" notions undoubtedly would bring fonh many intricate and difficult problems. These would show, I th ink, that my quick general statements about influence and what distinguishes the cases are not, scric!ly speaking, correct. But going into these details would ncce.~sitate going into these
details. So I wiU not. 21. Though perhaps it explains why I momentarily felt I had succeeded too well in constructing the vaccine case, and that perhaps one should perform the nondominant action there. 22. Bur it also seems relevant that in Newcomb's example nor only is the action referred to in the explanation of which state obtains (though in a nonex tensional belief context), bur also there is another explanatory rie between the action and the state; namely, that lx>th the state's obtaining and your actuaUy performing the action are both partly explained in terms of some third thing (your being in a cerrajn irutial state earlier). A fuUer investigation would have to pursue yet more complicated examples which incorporated ttus. 4. Interpersonal Utility Theory During 1981- 82, I presented these ideas in talks at Chicago, Michigan, Oberlin, Princeton, Sraniord, and Yale. I have benefitted from the comments I received then as well as from comments of the referees. friedman, M., and L J. Savage. I 948. "The Utility Analysis of Choices Involving Risk." joumal of Political Economy 56:279-304. Homans, G. 1974. Social Behauior, rev. e.d. New York: Harcourt Brace. Markowitz, H. "1952. "The Utility of Wealth." joumal of Political EcottOnry 60: 151-158.
Nozick, R. 1981. Philosophical Explanation.$. Cambridge, MA: Harvard University Press. Sen, A. 1979. " lnrerpersonal Comparisons of Welfare." In M. Soskin {ed.), Economics and Human Welfare, pp. 183-201. New York: AcadeoUc Press. Tversky, A., and D. Kabneman. 1979. "Prospect Theory. Econometrica 47: 263-291. 5. On Austrian Methodology 1. This essay was originaUy directed to an audience of nonphilosophers, and as a result some of the philosophical poims will not be news to readers with a Copyrighted material Notes and References ~ 351 background in philosophy. I have retained these points in the hope that some economists will find them illuminating and that even philosophers might find their application interesting. I have benefitted from discussion of the essay at the New York Seminar on Austrian Economics, the Society for Ethical and Legal Philosophy, and from a discussion of section IV with Richard Herrnstein. The major writings in the Austrian tradition are Carl Menger, Pri11ciples of Economics, Problems of Economics a11d Sociology; Eugen von B<ihm-Bawerk, History and Critique of Interest Theories, The Positive Theory of Capital; Ludwig von Mises, The Theory of Mo11ey a tid Credit, Huma11
Action; Frederick Hayek, Itldividualism and Economic Order, The Pure Theory of Capital, Prices and Production, Mo<etary Theory and the Trade Cycle. See also Ludwig von Mises, Socialism; Murray Rothbard, Man, Economy and State; Ludwig Lachman, Capital a11d Its StmctYre; MacroEconomic Thinking and the Market Economy; and Israel Kinner, The Economic Point of Vitw, Market Theory atld the Price System, A11 ss;ry on Capital, and his recent and imporranr Competition and E:ntrepreneurship. Not every figure accepts or places equal emphasis on e.ach of the features of the methodological framework. For example, while all pans of this essay apply to Mises (and to Rothbard), only section I is cle.arly relevant ro Hayek's views. 2. See F. A. Hayek, The Cormterrevolutio11 of Science (The Free Press, Glencoe, Ill., 1955), chaps. 1- 10, and Individualism a11d Eco11omic Order (Univer sity of Chicago Press, 1948), chaps. 2-4. For a sample of discussion in the journals, see the six essays by Watkins, Mandelbaum, and G<lldstein (rwo apiece) reprinted in Leonard Krimmerman (ed.), The Nature and Scope of Social Science (Appleton, Century, Crofts, New York, 1969). A srandard account of reduction is provided in Ernest Nagel, The Structure of Scitmce (Harcourt Brace, and World, New York, 1961), chap. 11. For complications
and refinements, see also Lawrence Sklar, "Types of lnterTbcorcric Reduction," British Journal for the Philosophy of Scimce 18 (1967), !09-124. 3. I leave aside the question of how much of such decision theory as has been developed should be acceptable within Austrian theory, though I do believe that much of it is, and that mistak~n views about the rdationship of preference to choice and action (see section 01, below) have led Austrian writers to ignore this literature. Indeed, one wouJd expect that writings on personal or subjec tive probabiliry would be most congenial to Austrians, the major proponents of the subjective theory of value. It is puttling that Austria.n writers have been concerned solely wirb "objective" probability, since there is no guarantee that an actor will act on those ratios or limits of relative frequencies. Only if such ratios are reflected in subjec.tive probabilities will a theory of human action be concerned with them, and if they are 1101 so reflected, the theory will have anyway to be concerned with the (divergent) subjective probabilities. On personal (subjective) probability, see Leonard J. Savage, The Found4tions of Statistics (John Wiley and Sons, New York, I 954), and F. J. Anscom be and R. J. Aumann, Copyrighted material 352 -<> Notu and References
A Definition of Subjtctivr Probability," Annals of Mathematical Statistics 34 (1963 ), 199-205. 4. I ignore here the issue of whether the situatiotr can be specified by utilizing only the concepts of Crusoe theory. 5. See R. D. Luce and H. Raiffa, Games and Decisions Uohn Wiley and Sons, New York, 1957); Thomas Schelling's treatment of coordination games in The Strategy of Conflict (Oxford University Press, 1963); john Harsaoyi, "A General Theory of Rational Behavior in Game Situations," Econometrica 34 (1966), 613-634; and "Advances in Understanding Rational Behavior," Working Paper CP-366 of the Center for Research in Management Science, University of California, Berkeley, July 1975, and the papers by Harsanyi cited therein. 6. See Luce and Raiffa, Games and De<isions, chap. 9. 7. The nonmethodological individualist might agree that the explanatory theory of n-person interactions suffices to explain all interactions, for he might believe that the ultimate explanation of n-person interactions lies in a general theory of multiperson ( > n) interactions. (I owe this preceding remark and the next paragraph to David Hills.) The methodological individualist, however, believes that the general theory of multiperson interaction won't make essential
reference to more than n persons ln that it will be nonvacuously true in some domain of no more than n individuals. Furthermore, the theory where n > 1 will completely satisfy methodological individualist strictures only if it does not anribute utility or reasoning to dyads or groups in a way which is irreducible ro individual utility functions and individual reasoning. It will nor countenance an irreducible g.roup mind, utility function, ere. (It is not obvious how to specify this "etc.") 8. See Richard B. McKenzie and Gordon Tullock, The New World of Economics (Richard D. Irwin, Inc., Homewood, Ill., 1975), chap. 18, "Riots and Panic," for a methodological individualist account of such behavior. 9. On emergence, see Nagel, Structure of Science, pp. 366-380; Carl Hempel, AspeciS of Scientipc xplatration (Free Press, New York, 1965), pp. 258-264. 10. Menger, Pri,ciples of Economics, chap. VID. 11. Mises holds that his money-regression theory demonstrates that no other origin of money is possible. "It follows that an object cannot be used as money unless, at the moment when its use as money begins, it already possesses an objc: aive exchange-value based on some other use. This provides both a refutation of those theories which derive the origin of money from a general agreement
to impute fictitious value ro things intrinsically valueless [here Mises foomores Locke I and a confirmation of Menger's hypothesis concerning rhe origin of rhe use of money" (The Theory of Momry and Credit, enlarged edition, Yale UniCopyrighted material Notes and References ~ 353 versity Press, 1953, p. 110). The prices tomorrow depend upon the amount spent on consumption, saved, and held in current cash balance. What is held in cash balance will depend upon expectations about pricS tomorrow (the very ones to be explained) and prices the next day. It is not, strictly, circular to explain prices on a given day in terms of expectations about prices for that very day, bur one might well want to avoid this, especially if one contemplates the possibility that aU such expectations are formed in accordance with (and only in accordance with) the very theory being developed. Mises stops the (unstrict) regress by having today's prices determining tomorrow's cash balances. Day by day we go back in rime until the point when it is only because of use value, and not exchange value, rhat something is held. Thereby, all prices are noncircularly e.xplained. Mises is able to introduce roday's prices as determining tomorrow's cash balances only if either (a) people expect tomorrow's prices to be
(roughly) like today's or (b) people arrive at their expectation of tomorrow's prices by staning with knowledge of today's prices, plus their knowledge of the causal factors operating, and thereby arrive at some view of how tOmorrow's prices will differ from today's, and so of whar tomorrow's prices will be. In each case, it is an expectation about tomorrow's prices which is crucial to determining the cash holdings. Mises's moncy-n:grcssion argumtnt shows one way such txpectarions may function. Howrver, it cannot demonstrate that Menger's account describes the only way money could arise, for it cannot demonstrate that expectations about tomorrow's prices can only be formed on the basis of today's prices. So, in panicular, it cannot show that a social con traer co11ld 110t actually give rise to (roughly) uniform expectations about the next day's prices. To demonstrate that, another argument would be needed. 12. Economics provides the prime arena of such explanations as nave thus far been offered. George Homans presents illuminating examples of methodological individualist explanations in sociology in his Social Behavior, Revised edition (Harcourt, Brace, jovanovich, Inc., New York, 1974), and explicitly endorses methodological individualism in The Nat1tre of Social Science (Harcourt,
Brace, New York, 1967). 13. But things are more complicated if the explanation of the subjunctives and act tokens e=ntially refers to previously holding subjunctives as well as to previous act tokens. 14. Different theoristS have held that capi(alist institutions are seJ(~ destructive, notably Marx and Schumpetcr. Most recwtly, Daniel Bell has advanced such claims in The Cultural Contradictions of Capitalism (Basic Books, New York, 1975). 15. See E. 0. Wilson, Sociobiology (Harvard University Press, Cambridge., Mass., 1975). Copyrighted material 354 ~ Notes and References 16. Some writers speak not merely as if different psychological types inhabit rhe different historical stages of human society, but as if different psychological laws function in these different stages. I find this far-fetched, but if it were true, and if the general law connecting the historical stages with the operative psy chological laws was a brute-macro-law, unexplainable by any deeper psychological law as it operated in different social circumsrances, then methodological individualism would be false. However, the explanations of behavior within each stage and of the transition of society from one stage to another might still
be methodological individualist explanations which usc the psychological laws of the relevant stage. What would not get a methodological individualist answer would be the question "why those laws?" 17. For a recent survey of the current state of the theory, see Roger Brown and Richard Herrnstein, Psychology (Linle, Brown, and Co., Boston, 1975), chaps. 1-3. I shall say something more about operanr conditioning below. 18. Mises himself holds that " leisure is a consumer's good" and " labor has disutility., are not necessary truths, and hence explanations which utilize these statements are 1101 claimed to be reducible to a completely a priori theory. 19. An illuminating survey of difficultie.s with the positivist arguments on synthetic necessary truths, by a defender of such truths, is Arthur Pap's Semantics and Necessary Tnuh (Yale University Press, 1958). The most influential recent critic of ne<:essiry and of the analytic-synthetic distinction is W. V. Quine. See his essay .. Two Dogmas of Empiricism" in From a Logical Point of View (Harvard University Press, 1953), his book \llord and Object (MIT Press, I 960), and his essay "Necessary Truth" in his volume of essays The \Vays of Paradox (Random House, New York, 1966). The most influential and ingenious recent defense and utilization of synthetic ne<:essary truths, and
of essences, ls Saul Kripke's mooog.raph, '1Naming and Necessity," in DonaJd Davidson and Gilbert Harman (eds.), Semantics of Naturall.anguage (0. Reidel, Humanities, New York, 1972). 20. Anarchy, State, and Utopia (Basic Books, New York, 1974), pp. 42-45, "The Experience Machine." 21. A survey is contained in B. F. Skinner, Science and Human Behavior (Macmillan Compan)', New York, 1953). A precise quantitative form of the law of effect is presented in R. J. Hermstein, "On the Law of Effect," Journal for the Experimental Analysis of Behavior 13 (1970), 243- 266, and "Quantitative Hedonism," Journal of Psychiatric Research 8 (I 971), 399-412. Iris often erroneously believed that Noam Chomsky's famous review essa)' of Skinner's book, Verbal Behavior-in Language 35 (1959), 26-58, reprinted in j . j . Ka11. and Jerry Fodor, The Structure of Language (Prentice-Hall, 1964)refutes Skinner. s gene.ral theory or at le.ast its application to human ~ings. However, many of Chomsky's criticisms deal with earlier stimulus-response theories, theories of drive reduction, ere., and not with Skinner's rheory of operant conditioning Copyrighted material Notes and References -& 355 (which is not an 5--R theory), and Chomsky's complainrs about
the circularity of the law of effect are mistaken. (S..,. Paul Meehl, "On the Circularity of the Law of Effect," Psychological Bulletin, 1950; Kenneth MacCorquodale. "On Chomsky's Review of Skinner's Verbal Behavior," joumal of the Experimental Analysis of Behavior 13 ( 1970), 83-99; and the articles by Herrnstein cited above_) Even the daim that an act is self-rein forcing has the consequence (as M~hl pointed out) that successive performances will increase its strength and that artificial duplication of its proprioceptive or brain effects could be used to reinforce other actions. Nor is it empty to claim that exploratory behavior is itself reinforcing, for not only is this claim plausible on evolutionary grounds, but tests could determine whether being given opportunity to engage in exploratory behavior itself functions as reinforcement. One of Chomsky's criticisms has carried great weight with many, namely that (even if the law of effect itself is not circular and empty of content) many of the applications of the law of effect are circular. For in many applications, the reason for believing a reinforcing event ha.s taken place is the very cfft:et it is to explain. Consider another case. A bridge collapses, and the question of why it collapsed is raised. Engineers hypothesize that the stress in a particular place
reached a certain point. They be~eve this occurred, because they have a theory, well confirmed elsewhere, that this causes bridges to collapse and is the most frequent cause of bridge c.ollapses. They then go on to explain the collapse of the bridge by referring to the stress. The collapse of the bridge is their reason for believing there was a certain stress, and the srress is their explanation of the bridge's collapse. Isn't this circular? It is not, for the two italicized terms in the previous sentence are different. It would be circular if they were the same, that is, if either (a) the collapse of the bridge is their reason for believing there was a certain stress, and the stress is their reason for believing the bridge did collapse., or {b) the collapse of the bridge is their explanation of the stress, and the stress is their explanation of the collapse of the bridge. But in the bridge example we are discussing, there is no such objectionable circularity. (See jaegwon Kim, "On Inference, Explanation and Prediction," joumal of Philosophy 61 (1964), 360-368.) The situation is similar with reinforcement. Given a well-confirmed theory, it will be reasonable ro infer some event of a type known to be reinforcing rook place, or to infer that a type of event which was known to take place is reinforcing. One would want to go ahead and check
this latter claim, but there seem to be no insuperable dif6cultie$ here. Thus, this criticism by Chomsky rests upon a mistaken methodological view. The theory of operant conditioning cannot be dismissed so easily. 22. For a discussion of ways in which empirical procedures come into the discovery of necessary (though not a priori) truths, sec Kripke, "Naming and Necessity. 23. On verstehen, see F. A. Hayek, The Counterrevolution of Science. Copyrighted material 356 -& Nores and References 24. See Ernest Nagel, The Struct..re of Science, pp. 480-485. 25. W. V. Quine, Word and Obiect, chap. 2. Putnam's essay considers the relevance of verstehen for the issues Quine discusses about translation, and so links up rhe rwo areas, but ir does nor go on ro rrear radical verstehett and irs implications for the empirical character of social science. 26. Or, if we view empathetic understanding as assigning prior probabilities, after a large number of cases we will be able to assess irs reliability as a prior probability assigner. 27. This scheme is nor identical, however, ro Mises' a priori scheme. For most people do nor think they always act ro reduce rhdr own felt uneasiness, ere. Hence, if Mises is right, rhese people's empathetic understanding of their ow11 behavior is sometimes fa ulty.
28. See Rothbard, Man, Economy, and State, p. 64. 29. Mises writes in Human Aaion: "However, one must not forget that the scale of values or wants manifests itself only in the reality of action. These scales have no independent existence apart from the actual behavior of individuals" (p. 95); "The scale of value manifests itself only in real acting; ir can be dis cerned only from the observation of real acting. It is therefore impermissible ro contrast ir with rea l acting and ro use it as a yardstick for rhe appraisal of real actions" (p. 102). 30. Perhaps Mises is denying rhis poinr in the middle of the lasr paragraph of page 119 of Human Action, but it is difficult tO be sure. However, on p. 122 he says," All parts-units-of the available stock are c.onsidered as equally useful and valuable if the problem of giving up one of them is raised." Here, then, we do have indif{erellc.e. Yet a choice will be made, perhaps at random. One particular object will be given up. Yet the person d~s not prefer giving up this one to giving up another one. Therefore, choice entails (at best) weak preference; it does not entail strong preference. Rothbard claims (Man, Economy, and State, p. 265), "Any action demon stratts choice based on preference; preference for one alternative ove.r others. There is, therefore, no role for the concept of indifference in
