7/31/2011 Work in progress, often updated and revised. Page numbers in Table of Contents may be inaccurate.

For latest update, completeness, and font readability please check at both http://myweb.uiowa.edu/butchvar and http://butchvarov.yolasite.com. The latter is more reliable. COPYRIGHT @ 2010 by Panayot Butchvarov. All rights reserved. Brief quotations are permitted if source is fully acknowledged.

ANTHROPOCENTRISM IN PHILOSOPHY Realism/Naturalism/Antirealism Panayot Butchvarov University of Iowa TABLE OF CONTENTS

Notes 401 15 Chapter One: INTRODUCTION 15 Part One: EPISTEMOLOGY AND ETHICS DEHUMANIZED 49 Moore's Principia Ethica was published in 1903, and soon became one of the signposts to the philosophy of the following 100 years. It is too early to judge how 20th century philosophy ended, but its beginning was remarkable. Both Moore's Principia Ethica and Russell's Principles of Mathematics appeared in 1903, the first volume of Husserl's Logical Investigations in 1900-01, and four of William Jamess major philosophical books in 1902-09. Wittgensteins Tractatus Logico-Philosophicus was written between 1914 and 1918. There was not a significant difference, except in style and temperament, between Anglo-American and European philosophers. The analytic/continental schism came much later. Both Russell and Husserl began as mathematicians. Frege was the German philosopher-mathematician to whom, by their own admission, Russell and Wittgenstein were both heavily indebted. In the Preface to Principia Moore wrote that his ethics was closest to the Brentanos. Russell studied and discussed Meinong in detail. James was much admired in Britain and in Europe, influenced Husserl and Wittgenstein, and was the subject of articles by Moore and Russell. 86 Principia Ethica remains the best example in 20th century philosophy of the sort of metaphysical ethics that is free of anthropocentrism. Moore dehumanized ethics even more clearly than Plato had done. But his book also inaugurated what has come to be known as analytic ethics, the sort of ethics characteristic of 20th century analytic philosophy. The relation of analytic ethics to Moores book rested largely on his thesis in Chapter I of Principia Ethica that the property good is indefinable. But Moore made clear he had no interest in what he called verbal and the tradition calls nominal definitions. They are the business of lexicography, he wrote. Yet it was just such definitions that analytic philosophers sought, sometimes calling them analyses. The most familiar example comes from analytic epistemology, not ethics: the definitions of S knows that p, which preoccupied epistemologists from the late 1960s to the early 1980s. The definitions sought and offered by analytic philosophers were not even lexicographic definitions, which record lexical facts and are tested by empirical investigation of speech and writing. Rather, they recorded their authors impressions of lexical facts, and were tested by the authors intuition of what would in some hypothetical situation, called a counterexample if it did not fit the intuition. The question How do I know what one would say in that situation, given that I am not in it? was usually ignored. For it could be answered properly only by appealing to what I or others have said in similar situations, and this would be to make an appeal, however amateurish, to lexical fact. Even the 1