economics or in any other praxeological science." However, (on pp. 18-19) he also writes, "in these examples, the units of the good have been interchangeable from the point of view of the acJor. Thus, any concrete pound of butter was evaluated in this case perfectly equally with any other pound of butter" (his italics the fin."t time, mine the second), and he continues on in the same vein on rhe rest of p. 19. 31. Mises, Human Action, p. 97. 32. Can't we consider the cost of A as tbe value of choosing among the re maining alternatives, making the implicit assumption that the value of choosing among aU of the remaining alternatives is equal to the value of the best alter native among them? Notice, though, that this involves introducing a choice Copyrighted material N ores and References -& 3 S 7 situation (choosing among jt~st those) which the person never acrually faces. And making the implicit assumption explicit would reraise the problem. 33. See David Lewis, Counterfactuals (Harvard University Press, Cambridge, Mass., 1973). I am not committed here, however, to the adequacy of Lewis's account, wbjch in any case I describe only roughly in the text. Furthermore, since the actual world is of course the world closest to the actual world,
Lewis's account bas the unfortunate consequence of making subjunctively true any conditional with true antecedent and consequent. We can avoid thjs con sequence by the following proposal, which has other desirable consequenc<s as well. Let the p-ne.ighborhood of a world W be the closest band of worlds where p is true, uninterrupted by a world in which not-p holds. More precisely, if p is true in world W, the p-neighborhood of W is the set of those worlds w such that p is rrue in w and thert is no world w' in which not-p is true whkh is at least as close co W as w is. If p is false in W, the p -neighborhood of IV is the set of rhose worlds w such that p is true in w and there is no world w' in which not-p is true which is the same distance from W as w is, and for any world w" which is between w and W and in which not-p is rrue, there is no world in which p is true which is as close co W as w" is. (The condition where p is false in W says that the p-neighborbood of W is the set of those worlds w where p is true, such that no notp world is equidistant from W) and such that any notp world between wand W is in the not-p neighborhood of W. ) However, when p is false, if the closest p-worlds to W are the same distance from W as are some not-p worlds, then a more complicated account of the relevant band must be
sven. Finally, we can say that a subjunctive with antecedent p and consequent q is true if q is true in each world of the p-neighborhood of the actual world. 34. For an account of dispositions which views them as placeholders for an underlying physical srrucrure, see W. V. Quine, The Roots of Referettce (Open Court, La Salle, Ill., 1973), pp. 8- 12. See also Isaac Levi and Sidney Morgenbesser, Belief and Disposition," American Philosophical Quarterly 1 (1964), 1-12. 35. The requirement that preference be transitive shouJd no t be rtad as,. "if X is preferred toY, andY is preferred to Z, then X should be preferred to Z," so as to license an inference to the constqucnt if tht anttctdent is satisfied. Perhaps it's the case that X shouldn't be preferre<lto Y. Also, if the person prefers X to Y, Y to Z, and Z to X (and so has intransitive preferences), three applications of the principle, read as above, would lead to the conclusion that he should prefer X to Z, Z toY, andY to X-another inrransitive triad! Instead, the requirement should be read as requiring preferences to hang together in a cerrain way, and as excluding preferences from combining in other ways, e.g., "it shouldn't be the case that X is preferred to Y, Y to Z, yet X isn't preferred
Copyrighted material 358 -~> Notes and References to Z." See Roben Nozick, The Nonnative Theory of Individ11al Choice (doctoral dissertation, Princeton University, 196-l), pp. 94- 98 (published in 1990 by GaJiand Press, New York). 36. Caution; see my re.marks below about me connection of preferences wim the subjunctives. 37. For a recent survey of the theory of choice sets, the theory of conditions such as these, see A. K. Sen, "Choice Functions and Revealed Preference," Rer,;ew of Economic Studies 38 (1971), 307-317. 38. As an antidote to Rombard's statement {Man, Econom)l and State, p. 15), " It is importanr to realize rhat there is never any possibility of meas11ring increases or decreases in happiness or sarjsfaction ... In order for any measurement to be possible, rhere must be an eternally fixed and objectively given unit with which other units may be compartd," readers may wish to .see D. Krantz, R. 0. Luce, P. Suppes, and A. Tversky, Fo11ndations of Meas~~rement, vol. 1 {Academic Press, New York, 1971). 39. This type of point will be familiar 10 readers of W. V. Quine and Donald Davidson. 40. See, for example, Mark Blaug, Economic Theory in Retrospect (Irwin & Company, Homewood, Ill., 1968), revised edition, pp. 503-
509; Robert E. Kuerme, Eugetr votr 8ohm-8awerk (Columbia Universiry Press, New York, 1971), pp. 25-34. 41. This can happen only if the person at that moment is indifferent for all things he can get between getting them then and getting them later. Otherwise, he'll do one of the (omer) time-preferred things. 42. Those who mink time-preference is a categorial component of action may think it strange ro attempt, as I do later, an evolurionary explanation of ir, and compare this ro an evolutionary explanation of preference, another (supposed) categorial component of action. However, neither is inappropriate. Since it is not a necessary num mat mere are preferences or timepreferences (or actions!), the world need not have contained them, their existence is a proper topic for explanation, and the correct explanation presumably will be some evolutionary one. 43. See Mises, Humatr Action, pp. 131-'132; Rothbard, Man, cmromy, and State, PP- 37-38. 44. That consumption takes time, and may be incompatible with other activiries, including other consumption, is put to differ<nt economic use by Staffan Linder, The Harried Leisure Class (Columbia Uruvcrsity Press, 1970). 45. I take these details from George Ainslie, "Specious Reward:
A Behavioral Theory of Impulsiveness and Impulse Control," Psychological 8ulleti11 81 {1975), 463-496, who also discusses in an illwrunating way various devices for getting past rhc earlier smaller reward without choosing it. Copyrighted material Notes and References ~ 359 46. The relevant que.stion is not what you actually wou.ld choose in tbe future if given the choice then (for your preferences might change), but bow valuable, according to your preferences now, getting it then will be wonh to you then. It is not clear how to capture this by a subjunctive about choices. 6. Socratic Puzzles This essay was stimulated by questions posed by Gregory Vlasros in his Socrates: lronist and Moral Philosopher (Cornell University Press, Ithaca, 1991), and its title derives from Vlastos's speaking, in his Plator~ic Studies (Princeton University Press, 1973), of cracking puzzles in Plato. Aher writing an earlier drah with tbe current title, I encountered Terence Irwin's review of Vlastos's Socrates, also entitled "Socratic Puzzles" (Oxford Studies in Anciellt Philosophy, volume 10, 1992). I hope Irwin will accept my reusing his title as a reinforcement of his tribute to Vlastos. I am grateful to Myles Burnye.at for his c.omments on an earlier draft of this
essay. 1. Whether or not this doctrine is an accurate description of everyone, it is a correct description of Socrates himself; it is inconceivable that Socrates would know what is best yet do otherwise. However, we should nor abandon Socrates too soon to the familiar objection that someone migbt know the good but be roo weak of will ro carry ir out. For what explains weakness of will? If that phenomenon itself arises due to, and cruciaJly involves, some particular ignorance on the agent's part, then Socrates' doctrine will not fall to the akrasia objection. Consider the psychologist George Ainslie's explanation of weakness of will in terms of timepreference, aod the crossing curves these can involve. (George Ainslie, "Specious Reward: A Behavioral Theory of Impulsiveness and Impulse Control," Psychological Bulletin 82 (1975), pp. 463~96. I first discussed this explanation in "On Austrian Methodology" (1977), in this volume as chapter 5, and I returned to it in The Nature of Rationality, Princeton University Press, 1993, pp. 14-18.) On this explanation, an akratic act might involve a failure of knowledge. Because of visual perspective, a nearer object can look larger than a mo.re distant one that actually is larger (where the actual size of the o bject is
measured at the object). Someone who thinks the nearer object larger makes a mistake of knowledge. ls time-preference like visual perspective, so chat rewards look smaller, the more distant they are in the future? If they thereby look smaller than they actually are (where their actual magnitude is measured at the reward), then isn't someone making a mistake, a cognitive mistake, who acts as if the nearer reward is the larger one? lsn,t this an illusion, akin to a visual illusion? (See Protagoras, 356c-7e.) I don't insist that the answer is dear, in this case. And Ainslie's explanation of akrasU. is not the only possible one. Still, we had better be sure that we possess an account of weakness of will that does not Copyrighted material 360 -<> Notes and References attribute any failure of kt10wledge to the agent before we contend that Socrates is obviously mistaken in holding that no one does wrong (or chooses the worse) voluntarily. 2. Vlastos, Socrates, p. 3. 3. Plato's Moral Theory, pp. 39-40. 4. About something where he hasn't claimed knowledge of the first kind, he might also be denying knowledge of both kinds. See Gregory Vlastos, "Socrates' Disavowal of Knowledge," Philosophical Qt<arterly 35 (1985), pp. 1-31.
5. Euthyphro's prosecution of his fa ther for impiery is shocking. Notice, howeve~ that if Euthypbro's futber did commit murder or otherwise do wrong in forgening about the murderer slave left tied in a field while the authorities were sent for, then according to Socrates' doctrine that a wrongdoer is better off being punished, the father will be bener off being punished for his act. 6. Vlastos, Socrates, pp. 174-176. 7. Such ao account would present necessary and sufficient conditions for piety, stating what is (nonaccidemaUy) common to all and only pious actions, perhaps in a way that provides a standard by which we can judge the piety of any act. So must the concepts F be recursive? Would a recursively enumerable conupr fail r:o provide the requisite standard? 8. See Robert No1jck, Philosophical Explanations (Harvard University Press, Cambridge, Mass., 1981), pp. 1n-196, and Steven luperFoy, The Possibility of Knowledge: Essays 011 Nozick's Theory of Knowledge (Rowman and Littlefield, N.j., 1987). 9. We need nor attribute to Socrate.s the very strong assumption that something is good only if it is good (and benefits one) in every possible context. Even a knowledge of the good cannot pass such a rest. Consider the context in which if you do come to have knowledge of the good, then
beings on alpha ccntauri will destroy your own mental capacities, or will destroy all life on c.arth. Socrates need only claim that something is good only if it is good in all normal contexts, whether or not there is some extraordinary context in which it leads 10 disaster, and that the ignorance of the good Socrates speaks of can lead to disaster in a normal context. 10. Vla<Os, "The Socratic Elenchus," Oxford Stttdies in Ancient Philosophy 1 (1983), pp. 27- 58. 11. Sec Vlasros, Socrate.s, pp. 113-1 14. See also footnote 44 in Vlastos's "Socrates' Disavowal of Knowledge." 12. Vlasros calls this "the puzzle of the elenchus" in the introduction to Socrates: lronist and Moral Philosopher, p. 15, but in stating only (B) he does not rc,ach all the way to an answer. Copyrighted material Notrs and Refrrences -eo 361 13. And once he denies the true statement p, has he then avoided inconsistency? Or is the person assumed to have other true belief-s that wilt maintain inconsistency with his existent false ones; does he have sufficient true beliefs about the central nature or essence of the concept under investigation, so that any false belief will conflict with one of these? But what stops him from giving up these?
14. The letters naming the three assumptions mark these facts: Vlastos's assumption speaks only of belief, the second one speaks also of knowledge, and the third one speaks of how inconsistencies are re.solved. 15. Or are they reachable but not in any finite number of steps? Should we redust Ltibnitz's notion of infinite analysis and Peirce's of the ideal limit of inquiry? 16. "The Paradox of Socrates" was published in the Queen's Quarterly (Winter 1958) and is reprinted as the Introduction in Gregory Vlastos, ed., The Philosophy of Socrates (Anchor Books, New York, 1971). In Socrates: lronist and Moral Philosopher, p. 44, footnote 82, Vlastos says that in earlier drafts of that volume's chapter on Socratic irony, he maintained the failure-of-love thesis but was convinced it was false (or unnecessary to introduce in explaining Socrates' behavior) by a student in his seminar, Don Adams. Vlastos does not state what reasons convinced him to change his view. The thesis is still worth discussing, I believe, even though Vlastos came no longer to maintain it. 17. But after he states this point in "The Paradox of Socra<es," Vlastos goes on to argue (pp. 15- 16) that the doctrine that virtue is knowledge is mistaken. However, a mistaken doctrine does not indicate a failure of love unless Socrates
was led to th.is particular mistake because it excluded other, more loving ways of saving souls. Vlastos then goes on to state that Socrates' flaw is not just a mistake in doctrine but a failure of love. But he offers no further argument for this. 18. Why is this better? The account of retributive punishment I offered in Philosophical Explanations, pp. 363- 393, might supplement Socrates' view. Even if reuibutive punishment effects no funher change in a person, it is bet ter for that person to be connected (via punishment for wrongs done) to correct values than to be completely disconnected. Best is to act on correct values, second best to repent oneself for flouting them, third best to be (even unwillingly) connected to correct values, and worst is to be completely disconnected. 19. Some of Socrates' tactics seem unfair. A person is asked a question, little suspecting whar lies ahead. He might have answered differently, or expressed hesitation, if he knew which aspects of the issues Socrates was concerned with. Unsuspectingly he answers, and then is pushed, nm to an unwelcome conse quence of what he really does believe, but to an unbelieved consequence of what be literally and uncarefully has said. No wonder that often Socrates' interlocutors were annoyed.
Copyrighted material 362 ~ Notes and References Yet Socrates is not merely a crafty lawyer who craps people into contradictions no matter what they believe or start out saying. For his questioning often is illuminating, and drives a reader to a deeper view even when he disagrees with Socrates' view. 20. Escape would injure the state, and injury is not to be caused. But may it not be caused in (legitimate) self-defense-Socrates did serve as a soldier, after aU- and is not Socrates under unjust attack? By his actions and acceptance of the benefitS of the legal system and the city, Socrates has consented to the city's laws and procedures. How do we know, though, rhe extent of implicit consent or commitment? Apparently Socrates had not consented to carrying out orders to inflict injustices upon others. Suppose I am told I've imp~citly consented to being killed unjustly; mighm 't I answer that my actions and receipt of benefits hadn't said that? How can the detailed content of an agreement, including whatever exception-clauses it contains, be read off actions performed in circumstances that did not exhibit any of the exceptional conditions? And is Socrates trying to kill the city, merely because the city would be destroyed if no one obeyed its edicts in any circumstanc-e,s? Even if a
generalization argument is appropriate here, was "No one should ever obey any edict of the city" really the maxim of Socrates' action? Socrates has received benefits, but is there no limit to what one must do (or accept) in gratitude for benefits received from parents or from ones city? 21. Jt is not a coincidence that Socrates developed this notion of a be.au tifully ordered soul while growing up and living in Athens at the very time the Panhenon was being constructed. Did Phidias's building influence Socrates' view, or was Athenian c.ulture the common cause that influenced them both? 7 . Experience, Theory, and Language I. \'fiord and Obiect, p. 5. 2. We need not worry here whether some scattered sentences might have it, e.g., "red here now." 3. Could a person deductively connect his body of beliefs 8 with experience E by adding to 8 rhe belief: if 8 then E wiU occur? He couldn't formulate his total body of beliefs so as to place it in the antecedent of this conditional; anything he could state in the antecedent would be only a tiny fragment of his beliefs, yielding something much smaller than a total theory as having experi entia( consequences. 4. Does Duhem's thesis have even the consequence that senttnces don't have
some experiential content of their own, even if this does nor exhaust them? To derive this consequence one would have to add to Duhem's thesis that indi vidual statemencs are not disconlirmed by experience, the additional claim that they also are not confirmed by experience. Copyrighted material Not<s and References ~ 363 This is an appropriate place to note that Quine's notion of stimulus mean ing does not delineate separate content, in the sense of Duhcm. For due to a person's assumptions and other current beljefs, much wiJI be in the negative (or affi_rmative) stimuJus meaning of a sentence for him at a rime, which Duhem would hold is not pan of the sen tenet's isolated conttnt. (And wouldn't Quine follow Duhem here?) 5. Compare Quine on dispositions as underlying mechanisms, in The Roots of Reference, pp. I 0-12; "Natural Kinds," 130-138. 6. Hilary Putnam, "The Meaning of 'Meaning,'" in K. Gunderson, ed., Min ruosota Studies in the Philosophy of Science, vol. 7 (Minneapolis: University of Minnesota Press, 1975). 7. As remarks on pp. 37, 48 of Word a11d Obiect hint Quine mighr do. 8. Yet the facts of scientific agreement ro be explained can be specified, I believe, so as not ro run afoul of this. Note also that within a psychological