Oxford English Dictionary is valuable mainly for the examples of usage it lists, not the definitions it distills from them. 86 Indeed, searches for definitions are alien even to current philosophy of language, because of three trailblazing developments in it half a century ago. The first was Quines attack on appeals to meanings generally. It is widely accepted today, but usually only pro forma. Phrases such as conceptual question, conceptual content, and conceptual connection still abound in the literature, with the noun concept explicitly used for the meaning or use of a word, not for a mental or a brain state. The second development, also widely accepted but just pro forma, was Wittgensteins relentless argument in the Philosophical Investigations, posthumously published shortly after Quines article, that words are not used in accordance with necessary and sufficient conditions. He gave game as an example, but his argument applies also to good, right, reason, know, exist, and other denizens of the philosophers lexicon, which, like game, are everyday words, not technical terms introduced as abbreviations of multi-clause descriptions. The third development was Chomskys linguistics, announced four years later. It marked a striking advance by stressing the biological, largely inherited, core of linguistic competence and urging the use in the study of language of the standard methods of scientific research. 87 Despite its inattention to what he meant by definition, analytic ethics did begin and develop in relation to Moores ethics, though by way of sustained disagreement, not agreement. Discussions of Principia seldom ventured beyond Chapter I, which alone was included in most anthologies. Usually ignored were the crucial Preface, where Moore explained what he meant by intuition and selfevidence, and thus what anyone calling him an intuitionist and foundationalist ought to mean. Also usually ignored were the beginning of Chapter II, where he explained what he meant by natural and nonnatural, thus what anyone calling his ethics nonnaturalist ought to mean, and Chapter V, where he explained his theory of right on the basis of the theory offered in Chapter I. 89 By intuitions, Moore wrote, he meant self-evident propositions, and nothing whatever as to the manner or origin of our cognition of them. And a self-evident proposition, he explained by plainly following the etymology of the word, is one that is evident but not by virtue of inference from other propositions. Moore did not say what he meant by evident, perhaps thinking it unnecessary. A proposition is evident, of course, if it is, or can readily be, seen to be true, either literally or metaphorically. Therefore, it may also be said to be known, in the serious and traditional sense of know, which sharply distinguished between knowledge and belief or opinion. The noun evidence, as used in court or in the lab, has a wider meaning, but the same root. Moore used self-evident for the propositions in ethics that state what kind of things ought to exist for their own sakes, i.e., are intrinsically good. He avoided the term justification, which became common in later analytic epistemology as a way of avoiding skepticism but acknowledging that little of what we say we know passes any traditional test of knowledge. 89 While we sometimes call it thinking to accompany a sentence by a mental process, that accompaniment is not what we mean by a thought. Say a sentence and think it; say it with understanding: And now do not say it, and just do what you accompanied it with when you said it with understanding! 149 A priori knowledge has also been defined as prior to experience in the sense of being innate. But would this mean that one who knows arithmetic, the paradigm of an a priori discipline, knew it actually before birth? If it means only that one knew arithmetic potentially before birth, then all knowledge is innate. Another definition is that a priori knowledge is knowledge based solely on reason. But what is reason? Presumably, not reasoning in the sense of inferring: a proposition known a priori may be self-evident, known without inference. By reason many mean intellectual (rational) intuition, a certain kind of awareness. Such awareness need not be supernatural or even unfamiliar. If you hear the door bell ring three times, you may be aware of how many times it rang, but you do not hear a number. When you see a books color, you may be aware of the generic property color, but not by seeing it in addition to seeing the books color. If you feel joy, you may be aware of its goodness, but you do not feel that goodness as you feel the joy. Such awareness is intellectual in the sense that it involves intelligence, not sense perception or introspection, though these also are kinds of awareness, and in that we can think about, but not picture or image, its objects. A posteriori and a priori knowledge are thus both based on 2

awareness, though on radically different kinds of awareness. This should not be surprising. What would be surprising is if one could have knowledge without ever being aware of what one knows. 166 It might help to consider arithmetic, the paradigm of an a priori discipline. Mathematicians do not observe or feel the numbers 2 and 4, nor do they perform experiments. Of course, they learn the names of numbers empirically, which is true of all language-acquisition, but mathematicians do not think that their knowledge of propositions (equations) such as 2 + 2 = 4 is based on sense perception or introspection. Is it then based on intellectual intuition? If not, are mathematicians aware of anything that is relevant when they consider an equation? Some say that mathematics is concerned only with words and symbols. But is mathematical knowledge then just a matter of words or symbols? 167 1. Logical experiences. 270 270 Chapter Ten: FACTS AND TRUTH 288 David Armstrong, A World of States of Affairs (Cambridge: Cambridge University Press (March 13, 1997), 136. 437 As 5.633 (Where in the world is a metaphysical subject to be found?) makes clear, Wittgenstein's reason for 5.631 (There is no such thing as the subject that thinks or entertains ideas) was not idiosyncratic. It was at least partly phenomenological, like Hume's and, later, Sartre's. The metaphysical subject is not to be found. A simple experiment might confirm the claim. Let the reader focus, not on this page or something remembered or imagined, nor on the readers nose or eyeglasses or anything else the reader happens to see, but on what sees this page. Few have claimed success in such an experiment. Few have claimed to find what sees this page, the subject of the seeing. Nor was the positive view of the self that Wittgenstein offered in 5.63 (I am my world) idiosyncratic. His description of the relation between the self and the world was not entirely unlike Humes (a heap or collection of different perceptions) or Sartres (an absolute, impersonal consciousness that constitutes the world). 338 If one finds these views absurd, one probably thinks of the self as oneself, say, P.B. Indeed, the assertion that P.B. does not belong to the world, much less that P.B. is the world, is absurd. But P.B.s existence is not in question. What is in question is the existence of the self to which our several impressions and ideas are suppos'd to have a reference that Hume looked for but could not find, of the subject that thinks or entertains ideas that Wittgenstein also could not find. P.B.s existence is not denied, jusy as Hume dd not deny Humes existence and Wittgenstein did not deny Wittgensteins existence. What is denied is the existence of that philosophical self. That there is no such entity is phenomenologically evident. It should be evident also from the grammatical monstrosity of expressions like the I (das Ich), I-hood (die Ichheit), and I-like (ichlich), which have marred some philosophical writings. 339 Although there is an obvious nonphilosophical distinction between P.B. and the world, there is no philosophical distinction between me and the world. The reason is that in the latter case each must be understood in terms of the other. The point without extension to which the self shrinks, according to Wittgenstein, is what determines the limit of the world, of the reality co-ordinated with that point. To say this indeed is to express a sort of solipsism, but a solipsism that coincides with pure realism. Such solipsism is not alarming, because the I in I am the world refers, not to Wittgenstein, P.B., or any other thing in the world, whether physical or mental, but to the world itself, even if we add, which Wittgenstein did not, that it refers to the world as cognized, viewed, in a certain way. It is solipsism because it says that only the world exists. It is not the traditional solipsism that might say that only P. B. exists. There is no such thing as the philosophical self, there is just that world. To say this, as Wittgenstein saw, is indeed to endorse a sort of realism, a pure realism. It is also to show how antirealism may be freed of anthropocentrism. The impression of absurdity plaguing antirealism vanishes when the implications of its most extreme variety, solipsism, are followed out strictly. 339 There is no room left for any sort of subjectivism, be it traditional solipsism in metaphysics, skepticism in epistemology, or egoism in ethics. There is no room for traditional solipsism because the self is only a point from which the world is viewed, and has no independent ontological status. There is therefore nothing relevant with which to contrast the world, nothing that would be left if the world did not exist. Without the contrast between me and the world, the solipsists assertion Only I exist becomes empty. It could mean either that only the world exists or that nothing exists, neither of which is what the 3

solipsist intended. Solipsism thus loses whatever content it seemed to have. It seemed to say that only I exist, not that either the world or nothing exists. 340 There is no room for skepticism about the external world: Scepticism is not irrefutable, but obviously nonsensical, when it tries to raise doubts where no questions can be asked (6.51). It is nonsensical because, again, there would be nothing with which the external world could be contrasted, nothing to which it would be external. And no room is left for egoism in ethics. We saw in chapters 3 and 4 that the implication for ethics of rejecting the philosophical self goes beyond the rejection of ordinary egoism. A complete rethinking of the subject matter of ethics becomes required: The sense of the world must lie outside the world. In the world no value exists... (6.41). 340 The fatal weakness of all subjectivism whether solipsism, skepticism, or egoism is its implicit commitment to a pre-Humean view of the self. Without a subject, there cannot be a subjectivism. Hume, Wittgenstein, and Sartre saw this clearly. Their rejection of subjectivism, though highly technical, was in agreement with common sense, which firmly disapproves of egoism, rejects skepticism without hesitation, and dismisses solipsism as madness. But what is directly relevant to us here is that if Hume, Wittgenstein, and Sartre were right, then the distinction between me and the world could not be taken at face value. To be employed in philosophy, it must receive a radical, nonanthropocentric reinterpretation. 341 1. Consciousness. 341 The rejection of the self is the first of the four steps to free antirealism of anthropocentrism. The second step, which Hegel and, much later, also Moore, Wittgenstein, and especially Sartre in effect combined with the first, is the rejection of consciousness if understood as an entity, presumably a relation between a self and its objects, that is distinct from those objects. The first step is a rejection of subject-object dualism, the second is a rejection of act-object dualism. Indeed, the former entails the latter. If consciousness is a relation between subject and object, and if there is no subject, then there is also no such relation. But, like the first step, the second step can be defended also on phenomenological grounds. 341 Just as the ubiquity of the pronoun I, which, as Kant noted, encourages the philosophical thought that there is a self, an ego, a special entity that engages in activities such as perceiving, thinking, and imagining, the ubiquity of the verbs perceive (and more specific verbs like see and hear), think, and imagine encourages the philosophical thought that there is consciousness, awareness, a special entity of which perceiving, thinking, and imagining are species, which is the selfs intending, its being directed upon, objects. Taken together, these two thoughts encourage a picture of consciousness as an arrow shot by the self and aimed at objects external to it. The first two steps reject this picture. Of course, they do not deny that there are people or that people perceive, think, and imagine. They reject the philosophical conception of thinking, perception, or imagination that rests on that picture. 342 There is no phenomenological ground for believing that there is such an entity as consciousness, just as there is not for believing that there is a self. The reader is invited now to attempt to find, not the entity that sees this page, but rather the readers seeing it the seeing itself, the visual consciousness. I believe few would claim success. Few would claim to find anything relevant in addition to the page. Hegel wrote that the truth of [self-certainty] contains nothing but the sheer being of the thing [Sache]. Heidegger glossed this as the truth of sense-certainty is not concerned with the knower, consciousness, or the I. 342 Wittgensteins views, mentioned earlier, about solipsism and skepticism were grounded in his rejection of the philosophical self, but they gained independent plausibility from the conception of consciousness defended earlier by G. E. Moore in the article The Refutation of Idealismand, much later, by Sartre in The Transcendencce of the Ego and especially Being and Nothingness. Wittgenstein presumably was aware of Moores article almost everyone in British philosophy during Wittgensteins first stay at Cambridge was. Moore argued in it that consciousness is diaphanous, translucent: [T]he moment we try to fix our attention on consciousness and to see what, distinctly, it is, it seems to vanish: it seems as if we had before us a mere emptiness. When we try to introspect the sensation of blue, all we can see is the blue: the other element is as if it were diaphanous. In general, Moore wrote, that which makes the sensation of blue a mental fact seems to escape us: it seems, if I may use a metaphor, to be transparent we look through it and see nothing but the blue. 343 4