theory of spoken utrerances of the sorr to be discussed below (in ternu of past conditioning and three term contingencies), nothing special wou.ld ride on the question of whether a panicular sound was a "'sign of assent" or nor. Quine uses the notions of assent and dissent to develop the notion of a verdict function, and notes that conjunction and disjunction are., though truthfunctional, not (completely) verdict-functional (Roots of Reference, pp. 7778). We can further delimit conjunction and disjunction by special verdict tables which use negation, which is verdict-functional as well as truthfunctional. In Quine's verdict tables (p. 77), let q be notp; we get six cases which don't arise, bur we also get dissent as the center entry for conjunction and assent as the center entry for disjunction. If to learn and and or is also to learn these special verdict tables, we have the law of non-contradiction and excluded middle centrally bound up with the very learning of these notions. (Contrast Quine's remarks on p. 80 of Roots of Referenee.) 9. Word and Object, p. 39. A more limired view of stimulus meaning as individuating what can be learned in ostension is taken by Quine in "Replies," p. 313. 10. I take the following listing of 'facts' to be explained by semantic theory
from jerrold Katz, Sema11tic Theory (New York: Harper and Row, 1972), pp. 5-6, who goes on to say (p. 7) "that the phenomena cited ... are semantic in nature and that together they provide us with a reasonable conception of what phenomena rhe subject of semantics is concerned with seems to me beyond serious doubt. 11. See Katz, p. 24, for this description of communication. 12. Noam Chomsky, "formal Properties of Grammars," in R. D. l.uce, R. Bush, and E. Galanrer (eds.), Ha11dbook of Mathe,UJtical Psychology, vol. 2 (New York: john Wiley and Sons, 1963), p. 330. Copyrighted material 364 -& Notes and References 13. The standard p<>sition within the philosophy of science would have it he the simplest linguistics plus psychology that is desired, just as it is the simplest geometry plus physics. A partial focus on the simplest linguistics corresp<>nds to Poinc.art's commitment to the simplest geometry. On the view to be de-scribed below, in which linguistics disappears within psychology, what would be desired is the simplest total psychology. 14. Chomsky, Aspects, p. 4. 15. Katz, Semantic Tl~eory, pp. 16-17, in a section entitled "Theories about the objeaive reality of language, rakes this view of language. See also de Saussure, Course in General Linguistics (New York: McGraw Hill, 1959),
p. 9; and louis Hjelmslev, Prolegomena to a Theory of Language (Madison: University of Wisconsin Press, 1963), pp. 5~. 16. Quine, Philosophy of Logic, pp. 15- 16; "The Problem of Meaning in linguistics," io From a Logical Point of View, pp. 49, 51-54. 17. Quine asks, "What is the rationale behind that infinite additional membership of K , over and above the finite part J?" and answers that the linguist rounds off. ("Problem of Meaning in linguistics," p. 54). Bur wouldn't this Quincan linguist reject a rounding off which faiJed to fit a complicated ut terance he could produce yet hadn't (with none like it being in J already)? "Without this notion (of significant sequence or p<>ssible normal urterance) or something ro somewhat the same effect, we cannot S3)1 what the grammarian is trying to do" (p. 52). Th.is should not be too worrisome if, even with the notion, we can't say why he's trying to do it. 18. There is no objection to desc(ibing patterns among the actual products of a process when this serves to explain some further facts identifiable apart from the theory which descri~ the patterns, vi~ . biologists doing taxonomy or describing the functions of organs or the physiology of organisms. Shall the set of sentences of the language be delineated in order ro specify rhe
output the psychological theory must have, thus serving as an eternal condition of adequacy on the psychological theory rather than as a comp<>nem of it? But for that purpose, one would not (contra linguistics) offer a specification which goes far beyond anything we know an actual person can do. 19. As Quine, I think, would agree. I discuss it briefly here only because it would provide a rationale for the grammariaots task as Quine states that task; and I sec no other rationale. 20. "Problem of Meaning in Linguistics," pp. 55- 56. 21. We cannot distinguish circumstances which trigger an existing disposition to Litter a sentence from those which teach the language and create such a disposition by the duration of ci.rcumstances C; conside( sentences such as "I have been -ing for n years,7' where presently you have been - ing for a far shorter rime. And any acquisition of a disp<>sition to say will itself be based Copyrighted material Notes and References ~ 365 upon a pre-existing disposition (neurological condition) to acquire the first disposition. In whar way is this preexisting disposition ruled out as the requisite disposition to utter S? 22. The Roots of Reference, p. 15. 23. B. F. Skinner, Verbal Behavior (New York: AppletonCentury-Crofts, 1957).
24. "The behaviors of speaker and listener together compose what may be called a total verbal episode. There is nothing in such an episode which is more than the combined behavior of two or more individuals. Nothing 'emerges' in the social unit. The speaker can be studied while assuming a listener, and the listener while assuming a speaker. The separate accounts which result exhaust the episode in which both participate." Skinner, Verbal Behavior, p. 2. 25. "Propositional Objects," in Ontological Relativity, p. 158; see also Roots of Reference, p. 4 1. 26. Words and Obiections, p. 284. 27. "The idioms of propositional attitudes have uses in which they are not easily supplanted" (Philosophy of Logic, p. 34). 28. A view of real.ist philosophy of science as an explanatory theory is presented by Richard Boyd in an unpublished manuscript. 29. Within his theory, Quine thinks "we give content to the ontological issue when we regiment the language of science strictly within the framework of the logic of truth functions and objectual quantification." Otherwise, saying constructions are innocent of referential intent "would be meaningless for want of a standard of referential intent" (Roots of Refermce, p. 136). But what contem does the ontological issue have when given content? What we quantify
over; but what is that standard a standard of? What (we think) there is; what our speech commits us to there being. But then Quine's remarks about why we shouldn't assess our referential intent via seeing what we quantify over with branched quantifiers are puzzling. For why should the fact that ordinary quanti. fication theory "determines an integrated domain of logical theory with bold and significant boundaries," uenjoys an extraordinary combination of depth and simplidry, beauty and utility, .. and admits of complete proof procedures for validity and inconsistency, whereas branched quantification theory doesn't (see Philosophy of Logi<, pp. 89-91; "Existence and Quantification," pp. I 11 113 ), recommend it over branched quantifiers as our criterion of ontological commitment? If these virtues are relevant to "referential intent," then either we need a sharper discussion ro show why, or "referenrial intent" doesn't mean what we thought it did. When the choice of criterion is so delicate, what is the value or purpose of the ontologist's (standing on the best current theory and) ttlling us what exists? Copyrighted material 366 -1> Notes and References 30. See Skinner, Verbal Behavior, epilogue I, "The Validity of the Author's Verbal Behavior, pp. 453-456.
31. See Skinner, pp. 114-115. 32. In the long interim between the writing of this essay in 1975 and ics publication in 1986, a piece was detached and published elsewhere instead. See "Simplicity as Fall-Out," originally published in 1983, reprinted here as Essay 8. References for Essay 7 Quine, W. V. "Epistemology Naturalized." In J. R. Royce and W. W. Roozeboom (eds.), The Psychology of K11otving (london: Gordon and Breach, 1972). Also in R. M. Chisholm and R. S. Swartz (eds.), Empirical Knowledge (New York: Prentice Hall, 1973). ---"Existence and Quantification." In Ontological Relativity a11d Other Essays. --- From a Logical Point of View (Cambridge, MA: Harvard University Press, 1953; new ed., 1980). ---"Narural Kinds." In Ontological Relativity and Other Essays. --- Ontological Relativity and Other ssa)s (New York: Columbia University Press, 1969). --Philosophy of Logic (Englewood, NJ: Prentice Hall, 1970; rev. ed., Cambridge, MA: Harvard Universiry Press, 1986). ---"Replies [to Critics]." In Donald Davidson and Jaakko Hintikka (eds.), Words and Objections: Essays 011 the Work of W. V. Quine (Oordrtcht: D. Reidel, 1969). --The Roots of Reference (La Salle, 11.: Open Court, 1974).
--- Word and Object (Cambridge, MA: MIT Press, 1960; paperback edition, 1964 ). 8. Simplicity as Fall-Out This essay was part of a larger one wrinen in 1975 (see Essay 7) and was deleted from that one in order to appear in a volume of essays in honor of Sidney Morgenbesser in 1983 (How Many Questions?, ed. Leigh Cauman et al.). Its publication in that volume was especiaUy appropriate because it was in Morgenbesser's classes rhat I first encountered the writings of Quine and issues about simplicity. 1. W. V. Quine, "On Simple Theories of a Complex World," in his The Ways of Paradox (Cambridge, Mass.: Harvard Universiry Press, 1976), p. 255. 2. The Web of Belief (New York: Random House, 1970), p. 47. Copyrighted material Notes and References ~ 367 3. I doubt whether Quine is correct in his particular claims about how sim plicity might artifactually be favored by our procedures. Some experimental setups allow us to get evidence for similarities but not for differences. Do none work in the reverse direction? And Quine's claim that if we add a parameter we have modified a hypothesis whereas if we change a parameter the hypothesis is refuted, seems merely verbal (is it "'refuted" or .. altered"'?), and too fragi.le to
support or produce even an artifact of the procedures. 4. See Hilary Pumam, "'Degree of Confirmation' and Inductive Logic," in P. A. Schilpp (ed.), The Philosophy of Rudolf Clrnap (La Salle, lll.: Open Court, 1963), esp. pp. 770-774. Refinements of our approach to sim pliciry, I hope, would apply to mort complicated and intricate versions of this simple model. 5. It is an oversimplification to say that onlr data gathered after a hypothesis is formulated support it. (Even the oversimplification should be put more accu rarely so that it ignores when the hypothesis was formulated, and concenn-ates on whether or not the hypothesis was designed to fit the data .) A newly ac cepted hypothesis (accepted on the basis of independent data gathered after its formulation) will inherit as support the data of the hypothesis it supplants, even though the new hypothesis was geared to explain the previously known data. So there is a historical change in whether data support a hypothesis; data that don't at one time support a hypothesis (because the hypothesis was dreamed up to fit those data) may come later to support it, when that hypothesis is accepted because it also fits new data that refute the previously reigning hypothesis. This phenomenon of inherited support makes it (even more) likely that the method of tenacity will yield much of the maxim of conservatism about
theory change. 6. More accurarely, that subset of the original barch conraining all those as simple as Hs, including Hs itself. 7. Note that we have presented, thus far, an invisible-hand explanation of simplicity. [See my Anarchy, State, a11d Utopia (New York: Basic Books, 1974), pp. 18-22.) There may be still another connection berweeo simplicity and un derdetermination of theories. Underdetermination of theories makes it too easy ro find a hypothesis that fits the data, and so the fact that one fits isn't (much) suppon for it. Whereas being a hypothesis that data gathered afterwards fit is hardec But there might be yet other bard conditions to fu lfill. Is a hypothesis supported if it fulfills any hardto-fu lfill previously formulated condition C, or must something independently be said for Cas connected with truth? To what ever extent the former gives support, then, having a certain degree of simplicity will count for a hypothesis, if it is nor easy to find a simple hypothesis that fits the data. 8. If "simpler than" was not merely some structural relationship berween hypotheses or theories, but also depended upon the way the hypotheses handled the data at hand, then the argument would not go through. For in this case Copyrighted material
368 -<> Notes and References H1. .. . , Hm might each be simpler than any other hypothesis to explain data Dto yet at the second stage still another hypothesis could be simpler than any one of these in relation to the data Dt + D2. (I owe this observation to Walter Gil bert.) Thus the argument applies only to notions of simplicity which satisfy 5. Note also that the conditions do not mention or require multiple generation of T's at any stage. Though presumably the more there are, the quicker there will be convergence to the Des.t. 9. Might the whole truth about simplicity be that standards of simplicity change to fit what we believe? No, for we still think some hypotheses that were rejected earlier (e.g., circular planetary orbits) are simpler than what we currently believe. There is no psychological process that creates a simplicity gradient that peaks at our current beliefs. 10. Gilbert Harman, "The Inference to the Best Explanation," Philosophical Review 74, 1 Uanuary 1965), 88- 95. 11. When we seek simplicity in this way, the total process whereby we end up with simplicity is no longer merely an invisible-hand process, though it is based on one. The invisible-hand character of even the original faJJ-out process would be undermined if part of the explanation of why (in that process) we
think up the hypotheses we do is that they are simple; that is, if we tend to think them up because they're simple. On the view we have presented, we think chcm simple because we rend to lhink them up. I need hardly mention that I do not claim that the view presented here is the complete story about simplicity. But so intractable is the topic of the real connection between simplicity and truth that it is progress, I think, to chip away at simplicity, reducing what remains for a (metaphysical?) theory to explain. 9. Invisible Hand Explanations This essay is dedicated to the memory of Raymond Lubitz, 1937-1984, A.B. Columbia (1959), B. Phil. Oxford (1961), Ph .D. in Economics, Harvard Uni versiry ( 1967), Assisram and Associate Professor of Economics, Columbia Uni versity (1967-1973), Federal Reserve Board, Washington, D.C. (1973-1984), and Chief of irs World Payments Economic Activities Section, Division of Inter national Finance; coauthor of Kenen and Lubitz (1971). 1. Daniel Dennen (1991) proposes such a disaggregated theory of the self. Question: what decentralized competing processes within an individual would give rise to a (relatively) coherent decision-..maker? 2. For instance, a theory would be interesting if it showed that, although everyone was aiming at a partern, either their actions animated by that aim
were not what produced the pattern, or if they did, that the pattern did not Copyrighted material Notes and References ~ 369 arise by the route everyone imagined-it was a side effect of their envisioned plans. 3. The evidentiaUy expected utility of an action is the weighted sum of the utilities of its (exclusive) possible outcomes, weighted by their conditional prob abilities given the actions. The causally expected utility of an act replaces these conditional probabilities by probabilities (of outcomes on actions) reflecting some direct causal influence. 4. Notice that, on this view, decision theory is not a theory of rational action, but of best or optimal action. An action would be rational if it were given rise to by a process that reliably yields optimal or maximizing actions. However, a person might happen to stumble confusedly upon doing such a maximizing action. In performing it, he would not be acting rationally, for his action would not be generated by a process that reliably produces optimal action~. 5. Invisible-hand explanations need not be a subclass of merbodological in dividualist ones. Suppose thar some panern arises at random in particular soci eties, and also that there exists an irreducible filter (not
susaptible tO individualist explanation) that eliminates all societies that do not fit that panern. Then there would be an invisible-hand (but not an individualist) explanation of why all societies fit that panem (see Nozick, 1974, p. 22). (I sharpen and discuss the notion of methodological individualism in Nozick, 1977.) In his Ely lecture, Kenneth Arrow (1994) refers to common inforrnation in the public domain as raising problems for methodological individualist expla nation, but it is not evident why either the existence or the consequences of such information (or of books!) must elude such explanations. References for Essay 9 Arrow, Kenneth ]. 1994. "Methodological Individualism and Social Knowledge." American Eco11omic Review 84(2): 1-9. Becker, Gary. 1981. A Treatise on the Family. Cambridge, MA: Harvard University Press. Cambell, Richmond, and Lanning Sowden. 1985. Paradoxes of Rati01111lity and Cooperation: Prisot~er's Dilemma and Newcomb~$ Problem. Vancouver: University of British Columbia Press. Dennett, Daniel. 1991. Co11sciousuess Explained. Boston: Linlc, Brown. Kenen, Petre B., and Raymond Lubitz. Intenltltional Eco11omics, 3d ed. Englewood Cliffs, N]: Prentice-Hall. Nozick, Robert. 1969. "Newcomb's Problem and Two Principles of Choice."
[See Essay 2.] --- 1974. Anarchy, State, and Utopia. New York: Basic Books. --1977. "On Austrian Methodology." [Sec Essay 5.] Copyrighted material 370 ~ Notes and References ---1993. The Nature of Ratiot~ality. Princeron, Nj: PrincetOn University Pre.ss. Ullmann-Margalit, Edna. 1978. "Invisible-Hand Explanations." Synthese 39(2}: 263-291. 10. Moral Complications and Moral Structures Parts of this essay were delivered in an address given to the twelfth annual meeting of the Board of Editors of the Nat,ural Law Forum, September 29, 1967. I. An especially clear and self-conscious example of attempting to carry our this task is Henry Sidgwick's Methods of Ethics (1907). 2. Goodman (1955), chap. Ul 3. He might not say "because it has features Th .. . , Tm," but rather "because it has features --" and fill in the blank with some though not all of the m T -features. The task of accounting for the shortening of the answer is similar to chat faced by an adhetem of the deductivenomological model as an adequate account of a certain kind of explanation (or as presenting necessary conditions for one important kind of explanation) in explaining why when one says ""Evenr E happens because __ .. one does not normally fill
in the blank with all of the things that the model says forms part of the explanation, not even with all of the initial conditions. For a discussion of the kinds of principles underlying the shortening of explanation answers (which is relevant to the point with which this footnote begins), cf. S. Bromberger, "An Approach to Explanation," in Buder (1965); "Why Questions," in Colodny (1966); and Hart and Honore (1959), chaps. 1-11. On the deduccive-nomological model of explanation, see Hempel (1965). Though I shall later in the text be discussing a model of the structure of moral views other than the deductive one, this alternative model also faces the task of accounting for shortened answers. The literature cited here is also relevant ro thar task for the alternative model, which I shall not attempt in this paper. 4. That is, principles of the form of (3) and (4) where it can be decided, with Little room for fid dling, oo nonmoral grounds whether the features mentioned in the principle apply to ao action. For other sorts of features the persons I am c-onsidering might be willing to assent ro exceptionless principles, e.g., "any action which shows lack of love of one's neighbor is morally impermissible." 5. Cf. The American Law Institute (1962}, sec. 3.02. No rules are provided
whereby it is to be determined whether "the harm or evil sought ro be avoided by such conduct is greater than that sought to be prevented by the law defining che offense charged," a!ld presumbly this is to be dedded by a jury. Copyrighted material Notes and References ~ 3 71 6. It is worth briefly noting one kind of theoretical solution to this particular reason for desiring judicial discretion. (''Theoretical" because it ignores lirnita tions upon the time and energy of legislators and le-gislatures.) In those cases, which are not constitutional cases and which needn't be decided immediately, where, under the doctrine of judicial discretion, the judge should use his discre tion (e.g., an obvious injustice is worked by the law as it "clearly" stands, the legislature did not foresee such a case and would not, one thinks, have intended the consequence required by the law in such a case, etc.), the judge finds the person gui.lty and throws the case back to the legislature. That is, he notifies the legislature that here is a case where he believes an obvious injustice is worked by the law as it "clearly" stands, and the legislature has a specified amount of time to pass new legislation in this area. If under the new law, the defendant's act is no longer an offense, he is then found innocent. If some act performed by
someone prior to the promulgation of the new law is an offense (only according to the new law), be is not tried. In the preceding paragraph, I consider only those cases where, under the doctrine of judicial discretion, the judge should use his discretion to avoid the ddcndant's being found guilty of an offense. In the cases where the law "clearly" holds the person innocent, many of the objections to ex posr facto legislation would apply to a judge's using his discretion to find the person guilty. (Though perhaps one would sometimes wish it 10 be done. Cf. e.g., Riggs v. Palmer, 115 N.Y. 506,22 N.E. 188 (1 889].) Similarly, though leSe< strongly, for cases where no particular finding is "clearly" required by the law as wrinen. In cases where two patties are opposing litigants, a more complicated procedure would be needed. Many detailed questions arise about the operation of a system such as the one sketched above which, in view of the theoretical character of the solution, are not worth pursuing here. The reason for mentioning such a "solution" is to point out that no analogous procedure is even theoretically available to us in the moral case. For, puniog it roughly, our task of accounting for an individual's moral judgments is like the one of formulating the rules and principles by
which, under the scheme, the legislature decides the cases thrown back to it, or the judge decides when an inj ustice would be worked by the nondiscretionary application of the law. 7. See his books, The Right and the Good (Ross, 1930) and Foundations of Ethics (Ross, 1939). 8. Whether the Rlist should contain (and contain only) features such that an action having some of these features, and none on the IV-lisr, is morally required, is a question that requires much discussion. What is important is that it contain features which may override featur<s on the IVlist. The host of complicated questions about how to stare this, and its consequences, will be considered later. Copyrighted material 372 ~ Notes and References 9. This seems an appropriate place to informally explain some symbols which sometimes are used at different places in the rext, wim which some readers may be unfamiliar. XU Y (the union of set X and Srt Y) is that set containing as members just rhose things which are members of X, or Y, or both. ~ is the set without any members, and is referred to as the null set. "X c Y" is read "X is a (proper) subset of Y" and is true if and only if each member of set X is also a member of set Y, and some member of set Y is not a
member of set X. "X c Y" is read "X is a subser of Y" and is true if and only if X c Y or X and Y have exactly me same members. A selection ser from a group of sets is a set containing exactly one member from each Srt in the group. "x l X" is read "x is a member of me ser X." "(3x) ( .. x ... )"is read "There is an x such mat ( .... < . )," e.g., "(3x) (green x)" is read "There is an x such that x is green" or "There is something which is greeo." "(x) { . .. x ... )" is read "For all x, ( . . . x . .. ),"e.g., "(x) (green x)" is read "For all x, xis green." "-> ... " is read "if--then ... " As an example of how things fit togemer (X)(X is a set-> (3y)(y l X ) or X =~) is read: For all X, if X is a set then there is a y such that y is a member of X, or X is equal ro the null set. 10. If an action with members of R and no members of W may be morally permissibl~ though not morally required~ it is not clear how to interpret~ < R1 I 1. I shall ignore in this paper any need for variables ranging over all me rationals or reals. 12. In this section "features" shall refer to features on the lists wimout variables, or specifications of features, with variables, on the lists (i.e., with c.onsta. ms substituted for each varia bid. If certain conditions are mtt, fe-atures with
bounded variables (the bounds being represented by equations) may be included also. I shall not pursue here the obvious way to do this. 13. The metaphor of balancing and outweighing leads one to want to consider a notion of exactly balancing, and to want to distinguish moral views which hold that in exact balancing situations rhe action is morally permissible, from those mat bold that in exact balancing situations the action is morally impermissible. Though our explanation of the inequality sign does not leave room for this distinction, it may be desirable to pursue the possibility of other structures which allow it. 14. Strictly speaking, since (2) is not imposed upon our fragmentary moral views, this intuitive argument for (3) should be put somewhat differently. 15. In the absence of (3), one might wish to define X Y as (Z) (Y > Z -> X > Z) & (Z)(Z > X -> Z > Y) & (3Z) (X > Z & Z > Y). Copyrighted material Notes and References ~ 3 73 16. For an explanation of this and other ordering terminology, see Suppes (1957), pp. 220-223. 17. Very difficult issues arise here, related to some about explaining the notion of "intrinsically valuable" when there are causal ot probabiliry connections between features, which must be considered in a full exposition of the theory.
18. It does not matter whether you agree with my moral conclusions about these particular two situations. What is important is that you can find two situations with structures parallel to these which you would view as I view these: e.g., in (1) the only way to stop P is to shoot him; in (2) you can also stop him by lying to him. 19. I am here assuming the truth of condition (5) of section IU. If this condition or some of the others accepted in section III are inadequate then more complicated principles must be formulated here. But even if the conditions in the pre"ious section art inadequate and must be replaced, we can adequately raise the issues we are concerned with here by using them. 20. One does not need a scale of measurement stronger than that gotten in section III, to state condition V. For ( W A - W s) > 0 can be stated, using the notation of section In, as w,. Ws. Similarly, one may stare condition VI below without assuming such a stronger scale of measurement. However, the statement of VII, below, does require the assumption of a stronger scale of measurement than we yet have available. 21. We might consider the stronger principle (VII') If (38) (B is an alternative available to the person & WA \Vs & (W,. - W8) 2!: (RA - R8)), then it is impermissible ro do A. This principle, unlike (VII), covers the case where ( IVA - IV8)
= (RA- R8). Given a scale of measurement we might be able ro interpret this, but we could nor; if my remarks in section Ill were correct, offer an intuitive argument for (VII') similar to that offered for (VII). For the action C which took up the difference between Band A would have to be such rhar IVc = Rc. And we have not seen any way to interpret this. 22. Paralleling the economists' notion of opporruniry cost, one may deny this. My claim, to be elaborated below, is that all morally relevant opportuniry cost of an action is alre-ady implic.idy built inro its \Vfearures. 23. I am not claiming char it is always permissible ro lie ro someone ro save oneself personal expense and inconvenience, ot that it alwa)'S is permissible when the R-features outweigh the W-features, but that it sometimes is in this latter case. No doubt, the pursuer?s performing a wrongful act makes it especially clear in this case. 24. That is, are there cases where C does not stand in the relation described by (VU) to A, yet it is impermissible to do A (if 8 is available) and if so, how are Copyrighted material 374 ~ Notes aod References these to be marked off from the cases where 8 srands in the relation described by (Vll) 10 A, yet-because of the personal costs which B involves- it is not
impermissible to do A? 25. This First Asymmetry thesis has the consequence that there is no duality which involves only the substitution of "'permissiblc>t for "impermissible,,.. and <" foe">," as there is in the simple structure with which we began in section II. Talk of asymmetries brings to mind the famous asymmetry between good and bad; viz, that all good things must come to an end. 26. Note that it does not follow from this chat in every possible situation there is at least one action available co a person which is morally permissible. It may be that a person can (wwngfully) intentionally put himself into a situation such that none of the alternatives in the situation he got himself into are morally permissible for him to do. 27. Notice that we are here speaking of whether something about the alternatives to A can prevent A from being impermissible even though WA > RA. In section VI we shall consider whether A can be morally permissible, even though IV A > RA, because of something about larger courses of action of which A is a part. 28. I owe this example, which led to a rewriting of an earlier version of this part of this section, to l'rofessor judith Thomson of the Massachusetts Institute of Technology. 29. This is not the same as saying that if the train were
proceeding along a side crack it would be permissible ro switch it tO the center track because a voiding the death of 40 persons outweighs killing 20 persons. 30. We might say that a feature of an ace, for our purposes, is alx>ut that act, and not about, even implicitly, others. Of course, the problems in explaining this are immense. 31. Cf. Von Neumann and Morgenstern (1953}, appendix; Luce and Rai.ffa (1957), chap. II; G. Debreu, "Cardinal Utility for Even-Chance Mi.xtures of Pairs of Sure Prospects," Revieru of Economic Studies 26 (1959), 174-177. For weakening of the strong conditions, relevant to artempts to parallel them in rhe moral case, sec M. Hausner, uMultidimensional Utilities," in Thrall, Coombs, and Davis (1954), an d R. j. Aumann, "Utility Theory without the Completeness Axiom," Econometrica 29 (1962}, 445-462. 32. Throughout (a)--(d') the labels of the form Fx and Wx are merely labels for sets of fea tures and do not say anything more thao thjs. I have used these labels, in a notation which is by now familiar, because I shaiJ soon discuss the sers of features for situations where the here-arbitrary labels mean what they have meant previously. 33. Perhaps the person could reasonably judge that, under the described circumstances, a particular A is impermissible if B is a\'ailable for re.asons quite
Copyrighted material Notes and References ~ 375 differenr from the sort we are here considering. One should incorporate conditions excluding such reasons into the antecedent of the principle. (I assume there are a smaU number of kinds. One does not want to require in the antecedent of the principles the proper kind of reason unless there is a way of specifying it which does not require reference to an apparatus beyond the one we had before numbers were introduced.) In this way one can hope to eliminate any counrer-examples to the principle which there may be, which are irrelevant to our concern here. 34. This is all put roughly. For detailed discussions of Representation and Uniqueness theorems, see P. Suppes and J. L. Zinnes, "Basic Measurement Theory," in Luce, Bush, and Gala mer (1963 ). 35. On structural conditions, see P. Suppes, "Some Open Problems in the Foundations of Subjective Probabiliry," in Macho! (1960), and D. Scott and P. Suppes, "Foundational Aspects of Theories of Measurement," Journal of Symbolic Logic 23 (1958), 113- 128. 36. For further details on the sort of existing system I am thinking of, see S. Siegel, "A Method for Obtaining and Ordered Metric Scale," Psychometrika 21 (1956), 207-216; P. Suppes and M. Winer, "An
Axiomatizarion of Uriliry Based on the Notion of Utility Differences," journal of Management Science 1 (1955), 259-270; and especially R. D. Luce and j. D. Tukey, "Simultaneous Conjoint Measurement: A New Type of Fundamental Measurement," journal of Mathematical Psychology I (1964), 1- 27. 37. If we let wz be the set-theoretic difference (IVA - IV~) then, given an addiviry assumption wh.ich is not unreasonable for these cases, one will get (for the case where IVA> RA ) the result that IVZ > R.,- IV~. And by using this, we may get a measure of the weight of rhe "features" which refer to the suitable alternative acts. 38. I have already argued that because of considerations about alternative courses of action taking up the same time interval as A does, RA > \V" is nor a sufficient condition for A's being permissible. 39. Suppose, for simplicity, a finite number of possible larger courses of action of which A is a part, A 1, A 2, .. . , A". Then if "is part of an act which - " is a feature of an act, rhen A would have as a W fearure "must be part of an act with fearures IVAl or with features IVA2 or .. . or with features IV.n. 40. Strictly, no pan of measure non-zero. I shall omit this in what follows. 41. One can de.finc an apparently weaker notion: A is weakly undercut, =
df. (X) (X is an act over t containing A -+ (3Y) ( Y does not contain A & Y undercurso X)). Similarly one can get de6nitjons paralleling the ones which follow by using "weakly undercuts" rather than "strongly undercuts" (and other obvious changes). Interesting questions arise about the relation of the "weak" notions to rhe "stroog"' ones. Copyrighted material 376 -<> Notes and References 42. (2) is equivalent to: If IVA> RA then [A is permissible if and only if (3C) (C is available ro the person & A is part of C & Rc > We & C is nor ruled impermissible by (! ))]. 43. In this case, principles (1) and (2) above. 44. Here, as below, I ignore the issues raised in sections IV and VI, and momentarily suppose that permissibility or impermissibility is determined by the way some general inequality goes. Such a simplification is useful as a fitSt step in an intuitive explanation of the different ways. 45. As in the case where you don't go to a dinner ro which you promised to go, in order to minister to an acddent victim whom you encounter on the way to the dinner. You must call your hostess "as soon as possible" (compatible with your best ministering), make explanations, etc. 46. Though perhaps the person saved is so required. This raises the question of whether in all such cases someone (though perhaps not the
person who did the act) is morally required, if it is possible, to make amends or reparations 10 those harmed by overridden W -features of an act. Pursuing this issue leads one to issues similar to some discussed by welfare economists; see Little ( 1960), chap. VI. 47. When one re.aches the point where one can make amends or reparations or o ffer explanations, some i'eacures override or overcome "'not making ame.nds or reparations and not offering explanations after a I situation" which do not override or overcome "not making amends or reparations and not offering explanations after a [I situation," and the second outweighs or overshadows some R-fcatures that the first docs not. 48. Take the first person who T-ed someone else. U his action is morally impermissible, he is from that time forth, a T-person. If his action was morally permissible, take the next person to T someone. The first one who does it impermissibly becomes the fi rst T-person. (l ignore here tbe possibility that two persons simultaneously and impermissibly T each otheL) The only way to become a T -person is by T -ing someone who is not a T -person, where the IVfeatures of this act of T -ing are not overridden, overcome, etc. On the view we are cons.idering, it is permissible tOT aT-person; that is, the \\'-
fea ture, T, of an action has no moral weight when the aCtion is done to a T -person. I do not wish co discuss here whether, for some feature T on the W -lise, such a view is correct. It is a possible view, and an adequate account of the sort of moral structures we ate discussing must be able to handle it. 49. I shall not pause ro consider the different views that fall under theretributivists' view that it's good that someone who has committed a wrong suffer; e.g., is ir good that be suffer, or is it good that he suffer because he's committed the wrong (that his suffering be due to his having committed the wrong), Copyrighted material Notes and References ~ 377 or is it good (that his suffering be due tO his having committed the wrong, and be know that his suffering is due ro having committed the wrong)? And if more than one of these is held to be good, do they differ in degree of goodness? SO. V-VIII have only been sketched. The details needed to state them correctly would rake us away from our major concern here. 51. To be sure, the lawyers must write something down completely, without open-ended lists. (Though perhaps section 3.02 of the Model Penal Code is meant to open the ends.) I should note that nothing I say here depends upon a claim that, given its content, section 3.04 of the Model Penal Code cannot
plausibly be reduced to a simpler structure. 52. Distinctions analogous to those discussed, for example, in R. Chisholm, "Supererogation and Offense: A Conceptual Scheme for Ethics," Ratio 5 (1963), 1-14, and R. Chisholm and E. Sosa, "Intrinsic Preferability and the Problem of Supererogation," S)7zthese 16 (1966), 321- 331. 53. Suggestive material can be found in levi (1967), chap. VIII. 54. I ignore the complications introduced by section IV about alternative courses of action, and also ignore the complications introduced in section VI about longer courses of action. These extra complications just reinforce the point to be made here, a.nd would unduly complicate my exposition. 55. Persons who hold that if someone is to have made a moral judgment that an action is impermissible, then he must have certain kinds of moral reasons in support of the judgment, must denr that in this case the person has made a moral judgment. 56. See Chomsky (1965), chap. I, sees. 1 and 2; and for some critical remarks, G. Harma.n, "Psychological Aspects of the Theory of Syntax," journal of Philosophy 64 (1967), 75. 57. One year after the completion of this essay, I am led to add a gnomic footnote, which I hope to explain elsewhere. It now (September, 1968) seems to me that if one were completely successful in carrying through the program of
this paper (which is far from having been done here), one would have produced a Tractatus Logico-Ethiws. What is needed, perhaps, is an Ethical lnvestigauons. References for Essay I 0 American law Institute. 1962. Model Penal Code: Proposed Official Draft. Butler, R.]., ed. 1965. A11alytical Philosophy, second series. Oxford: Basil Blackwtll. Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Copyrighted material 378 --~> Notes and References Colodny, R. G., ed. 1966. Mind and Cosmos. Pittsburgh: University of Pittsburgh Press. Goodman, Nelson. 1955. Face, Fictio>~, and Forecast. Cambridge., MA: Harvard University Press. Hare, R. M. 1963. Freedom and Reason. Oxford: Clarendon Press. Hart, H. L.A., and A.M. Honore. 1959. Causation in the Law. Oxford: Clarendon Press. Hempel, Carl G. 1965. Aspects of Scientific Explanation. New York: Free Press. Levi, Isaac. 1967. Gambling with Tmth. New York: Alfred A. Knopf. Little, I. M. D. 1960. A Critique of Welfare Economics, 2d ed. Oxford: Oxford University Press.
Luce, R. D., R. Bush, and E. Galant, eds. 1963. Handbook of Mathematical Psychology, vol I. New York: Wiley and Sons. Luce, R. D., and H. Rai.ffa. 1957. Games and Decisions. New York: Wiley and Sons. Macho!, R. E., ed. 1960. Information and Decision Processes. New York: McGraw Hill. Ross, W. D. 1930. The Right and the Good. Oxford: Clarendon Press. --- 1939. Foundati<ms of Ethics. Oxford: Clarendon Press. Sidgwick, Henry. 1907. The Methods of Ethics. 7th ed. london: Macmillan. Suppes, Patrick. 1957.1ntrod,ction to Logic. Princeton, N.J.: D. Van Nostrand Co., lnc. Thrall, R. M., C. H. Coombs, and R. L. Davis, eds. 1954. Decision Processes. New York: Wiley and Sons. Yon Neumann, J., and Oscar Morgenstern. 1953. Theory of Games and Economic Behavior, 3d ed. Princeton, N.J.: Princeton University Press. 11. On the R.andian Argument I. The main sources are Atlas Shmgged, especially Galt's long speech, and the essay "The Objectivist Ethics" in her book of essays, The Virtue of Selfishness. The other essays in this book are helpful also (later references to some of her other essays will be to essays in this book), as are her other books, including Introduction to Objectivist Epistemology, in understanding
her views. Since I shall be quire critical of Miss Rand's argument in the remainder of this essay, I should here note (<.specially since she has been given a largely viruperative and abusive hearing in print) that I have found her two major novels exciting, powerful, illuminating, and thoughtprovoking. These virtues, tvc.n combined with a "sense of life" that is worthy of man, do not, of course, guarantee that her conclusions are true,, and even if we suppose they are true, all this does not, of course, guarantee that the actual arguments offered wiH Copyrighted material Notes nd References ~ 379 be cogent, that they will prove their conclusions. Nothing I say in this essay is meant to deny that Miss Rand is an interesting thinker, worthy of attention. 2. Note in passing that persons rna)' have innate preferences unconnected with their own welfare and survival. There is a natural selection argument that, in the evolutionary process, preferences that tend to help keep you alive to the reproductive ge will be selected for, but also there are natural selection arguments that innate preferences for behavior which is reproductive behavior, and for behavior after reproduction which enhances the chances of the progeny's survival, even to destroying oneself to guard one's young, will be
selected for in the evolutionary process. So beings may be preprogrammed or predisposed to do things which lessen their individual chances of survival. 3. Nathaniel Branden, "Rational Egoism: A Reply to Professor Emmons," The Personalist, Spring 1970, p. 201. 4. It may seem a necessary truth that "'right,"' "ought," "should,~'> etc. arc to be explained in terms of what is (intended to be) productive of the greatest good. So, it is often thought, that what is wrong with utilitarianism (which is of this form) is that its conception of good is too narrow. It doesn't, for example, take rights and their nonviolation into account in the proper way, bur leaves them a derivative status. (Hence many of the counterexample cases to utilitarianism: punishing an innocent man, to save a neighborhood from a vengeful rampage, etc.) But even if one includes the nonviolation of rights into a theory in a primary way (and lhose of us for whom there is some de,sirable society in which we would choose to live, even though in it some of our rights are sometimes violated, rather than move to an island on which we could survive, alone, do nor think that it is the sole greatest good), one may include it in the wrong place-, in the wrong manner. For suppose we build into the des irable tnd-state to be achievc.d some condition about
minimizing the amount and significance of the violation of rights which takes place in the society. We then have something like a utilitarianism of rights, which could still require us to violate someone's rights if this act of violation leads to a minimizing of the amount of violation of rights in society. It might do this perhaps by deflecting others from their intended action of gravely violating people's rights, removing their motive for doing it, diverting their anention, and so on. (A mob rampaging through a part of town, killing and burning, will be violating the rights of those living there, and so one might try to justify one's act of punishing someone you know to be innocent of the crime which enraged the mob, on the grounds that this action will lead to a minimizing of the weight of the violations of rights in rhe society.) As against this conception, which builds rights into the end-state to be achieved, one might say that their proper place is as a constraint upon the actions to be done. So the structure of the view is: Among those acts available to you which don1t violate constraints C, act so as to maximize goal G. Copyrighted material 380 -& Notes and References This differs from one which says: act so as to maximize G, and which tries
to build the side-constraints C into the goal G. For the view with the sideconstrai. nrs forbids your violating having the minimizing of the violation of these constraints in the goal, allows you to violat'e the constraints in order to lessen the total amount of violation of such constraints in the society. I ig nore questions about finding a way of pulling the side-constraint view into the form of the goal-without-side-constraints view, e.g., perhaps by having a distincrion made., in the goal, be:rween your violating the copsrraint, and someone else's doing it, where the former is given infinite weight in the goal, so that no amount of stopping someone else's doing it can outweigh your violating the constraints. Even with this possibility, notice that indexical expre.ssions ("my doing something") appear in the goal. I also ignore questions about whether sometimes you are allowed to violate rights of innocents in order to prevent monstrous deeds by others. For example, suppose that the only way to have stopped the Naz.is from conquering all was to have used some weapon on them which would have also killed innocent people in the city attacked who couldn't leave and weren't choosing to stay there. Perhaps avoiding great moral horror swamps people's rights, so that one would be justified in doing
something one knew would kill innocent people, in order to stop the horror. (~< Michael Walzer's paper "World War II: Wh)' Was This War Different?" delivered before the American Political Science Association, September 1970 meeting.) The possibility of "swamping" of rights ma)' lead one to think that it is the goal structure, with a weighting of consrra ints inside the goals, which is ap .. propriate, rather than the structure with absolute sideconStraints. But other structures are possible, e.g., a side-constraint structure with pdnciples governing the setting aside of that whole structure. (But since many of the set-aside structure's features will have co reappear again in its replacement, is this the proper way to view things?) These issues, connecting with some discussed in "Moral Complications and Moral Structures" (Essay 10), are very complex, and I hope to discuss them in detail on another occasion. These considerations enable us to clarify a difficu lty with the classical con ceptioo of the night-watchman state, which limits the legilimate functions of the state to the protection of irs citizens against violence, theft, fraud, and to the enforcement of contracts, etc., but specifically exclude.s redistributive tasks. The difficulty (which Jlirst presented in a paper, A Framework for Utopia,"
delivered at the APA Eastern Division Meeting in 1969, with Sidney Hook as commentator} is that the protection of people in this way costs money (for de tectives, police to bring criminals into custody, courts, prisons). How is it to be paid for, and, in particular, is there some non-redistributive way of financing it? We might imagine a S)'Stem in which the state provide.s these protective services only to those who specifically pay for them, with perhaps different packages of protective services offered for different fees. People who don't buy a proCopyrighted material Notes and References -& 381 tection conuacr don't get protected. Now consider this system of individuals purchasing protection from the stare combined with a voucher system (cf. Mil ron Friedman's school vouchers, which, of course, allow rhe people a choice of schools) funded by taxation, under which all people, or those in need, ere., are given vouchers, backed by the funds collected by raxarion, which can b<o used only for rhe purchase of a protection policy &om rhe srare. lr i.s clear that the latter system is redistributive (compare it to the former), yet this Jar ter system is equivalent to the system of the nighrwatchman srate. So, unless some other nonredisuibutive method of financing protection for everyone can
b<o found (and, on the face of ir, it is difficult ro see how it can be), the night watchman state itself performs some redistributive functions (viz., having some people pay for the protection of others). And, it might well 1J<, asked, if this re distributive function of the state (redistribution i.n order to protect everyone) is legitimate, why is redistribution for other attractive and desirable purposes nor legitimate as well? On the other band, if the nonredistriburionisr accepts some thing like the first system described, then his position will seem to many ro be inconsistent, as well as monstrous. I wish here to focus only on rhe first of these charges. The proponent of the night-watchman state is greatly concerned with the violation of rights and with protecting these rights; so that he makes the protection the onlr legitima<c func tion of the stare, protesting that other functions are illegitimate b<ocausc their performance will itself involve the violation of rights. How can he then, given his paramount placing of the protection and nonviolation of rights, move to th first position which would, it seems, leave some persons' rights unprotected or ill-protected; how can he do this in the name of the nonviolation of rights? This objection assumes that the proponent of the first system is a "utilitarian of rights,,. whose goal is, say, the minimi:t.ation of thr amount
and significance of the violation of rights in the sociery, and who will pursue this goal, even through using means which themselves violate people's rights. But if he places the nonviolation of rights as a constraint upon action, rather than building it into the end-state ro be realized (it may, let me hasten to add, occur in both places), and his conception of rights is such that your b<oing forced ro contribute to another's welfare violates your rights, and is such that someone else's not providing you with something you need greatly, essential to the protection of your rightS, docs not itself violate you( rights (though it may faciUrate, or avoid making more difficult, someone else's doing so) then this theorist wiU be led to something like rhe first system of individuals purchasing protection for them selves (and those others they voluntarily choose to do so for). And this wiU 1J<, a consistent position. (That it is a consistent position does not, of course, show that it is an acceptable position.) Given the subject of this essay, we do well here to mention Ayo Rand's pro posal, in her essay entitled "Government Financing in a Free Sociery." Briefly put, her suggestion is that the government enforce the conttacts of only those Copyrighted material 382 ~ Notes and References
who have paid a special fee (others may sign contracts backed by trust, de sire to continue their good reputation etc., but oor. by governmeot's powers of force) and that the government's other legitimate functions be financed our of the funds so collected from those who voluntarily choose to make use of (and pay for) the government's contractcoforcing powers. Bur it is difficult to see why she finds this a legitimate solution to the problem. For the government has a legal monopoly on the use of force, and the proposal involves the government charging extra to erlforce contracts, in order to cover the cosrs of its orher protective functions. (Compare what it would charge for enforcing contracts if it did nor take on these other protective functions.) Why is this nor illegitimate forcible redistribution, and isn't talk of people voluntarily dealing wirh the legal monopoly (and voluntarily paying the higher fees to cover rhe legal monopoly's other functions?) roo quick? If in the United States today, the post office, with a legal monopoly on the right to carry the mails, charged one dollar per let<er, in order to cover the costs of other redistributive activities (and thus charged more than even a monopolistic post office, covering its costs, which was singlemindedly devoted to delivering the mail), would Miss Rand think that a legitimate
me,ans had been found to finance redistributive activities, say the educational costs of college students? Could anyone who otherwise thought that there was a problem with governmental redistributive activities feel that problem avoided by tying redistributive activities tO the provision of some service protected by a legal monopoly? To be sure, there arc powerful arguments for the state's having a legal monopoly on the use of violence (rather than allowing, e.g., private contract-enforcing firms to use force ro enforce conrracrs rhey have been hired to enforce), as rhere are not for rhe post o ffice's having a legal monopoly rode liver the mails. But since the arguments for the state's having a monopoly on the use of violence do nor involve redistributive considerations, it is difficult to see why it is legitimate to allow the monopoly thus established to exploit its monopolisitic position in order to pursue redistributive aims. These considerations are not addressed to and will have, perhaps, little interest for tbose who find no problem with the state's forcible redistributive activities; they are meant merely ro point out that those who do find a problem cannot avoid it by slipping the state's redistributive activities in through the (only) back door, in the fashion of Miss Rand. 5. Rand, "The Objectivist Ethics," p. 27.
6. Sec also our discussion in section V, which considers a case where John Galt seems not tO act on ( I ). 7. If rights are explained as "conditions of existence required by man's na ture for his proper survival," then we can reconstruc-t the Randian argument for a right to life, but this argument will not answer tbe question: why should I not violate another's right to life; why should I not intervene by force to eliminate one of the conditions of another man's existence required by his narure for his Copyrighted material Notes and References ~ 383 proper survival? That is, with this explanation of rights, argument is needed for (6) above. If we assume that rights are not to be violated, and others should nor forcibly intervene in the exercise of someone's rights, then argument is needed to the conclusion that a person does have a right to his own life, that is, that others shouldn't forcibly intervene in it, even granting thar irs maintenance is his highest value. Taking either approach, we face the que.stion of why one person shouldn't intervene by force to thwart the conditions of another man's existence. A similar que,stion is raised by Mortimer Adler's argument in his The Time of Our Lives: The Ethics of Common Se.zse (1970), where there is a transition
(chap. 14, sec .. 2) from an individual's moral obligation to pursue his own real good to an individual's moral right (entailing obligations of others not to interfere) to pursue his own real good. He attempts to bridge this gap (chap. 16) by grounding our moral obligations nor to violate the rights of others on our obligation to make a good life for ourselves. He argues as follows: (I) Each of us needs civil peace as a means of making a good life for ourselves. Every universal act which injures other men or the community is a breach of civil peace. "Hence when I act unjustly toward others or act in any way that is contrary to the good of the community, I am injuring myself. It may nor appear to be so in the short run; I may gain apparent goods by my injustice toward others or by criminal activities that injure the community itself. But in the long run, I may have gained these apparent goods only at rhe loss of a real good rhar I need-the civil peace of the community in which I live. It is only in the short run that injustice can appear to be expedient. In the long run, which is the omen of my whole life, the just tends to be the expedient" (p. I 73, my italics). Note that my individual conduct breaches the peace; it docs not de,stroy it. And it is an undestroycd pe3cc (rather than an unbreached one)
that it is said I need. (2) Under ideal conditions "when an individual seeks only those things which are really good for him, he does not infringe on or interfere with the pursuit of happiness on the part of others through their seeking the same real goods for themselves" (p. I 74). For Adler says, consider that one man's pursuit of real self-improvement (which is the greatest good) cannot interfere with another's similar pursuit. But if there are scarce (in the economist's sense, which does t~OI entail that the situation is non-ideal, in Adler's sense; cf. pp. 177178) resources of self-improvement (books, musical instruments), then such interference is possible; similar!)', with teachers, with the added possibility of forcing one of the few able people in a subject (who would rather be improving himself) to instruct you. It is een easier to see how this might happen with rhe goods other than self-improvement that Adler discusses. (3) Under non-ideal conditions, men may nor obtain the privileges which enable them to lead good lives by force or fraud (footnote 3, pp. 309- 310). Copyrighted material 384 -& Notes and References No reason connecting this constraint to an individual's obligation to pursue a good life for himself is offered by Adler. (Nor is any other
reason offered.) It would be interesting to hear what reasons he would produce, in view of his statement (pp. 171- 172): If it could be shown, as I think it cannot be shown, that the individual's obligations to others and to the community are independent of his obligation to make a good life for himself, then the discharge of such obligations would impose a burden on him that might interfere with or even frustrate the pursuit of his own happiness. I would like to add, in passing, that those who regard their duties to others or to the community as independent of their obligation to make a good life for rhemselves are either sentimentalists or thoughtless do-gooders"!! 8. Or, to go to a case which is merely meant to caution one about how one formulares the harmony condition, each of two boxers, who have promised to win the fight, and who each have contracted with different outside parties to win, should win, and furthermore, it is permissible for one to thwart the other's attempt to do what he should do. 9. I say "stronger" re.alizing that she would deny that it is a stronger claim, because she thinks that there is no additional step needed to get from somerhing being in one's interests to its being something one should and ought to do. 10. We should notice here a slide many wrirers in and readers of
this tradition make, from X is a (morally) better act than Y, to one is (morally) better doing X than Y, to one is better off doing X than Y. An argument that tight conduct and self-interest don't diverge requires independent ways of identifying each, plus an argument that these tw.o independently identified things always go together; at any rate, so it must go if the ordinary view that self-imerest and rightrtess are distinct notions is to be adhered to. Alternatively, it might be argued that they don't diverge because there are nor two independent notions, but rarher one of them is primary and the other one can be explained in a coherent fashion only in rerms of the first (in such a manner as to yield their nondivergence). So, one might explain "morally right" or "morally ought" in terms of "self-intertst," or against such a background (Rand). Or, one might attempt to argue that the notion of self-interest irself cannot be coherently explained except in terms of morally right (with a particular form, and perhaps content) in such a fashion rhat they don't diverge. This latter possibility is an exciting and tanralizing one, which should be pursed. 11. See her essay, "The Conflicts of Men's Interests," in The Virtrte of Selfis/ mess. 12. It is sometimes assumed thar the unconstrained (constrained)
egoist can give no weight to the interests of others except as they are reflected in his own interests (and the constraints). But a theorerically possible egoist posirion could require: Maximize self-interest (subject to constraints) and among rhose actions which tie so far as (constrained) self-interest is concerned, pick that one which Copyrighted material Notes and References ~ 385 best satisfies condition C (where C may be any condition, e.g., the utilitarian on.e). Here we would have a lexicographic ordering, with the (constrained) egoistic score occupying the first place in the vector, so that nothing can override considerations of self-interest, and other things are given some weight later on in the vector. I ust' "'constraint" here as it is used in linear programming, to mean bound on another maximizing or goal-directed process. 13. And not a path that Miss Rand would follow, I think. (Sec "The Objectivist Ethics," pp. 28-30.) B11t one cannot n1erely say, to I1Se Miss Rand's terminology, that happiness is the purpose of ethics, b11t not the standard. For the problem here is that it is known that the acrion will not achieve this purpose, when it is guided by Galt's valuing Oagny Taggart's life above his own. And in the science fiction case below, the purpose of happiness is achieved by
his not following the standard, without his making whims or irrational desires his guide to conduct. What the egoist condition on desires is which allows non-selfcentered valu ing of anoth.r's life above one's own, or the valuing of the triumph of justice and right above one's own We, bur rules out other, altnistic, de.sires as incom patible with rational egoism is a puzzle. I discuss the position in the text because when the components of Miss Rand~s view, assumed to be indissolubly intertwined, part company, it is one path that can be taken to r<eonstruct the position (and has been taken in conversation with me by persons who term themselves followers of Miss Rand, for whose intellectual meanderings she is, of course, not responsible). As I have said, it is not a path that the preponderant thrust of her view (the components of her view which, I assume, have major weight for her) would lead her to take, I think. But io that case, it is unclear what alternative teleological argument she wouJd offer for a life lived in accordance with the virtues of rationality, honesty, integrity, pride, productiveness, justice, and independence. 14. If "happy" is what it is. More likely terribly sad, hoping that they'll reach safety. 15. Privacy invasions raise an interesting problem for libertarian theory, for
it seems that we can imagine such invasions without any of the particular sorts of physical invasions which libertarians tend to focus upon. For example, suppose rhat there is one telepath who picks up emanations from you which can't be screened in. He knows exactly what you are doing, and thinking, at any given moment. Perhaps be broadcasts you on a particular T.V. wavelength. Anyone can tune in, any time, to see what you are doing and thinking at that moment. On Saturday night, aher you are asleep, there are the highlights of your week. No thought or action of yoUis is private, yet the teleparh has not irwaded you or yo11t property in any way (as libertarians speak of invasion). Are there any grounds to legally forbid the telepath f$om so operating, which the libertarian can formulate? Would we wish to forbid it in this case? Copyrighted material 386 -& Notes and References It might be said that such cases are impossible, and it is not a difficulty with a view that it does not handle some impossible case as we would wish. But it is, I think, an objection to a view if it doe.s not handle this kind of case correct!) (as it would be to show that a consequence of a moral view was that if there were anyone wbo could travel faster than tbe speed of light, then it would be morally
permissible for him to murder whomever he wished), even though the task of marking off which kind of impossibilities cause trouble for a moral theory still requires doing. 16. By inadequacy I mean that the view does not handle some particular cases as the reader, in his considered judgment, would wish to see them handled. Of course, if the reader insists on seeing each example through the principles he accepts ("Well, since it doesn't violate principle P, it's o.k.") then it will be impossible to produce what he will accept as a counterexample to the principles. Issues about how nondogmaticall) to hold principles so that the)''re open to counterexamples (as well as to counterarguments: but even here, one sufficiently attached to a consistent principle P could deny any statement Q from which not-P follows), but still to hold and accept them rather than merely contemplating them, are of great interest and merit extended discussion. Followers of Miss Rand should not scorn holding principles in this fashion, if I am right in thinking that: (a) They do nor possess a knock-down deductive demonstration of their prindples. (b) A large pan of the atttacrion of the Randian view for people is the way it handles particular cases, the kind of considerations it brings to ~ar, its "sense
of life." For many, the first rime they encounter a libenarian view saying that a rarional life (with individual rights) is possible and justified is in the writings of Miss Rand, and their finding such a view attractive, right, etc., can easily lead them to think that the particular argmnents !vliss Rand offers for the view are conclusive or adequate. Here it is not the argument which has led them to accept the view, but rather the way the view codifies, integrates, unifies, extends many of the judgments they want to make, feel are right, and supports their aspirations. If this is so, then one should hold the view so that it is open to challenge from just that son of data that bas provided its main support. Here we do well to mention a problem which bas received little discussion in the literature on ethics. Some writers on ethics have viewed their task as offering moral principles which would unify and account for the particular moral judgments we make, often adding that the reciprocal process of formulating such principles, and modifying particular judgments, itself is the process of moral justification. We may ask the philosopher of science's question: Does rhe data uniquely determine one theory which accounts for it, or are there alternative theories which equally well account for all the data we have or could
have? Asking this question forces us to clarify the notion of all possible moral data: is it, for ead1 act in each situation, a pairing of the act in the situation Copyrighted material Notes and Refrrences -& 387 with a judgment of moral permissibility? Here it seems plausible to think that alternative different theories will equally well account for all this data (as plausible as to think that alternative physic,.] theories can account for all possible observational data). Perhaps adding into the data other types of parTicular judgments (e.g., of persons' characters) will help, but, one suspectS, nor much. More promising is the claim that in the data we have not only particular judgments of particular actions bur also, often, some (partial) reason offered for rhe judgment. And so the theory must not only yield the particular judgment but also, when we have them, certain sorts of reasons for the judgmem. (I say "certain sorts" to mark a problem: if we offered as our reason for a judgment, "because it follows from T" where T is a complete fundamental moral theory, then any theory which accounted for our judgments would have to be, contain, or yield T. So it is a more delimited sort of reason which is needed.) Placing this into the data, is it still the case that there are alternative theories which equally well ac
count for the data, where no one theory accounts for and reduces all the others. If so, the various well-known positions in the philosophy of science as to the content of theories become available as options, and one may pick an option in ethics different from the one one holds with regard ro a particular theory of physics. The whole area is open, and would repay derailed investigation. 17. Independently, Thomas Nagel argues against the principle of experiential ethics in his paper "Death," Nous, 1970. 12. Weighted Voting and "One-Man, One-Vote" My comments in this essay refer to a paper by William H. Riker and lloyd S. Shapley given at the 1965 meeting of the Society for Political and Legal Philosophy. The published version of their talk appeared as "Weighted Voting: A Mathematical Analysis for Instrumental Judgments" in Representation, edited by J. Roland Pennock and John W. Chapman (New York: Atherron Press, 1968), pp. 199- 216. 1. "One wants to equalize" shall in this paper mean "one must equalize in order to conform ro one reasonable interpretation of the Court's decisions." I shall nor here be concerned with the desirability of this equalization, and shall nor discuss whether in constructing a system of rtprcsenrativc institutions one should, under all circumstances, build into it such equalization.
Nor shall I discuss the question of under what circumstances. one shouJd reconstruct an ex isting systc.m of representative institutions in order to achieve such equalization. 2~ The situation is somewhat more complicated, as is pointed out in John F. Banzhof IU, "Multi-member Electoral Districts: Do They Violate the 'One Man, One Vote' Principle?" Yale Law Journal, 75 (1966), J309. Questions paralleling those I raise arise for the more complicated formulation. Copyrighted material 388 ~ Notes and References 3. To say that important purposes are served by repre.sentarion by geograph ic.al distrit'ts is nor to say that one never should want to modify such a system. For example, if persons working in New York City bur residing elsewhere must pay some income tax to New York City, and one takes seriously the slogan "No taxation without representation," then one may think it desirable to give such persons some vote (weighted less than one) for the representatives in the New York City Council of the districts in which they work. 4. Interesting questions arise about whether in this situation the legislator's power should be proportional to rhe number of eligible vottrs, the number of registered voters, or the number of actual voters in his district. I shall not
pause to discuss these questions, nor shall I discuss the claim that the legis lator's power should be proportional ro rbe number of votes he has received, which presumably would require something like a system of proportional rep rcsentation. Though I shall not discuss systems of proportional representation, I should mention that they seem to me to raise very impo rrant questions for and perhaps about political philosophy. I conjecture that one can list conditions on presentative institutions which are jointly satisfied only by a system of pro portional representation, and which are such that each seems reasonable and desirable and indeed required by justice. But political science books tell us that systems of proportional representation work out t.erribly and so we reject such systems. Still, it may be that when we go back to our conditions, though we see which musr be rejected tO avoid a system of proportional representation and keep (an idealization of) our current system, we will feel that these conditions, which must be rejected if we are to avoid proportional representation, are con ditions of fairness and justice. Yet we are reluctant to conclude that we should sometimes choose unfair and unjust institutions over fair and just ones. I do not claim that there can be no adequate the.ory of justice to resolve this difficulty,
but merely that it pose.s a problem for and perhaps about a theory of justice. 5. A legislator might wish tO split his votes on an issue, if this is permitted by the sysrem, to reflect deep divisions within his district. 6. As I understand it, there are two (related) rationales for the Shapley power measure: ( l) The three conditions which only the Shapley measure satisfies. For a statement and discussion of these conditions, which raise some questions about the adequacy of the th.ird condition, see R. Duncan luce and Howard Raiffa, Games and Decisions (New York: Wiley, 1957), pp. 245-250. (2) The rationale in terms of random coalition formation. Does the Shapley measure yield intuitive results for aU systems of voting, e.g., for the system in wh.ich the legislators are ordered, and each may vote for a bill, against it, or abstain. If a legislator votes for {against) a bill, and all legislators before him in the order abstain, then the bill passes (is defeated). If no one votes for the bill, it is defeated. Under this arrangement, it seems that {a) each Copyrighted material Notes and References ~ 389 legislator has more power than anyone who follows him in the order, and (b) the 6rst legislator could not have more power. I believe the Shapley measure cannot yield both (a) and (b), and it would, I think, lead to the rejecrioo of
(b), thus not giving the 6rst legislator the maximal amount of power. This may lead one to question the imposition of Shapley's second condition (that the individual values of the game form an additive partition of the value of the whole game) in measuring the power of legislators in legislative bodies. 7. If one ignored everything but voting power on the floor of the legislature, e.g., that some districts would be represented on committe<:s less than proportionally to their size. 8. It is of some interest to formulate the conditions which a gerrymandered system of districts violates. If the intention of rhe person forming the districts isn't relevant (but rather the effects of certain districting) and the geographical and population-density conditions fo rmulated by political scienrists are not desirable in themselves but are meant to make unlikely the violation of further conditions, then one wants to know what these further conditions ar<. One suspects that th= further conditions may na!Urally tie in with systems of proportional representation. 9. For this example, I assume that all the legislators must vote, and that each must casr his votes as a bloc. 10. When m is the total number of legislators, x = m!. 11. Where x = m!. 12. One actually wants something more than this; namely, that a
legislator's power be between zero and one (excluding the endpoints) if he has associated with him some though not all of the persons associated with the legislators. 13. It is not, I think, clear what facts may legitimately be included in the relevant structure for determining measures of power. Should one include the fact that certain methods of election make it more likely that the rwo representatives of a multimember district will vote in the same way than that two randomly selected representatives will, that the number of political parties and the degr of parry discipline affects the likelihood of different permutations, that some representatives look to others for guidance on specific sorts of issues, that some legislative committees recommendations are rare-ly rejected, and so forth. 15. Why Do Intellectuals Oppose Capitalism? An earlier version of this essay was given as part of a lecture series at Trinity College, Connecticut. This (revised) version was submitted in 1984 for publication in the volume of essays from that lecture series but, by some accident, it was the ea.rl.ier manuscript that got published in The F11ture of Private Enterprise, ed. Craig Aronoff et al. (Atlanta: Georgia State University Business Press, 1986). Copyrighted material
390 ~ Notes and References 1. See B. Bruce-Biggs, ed., The New Class? (New York: McGraw-Hill, 1981). 2. See Mancur Olson, The Logic of Collective Action (Cambridge, MA: Harvard University Press, 1965). 3. ludwig von Mises, The Anti-Capitalistic Mentality (Princeton, NJ: Van Nostrand, 1956). 4. It is ironic that we see anti-capitalist feeling as a consequence of the schooling system, when recent radical writers see that system as molding people {or capitalism, to be docile and obedient instruction followers, acceptors of hi erarchy, keepers of schedules, etc. A school system, of course, could have both effects, whether intended or nor, molding some to fit intO the economic system and others to rc.scnt it. 5. We now can see why school athletes do not tend disproportionately to rurn against the capitalist system, eveo though they roo may experience a drop io social position after the school years. It was the informal social system which earlier treated them so well, and while they later may regret or resent the pref erences of the consumers in the market, they will not have a prior arrachment to any mode of distribution other than through the aggregate of individuals' preferences.
6. Paul Goodman, Compulsory Mis-Education and the Community of Scholars (New York: Vintage Books, 1966). 7. Nor resrrining his point to schools, joseph Schum peter discusses how "a capitalist-bourgeois sociery will find it difficult 10 bring the imellecruals to heel ... In defending the intellectuals as a group ... the bourgeoisie defends itself and its scheme of li fe. Only a government of nonbourgeois nature and non-bourgeois creed-under modem circumstances only a socialist or fascist on<>-is strong enough to discipline them. In order to do that it would have to change typically bourgeois institutions and drastically reduce the individual freedom of all strata of the nation. And such a government is not likely- it would not even be able--to stop short of private enterprise. From this follows both the unwillingness and the inabiliry of the capitalist order to control its in rellecrual sector effectively"; Capitalism, Socialism and Democracy (New York: Harper, 1950), 150-151. 16. The Characteristic Fcatu.rcs of Extremism 1. Geula Cohen, "The Passion of the True Believer," Co,gress Monthly 54, no. 5 (july/Augustl987), 10-11. Copyrighted material CREDITS L "Coercion," copyright C 1969. From Philosophy, Scimu, and Method, ed. S. Morgen~r er al., 440-4 72. Reprinted with permission of Sr.
Martin's Press, foe. 2. "Newcomb's Problem and Two Principles of Choice," copyright (I 1969 D. Reidel Publishing Company, Dordrecht, Holland. From Essays in Honor of Carl G. Hempel, ed. N. Rescher et al., 114-146. All rights re served. Reprinted by permission ol Kluwer Academic Publishers. 3. " Reflecrions on Newcomb's Problem; from the "Mathematical Games" column in Scitnri{ic A"rericatl (March 1974, 102-104, 106, 108) by Marrin Gardner, copyright C 1974 by Scientific American, Inc. All rights res~ rved. 4. " Interpersonal Uriliry Theory," copyright C Springer-Verlag 1985. Social Choice and \Tie/fare 2 (1985): 161-179. 5. "On Austrian Methodology," copyright C 1976 by Roberr Nozick. Synthese 36 ( 1977): 353-392. All rights reserved. Reprinted by permission of Kluwer Academic Publishers. 6. "Socratic Puzzles," Phro,.esis 40 (1995): 143-155. Reprinted with permission of Van Gorcum Publishers. 7. "Experience, Theory, and Language," copyright C 1986 by The Library of Living Philosophers, reprinted by permission ol Open Court Trade & Academic Books, a division of Carus Publishing Company, Peru, IL, from
The Philosophy of W. V. Quint, ed. L<wis Hahn and PauiSchilpp. 8. "Simplicity as Fall-Out," from How Man)! Questions! Essays in Honor of Sidrrey Morgenbu.ser, ed. L<igh C.uman, Isaac L<vi, Charles Parsons, and Robert Schwam (Indianapolis: Hackett, 1983), 105-119. 9. "Invisible-Hand Explanations, American E.eonomic Review 84, no. 2 (1994): 314-318. 10. "Moral Complications and Moral Structures, Natural Law Forum 13 (1968): 1-50. 11. "On the Randian Argument." Tl>t Pers011a/ist 52 (Spring 1971): 282- 304. Copyr t-Ied P'ater a 392 -b Credits 12. "Weighted Voting and 'One-Man, One-Vote,'., from Representation: Nomos X , ed. J. R. Pennock and J. W. Chapman (New York: Atherton Press, 1968), 217- 225. 13. "Goodman, Nelson, on Merit, Aesthetic," journal of Philosophy 69 (1972): 783-785. 14. "Who Would Choose Socialism?" Reason, May 1978, 22-23. 15. "Why Do Intellectuals Oppose Capitalism?" a revised version of an essay first published in The Future of Private Enterprise, ed. Craig Aronoff et al. (Atlanta: Georgia State University Business Press, 1986). 16. "The Characteristic Features of Extremism," Congress Monthly 54 (july/ August 1987): 12- 13.
17. "War, Terrorism, Reprisals- Drawing Some Moral Lines," originally ap peared as ''Total War, Nuclear Deterrence, Terrorism, ReprisalsDrawing Some Moral Lines," Re<Json, December 1978, 19. 18. "Do Animals Have Rights?" originally appeared as About Mammals and People," New York Times Book Review, November 27, 1983, 11; copy right ~ 1983 by The New York Times Company. Reprinted by permission. 19. "Fiction" originally appeared in Ploughshares 6, no. 3 (Fall1980): 74-77. 20. " R.S.V.P.- A Story," Commentary, March 1972,66-68. 21. "Testament," from "Two Philosophical Fables," Mosaic 12, no. 1 (Spring 1971): 24-28; published by the Harvard-Radcliffe Hillel Society. 22. "Teleology," from "Two Philosophical Fables," Mosaic 12, no. 1 (Spring 1971): 24-28; published by the Harvard-Radcliffe Hillel Society. Copyrighted material NAME INDEX Adler, Monimer, 383-384 Ainslie, C.,orge, 358n45, 359n I Anscombe, F.)., 351n3 Aristotle, 6 204,255-256, 2B2 Arrow, Kenneth, 369n5 Asimov, Isaac, 8.1 Aumann, Robert, 351n3, 374n31 Bali,John, 81 Bamberger, William, Z2 Banzhof,John,387n2 Bay, Christian, 2J Be<ke~ Gary, J.2Z.
Begin, Menachem, 304 Bell, Daniel, 353n 14 Benacerraf, Paul, l!6. Blaug, Mark, 3S8n40 BOhm-Bawerk, Eu.gcn von, 110. 135. 35lnl, 3S8n40 Bohr, Nils, T1 Boyd, Richard, 36Sn28 Branden, Nathaniel, 379n3 Bromberger, Sylvan, 370n3 Brown, Roger, 3S4n17 Bruce Biggs, B., J90n 1 Buchanan, james, 344n46 Burnyeat, Myles, ill Chisholm, Roderick, 377n52 Chomsky, Noam, ~ 171, 172, 178, 354-35Sn21,363n12,364nl4,377nS6 Cohen, Geula, !!. 296, J90n I Crick, Francis, 284 Dahl, Robert, 338n23 Davidson, Donald, 358n39 Debreu, Gerard, 374n31 Deonett, Daniel, 368nl Descanes, Ren~, 314, 327-328 Dreben, Burton, ill Duhem, Pierre, 157- 163, 166, 362n4 Dworki n, C.,rald, 333 Ellsberg, Daniel, 349nti Feller, William, 349nl9 Fermat, Pierre de., 276 Fredericks, George, &l Freund, Pau~ 266 Friedman, Benjamin, i8 Friedman, Milton, 101 -102, 105-106,
ill Gardner, Marrin, it, 74 Gendkr, T ama_r., l Gettier, E., ~ Gilbert, A. S., Z2 Gilbert, Walre~ 368n8 Goldst<in, Leon, 351nl Goodman, Nelson, !!, t 204, 273-276, 370n2 Goodman, Paul, 294, 390n6 Goodrich, Ralph, all Gordon, David, lJlZ H31e., Robert, 2.3. Hare, R. M., 243 Copyrighted material 394 ~ Name Index Harman, Gilbert, 368nl0, 377n56 Harsanyi, John, 96-98, 352n5 Hart, 1:1. t. A., Z, ~ 333nn2,3, 336n 12 Hausner, Melvin, 374n31 Hayek, Frederick, 110 , 284, 351nnt,2, 3SSn23 Hjelmslev, Louis, 364nl5 Hempel, Cad, .it, ~ 344n49, JS2n9, 370n3 Heppe, Robert, 2.2 He.rmsrein, Richard, J51n l, 354nn17.21 Hills, David, 352n7 Holland, John, 5 Holyoak, Keith, 5. Hom-ans, George. 97. 353n12 Honore, A. M., 1.2: 333nn2,3 Hospe:rs, John, ill Howe, Irv ing, 304
Irwin, Terence, 146. ill Jeffrey, Richard, 346n4, 348n I 0, 349n20 Kahneman, Daniel, 103. 1.!li Kalelkar, Mohan, Z6 Kanr, lmmanue.l, :l.i Kaplan, Abraham, 338n23 Ka!z,Jerrold, 172, 363nnl 0,11, 364nl5 Kepler, Johannes, 276 Kim, Jaegwon, 3SSn21 Kirz:ner, Israel, .351 n 1 Kraott, David, 358n38 Kripke, Saul, 164. 276n, 354nl9, JS5n22 Kruskal, Marrin David, ~ill Kuenne, Robert, 3S8n40 Kuhn, Thomas, 1Jl Lachman, Ludwig, 3Slnl L.asweiJ, Harold, 338n23 Lcibniz, G. W., 284, 361n15 Levi, lsaac, 3.S7n34, J77n53 Lewis, C. S., !l2.=..8.J. Lewis, David, 357n33 Linder, Staffao, 358n44 Lubitz, Raymond, ill luce,, R. D., 344n46, 349nn15,!9, 352nn5,6, 358n38, 374n31, 375n36, 388n6 lupcr-Foy, Steven, 360n8 MacCorquodalc, Kenneth, 355n21 Mandelbaum, Maurice, 351 n2 Markowitz, Harry, 102-103. 10$-106 Marx, Karl, JSJn 14 McKen:r.ie, Richard, 3S2n8 Meehl, Paul, 35Sn21 Menger., Carl, 110. 114. ~ 351nl,
352nnl0,11 Jvllles, Richard, 8.0. Mill, John Sruart, 10.i Mills, C. \Vright, 1.0 Mise, Ludwig von, 110. 119. 120. 122. 125. 129, '132-133. '135-136, '138. 285-286, JS!nl , 352nl1, 354nl8, 356nn27,29,30,3t, 358n43, 390n3 Morgenbesser, Sidney, .l-4. 2.., 3S7n34, 366 Morgenstern, Oscar, &2 Nagel, Ernest, 2, 344nSO, JS I n2, 3S2n9, 356n24 Nagel, Thomas, 387n17 Nash, John, 22 Newcomb, William,1. 45-46, ill Newton, Isaac, 273-276, 284 Nietzsche, Frederick, 11 Nisben, Richard, S. Olson, Mancur, 344n46, 390n2 Oppenheim, Felix, 333nl, 344n51 Pap, Arthur, 3S4n19 Pauling, linus, 284 Peirce, C. 5., 361n15 Pitkin, Robert, 8!t Plaro, 150, 260, 283, ill Poincare., Henri, 364nt3 Putnam, Hilary, 123, 164, 363n6, 367n4 Copyrighted material Quine, W. V., [, 123, 156-181, 183, 188, 354n19, 356n2S, 357n34, 358n39, 363- 366, 366nn1.2. 367n3 Raifla, Howard, 344n46, 349nn15,19, 352nn5,6, 374n31, 388n6
Rand, Ayn, 7-8 249-264, 378=387 Rawls, john, 1 Reganl Tom, ~ 305-310 Riker, William, 79. 266-267, 270, ill Rosenwald, Kun, 8.3: Ross, W. 0., :ill6 Rothbard, Mutray, 119, 135, 351nl, 356nn28, ,1Q. 358n38 Saussure, F. de, 364nl5 Savage, L J., '101 - 102, 105- 106, 133, 349n20,35!n3 Schelling, Thomas, '191, 34 Ln39, 352n5 Sch.r:Odinger, Erwin, 76-77 SchumJl"t<t, Joseph, 353n 14, 390n7 Sen, Amanya, 93, 358n37 Shapley, Lloyd, 266-268, 270, 387-389 Sidgwick, Henry, 3 70nl Skinnei, 8. F., 12 1-122, 125, 172, 175, 179, 306, 354n21, 365nn23,24, 366nn30,31 Sklar, Lawrence, 3Sln2 Smith, Adam, W S<x:rates, 2-3, l.L. 145-155, 359 362 Sophocles, 26!1 Name Index -& 395 Sosa, Ernest, 3 77n52 Suppt:s, Patrick, 344n49, 358n38, 373n16, 375nn34,35,36 Tarsk~ Alfred, 1.&Q Thaprd, Paul, .1 Thomson, j udith, 374n28 Tukey, John, 375n36 Tullock, Gordon, 344n46, 352n8 Tversky, Amos, 103, 105 , 358n38
Ullian, joseph, 183, 1.8.8 UllmaoMargalir:, Ed_na, W VJasms, Gregory, 145, 146, 147, 150-153, 359-361 von Ne.umann, john. ~ 374n31 Walzer, Michael, J... 300-304, l8ll Watkins, J. N., 351n2 Warson, James, 284 Weiser.. Frederick von, llO. Weiss, Stephen, Z6 Wertheimer, Alan, ~ Whiting, Nathan, 0: Wilson, E. 0., 353nt5 Winet, Muriel, 375n36 Wiugenstein, ludwig, W Zeuthen, Hans, 96, 28 Zinnes, joseph, 344n49, 375n34 Copyrighted material Copyrighted material SUBJECT INDEX action: theory of, 111 - lJSj altcrnar.ive, 211- 220, 250-252; larger courses of, 225- 232. &e also reasons for action aesthetic muir, 273-276 alt.rasia, 3S9n1 amends, 2 3 l-2 34 anchoring.. 8:2 animal rights, ~ 305-110 a priori knowledge~ 193; theory of action, 119- 126 Archimedean axiom., 87-88. 11lZ artifacts~ 18}-184 Austrian onomic theory, 110-14 1, 35(}-359
backwa_rds cau.salicy, ~ an lla.ldwin effett, W bargaining, 9 S 1 00 behavior, 120-123 belief revisionl t Sfbl 52 blackmail, 340n33 book reviews, Z l!lO burden of proof, :a4 capitalism, Z. 249, 280-295 causal decision theory, 62-65. Z8 choicr, 5-6. 9-10. 37-44 45-84, 126-135. 250-252 coercion, ~S-6. L 15-44 333-345 Columbia College, H 2dJl commodit), 122-128 conditional expected utiJity, ll conditioning. 12lk122 constitutive conditiON, ill context, ~ 308-309 cost, 129- 130, 13!b139 cranks, 305-306, 310 decision theory, 3-4, 45-84, 19.1-196, 202, 362n3; causal, ti2dU deductive moral structure. 204-202 definition, 147-148, 337nl8 descriptions, 128 122 determinism, 82:83 disposition, 131. 173-174 disproportion, 299. 305- 306 distribution, 286-293 donlinance principle, 40. 46-84. lli double-discounting, 140-141, 1.22 doubleeffect, 301-302
economic tbeory, 85- 109; methodology of, 1 10-141 egoism, 26.1 elench,.s, ~ 149-152, lli embodimen4 method of. 154:1 55 entitlement, sense of, 286-293 essence. lli ethics, 201-264, 300-310, 321-323 evil, problem of, lli evolutionary theory, t, 117. 140-141. lSS-159, 191-193, 379n2 examples in philosophy, 66.-70 exceptions, 205- 207 exchangc, 23=24. 129-130 Copyrighted material 398 -i> Subject Index expecred course of events, 24-31 . 339n28, 340nn31,32 expected utility, 48-84, 101-103; conditional, iS experience, I 57-169 experiential ethics, 262- 264 explanation, 9- 10. 60-61. 67-70, 131 , 172-176, 187-190, 28().-295, 354n18, 370n3. &~also invisible hand explanations; reduccion exrnmism, ~ 296-299 fact$, 157- 158 fiction, philosophical, !Q. 313-330 fix.ed stares, iZ::6.1 gamble, 10!- 103 game theory, 112-114 God, 316, 324-330 grammar, 170-174
guerrilla warfare, 303- .304 happiness, 2 6 2- 264 help chain, l.2.i. human being, 307-309 hypotheses, 184-190 inconsistency. 150:-1 51 independent measurement procedures, 88-89, ~ 107-109 inderenninacy of translation, 161-167 indiffetence, 93- 95, 127-135, Sua/so preference inRecrion pointS, 105. 107-1 08 institutions, 35-36, 114- 117 inrellcctual, Z., 28().-295 interaction, 111-1 15; rwo~person, 111113; three-person, 1 13; nperson, 1 13- 11 s interpersonal comparisons- of utiliry, .L 85-109; procedure, 85- 93; l e-ro point, 93-95; unit, 95-109 imersteUar communications, 317-32 3 inrerval scale, 87, 93, 213,220-224 invisible hand explanations, L 9, 187190, 191- 197, 367n7, 368nl1, 368- 370 Israel, 277-279, 302-303 kibbutz, 277-279 knowledge, 146- 148, 1.10-l.SI labor, llZ language, 157-1 81 law, 2!16. legislation, 36, 26.1 270 leisme, 1.36==1.\7 libertarianism, 7-8. 300-303
liberty, 15-16, 31-36 life, 250-261 linguistics, 169-179, ill logicaJ positivism, ~ ill lying, 211- 212, 216 217 marginal utility, U6 market, 120, 195, 28().-295 Marxists, 2.8.3 maximization moral sr.rucrure, 202-204. 379n4 Maxw~Ws d~mon, 8.1 meaning, 157-169; experiential, IS7-163; stimulus, 167-169, 178-179 meaning of life, 329-330 measurement, 220-225 mental retardat-ion, 306-308 metalanguage, 180-181 method: of examples, 66-70; of economics, 110-1 41 ; Socratic, 149152, iSS; of embodiment, 154-155; and simplicity, 182- 190 methodological individualism, ~ 110-118, 119, 352n7,369n5 minimal state,, 19 1, 379-382 Model Penal Code, 236- 237 money, 10().-109, 114, 352.nll moral: structures, 4-5, 201-248, 37()... 378; impermissibility, 204-248; permissibility, 207-248, 301-302; Copyrighted material moral (continued) asymmetry, 218-220, 231; undercutting, 227-232; amends, 233234; completeness, 242; spccjalncss,
255-256; generaliurion, 257-258; status of species, 307- 309; context, 308-309 moral weight, 1, 209-248; extra, 2B215; measurement of, 22(}-225, 231 necessary truths, 12G-123. 354nl8 neoconse.rvatives, ill Newcomb's Problem, 1, 45-$4, 196, 345-350 night-watchman state, 380=382 noncombatants. 302-304 nonnative principles, .L 6-7. 201-248 nuclear weapons, 304 offors, 5-6, 23-31 , 339-340, 343nn44,45. See also preoffer sinuuion; threats ontology, 313-316, 365n29 operant conditioning, 115. 121-122. 125. 306, JS4n21 other minds, 324-328 outweighing, 207-248 overriding, 232- 239 paternalism, 35-36. 342n43 phoron, 1 3()...1 31 poem, ~ 224-225 political philosophy. t-2. 154, 191 , 265-270, 362n20, 38G-382, 388n4 powe~ 26.5-270 pudiction of choices, 4 S-84 preference, 1 tO, 126-135, l2li. See auo ind ifference; rime-preference preoffer situation, 40-44 prima facie duties, 206-210 Princeton University, i, 2 principles, 42, 48-50, 55. 57-58, 62-64,
@. 201-248, 297, 301-302 prisoners' dilemma, 40, 61~2. 283-284, 344n46 Subject lndex -& 399 privaq, 263. 385nl5 probabilities, 45-$4, 123i probabilistic indepen.dence, .iJ psychology, 119-126, 135-141, 175- 179 public goods, 344n46 purposive behavior, I 20-12 3 puzzles, 37-39 45-48 74-75, 85-$6, 111-114, 135-141, 145- 155, 157, 163-166, 179- 181, 182- 183, 203- 206. 255-256. 281-282, .103-304, 307-309, 313-316, 324-328, 329-330, 340n31, 379-382 rondomness, UD. ratioruality, 193-194, !96, 2.S.S-256; of belief, I.t 196i conditions, 131 - 1 34 rational man, 39:::44 reasons for action, 17-18. 33-34, 36-37, ~ redjsrriburjon, 380::382 reducdon, 111-118. 156, 1.6..S. reliability, ill retribulive punishment, 92-93, 361nl8 Riggs v. Palmer, 371!16 right .. making features; 207-248 rights, 133, 258-261 , 294, 301-310, 379-382 risk, 105-106 Robinson CruSOt theory, 111-I 14 ruling class, W school, 286-295
SchrOdinger's cat, 76- 77 self-interest, 249-264, 283, ill selfsacri6ce, 261-264 Shapley power index, 6. 388n6 side-constrainrs, 379n4 simple moral structure, 20 7-210, 2.18. simplicity, . z. 182-190, 366-368 skepticism, 314. 324-328 socialism, Z, 277-279. 283- 284 social science, 118. lli sociology, 286-293 Socratic ignorance, 14 S-148 Copyrighted material 400 -& Subject Index Sa<:raric method, 149- 152, 155 soul, state of, 152-155 specialness, 255-256 species, moral status of, 307-309 stimulus meanings, 167- 169~ 178-179 strategy, 32-34 subjunctives, 115, 129-134, 151, JS3n13, 357n33 suboptimal panems, 34-36 sunk costs, 120 teleology, 68- 70 terrorism, 298, 302 threats, 5-6, 16-22, 23- 31, 31-44, 9697, 334n7, 338- 342. See also offer>; warnmgs time-preference, 110,135-141,192 tips, 36 tracking, 147- 148, 151 transitivity, 209, 357n35 translation, 175-177
truth, 179-180; and simplicity, 182-190 two-person inte raccjon, 111-114 uncertainty, 138-139 undercutting, 227-232 underdetermin.ation, 157, 163-167, 179- 180,184-187,327,386-387 utilitarianism, S, 203-204, 379n4 utility theory, 3, S, 202, 203-204; and choices, 37-44; interpersonal, 85- 109; and money, 100-109; shaping of utility, 117-118; diminishing marginaJ utility, 126; of future: reward, l39-140 values, 250-261 verstehen, 123- 125 voluntar)' action, 37-44 von Neumann- Morgcn.stern conditions. 87- 88, 98- 100, 101- 102, 104-106, 108, 133, 221 voting system, 265- 270, 387- 389 warfare, 300-304 warnings, 17, 31 - 37. See also threats weakness of will, 359n I weahh maximization, 192- 193 weighted voting, 265-270, 387-389 will, 37-44 wrong~making features, 207-248 tero point, 87, 93-95, 107- 108 Copyrighted material Copyrighted material (. .. ,_ I ,. ''
, .. -~. / ' .. . ;r . :~-~.. u ' ' ' If " -<( .-., ' ' ' ~ ''. ~ v ' ,, ; ~ . '. ._ .. "' . ---fT.~--- _.,-. ... : .-: . .. ' w .. !";'!.' ' ~ ~ ,, .1!!!: '
'' .... -' ' .. ' ' n '' ' ;,. ...; .:~ !. ' . ' .-.. . -, --~ ' . ' . ... ' - 'r _..' -~ :'~ . ;; ' ; ,, !i' c ', , .. , -.,_. ',_ I' . t~ . . , , .

Socratic Puzzles

Uploaded by

Copyright:

Available Formats

Socratic Puzzles

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Socratic Puzzles

Uploaded by

Copyright:

Available Formats

......

2 Newcomb's Problem and Two Principles of Choice 45

3 Reflections on Newcomb's Problem 74 4 Interpersonal Utility Theory 85 5 On Austrian Methodology 110

...... SOCRATIC PUZZLES INTRODUCTION

CHOICE AND UTILITY

Copyrighted material Copyrighted material

consequence if Q doesn't do A, and Q does one of the B1 in order

A I> and if he prefers doing one of the old alternatives (A 1) to doing

normal course of events in assessing whether we have a tbreat or an

Copyrighted material 28 -f> CHOICE AND UTILITY

strongly tempted to say, when Q pays P $10,000 because he accepts the

B. Goes out of business

I. Union wins II. Union loses (b) (a)

Copyrighted material 32 -& CHOICE AND UTILITY

Copyrighted material Coercion -& 41

If we concentrate solely on the choices made in the threat and offer

2 NEWCOMB'S PROBLEM AND TWO

of Pure Reasm-t, .\41.1., B4 50

You have a choice between two ac.tions:

Thus we get the matrix, T. A: 0)2 OJ Ot9 06

A and B, and the states Sh ... ,

s. are contrived states, which some

Newcomb's Problem and Two Principles of Choice ~ 53 St l win $50 I lose $6

the other action. Either tendency is genetically dominant over a

expected utility) No dominant action is available (II) Maximize

You might also like