Several decades later, Sartre eloquently described the role of consciousness in constituting the world as the revelation of the objects conceptualized as a world. Consciousness exhausts itself in its objects, he wrote, precisely because it is nothing but the revelation of them: Consciousness is outside; there is no within of consciousness. Indeed, as we saw earlier, Sartre held that consciousness is always also self-consciousness, but not in the sense of being consciousness of a self, or even of another state of consciousness, but rather in the sense of being nonpositionally (nonthetically) conscious of itself by being, so to speak, a conscious consciousness, e.g., a conscious seeing of this page or a conscious, as contrasted with an unconscious, desire. Sometimes a state of consciousness is positional, thetic, self-consciousness, meaning by this that it has a consciousness as its object, not itself but another state of consciousness. Even then, however, consciousness, though positional, exhausts itself in its object, which in turn exhausts itself in its own object. Whether perceptual or conceptual, consciousness is not a thing. One may even go so far as saying that it is nothing. To use a word Heidegger had applied, consciousness is only the lightening of its objects, like the coming of dawn, which lightens, reveals, the rocks, bushes, and hills that had been invisible in the darkness of the night, but is not itself an object. 343 If Humes and Wittgensteins denial of the existence of a self, and then Moores and Sartres conception of consciousness, are accepted, we can grasp more clearly the conception of ourselves needed for understanding the key thesis of antirealism that, as Nelson Goodman put it, we make the world. 344 We make the world is, of course, a metaphor. It means that what the world is for us depends, at least in part, on our cognition of it. It depends not only on how we perceive but also on how we conceptualize and describe it. As I suggested in the Introduction, the thesis of antirealism, so understood, is not entirely foreign to common sense. We can easily understand what would be meant by saying that the world of the fly is quite different from our world, and that the world of the octopus is even more so, as long as we are aware that the sense organs of a fly or an octopus differ radically from ours and that therefore a fly or an octopus perceives the world not at all as we do. We can easily understand what would be meant by saying that our world is probably quite different, perhaps indescribably so, from the world of an intelligent extraterrestrial life form, as long as we are aware that almost certainly it would be not only perceived but conceptualized very differently. Even radically different human cultures are often said to have, live in, different worlds because the concepts through which they understand the world are different. 344 While Kant held, roughly, that the world is dependent on our consciousness of it, Hegel held, also roughly, that the world is that consciousness. Their positions are likely to be found more plausible if we do not limit consciousness to perception, and then if we suppose that perception necessarily involves conception. The defect of empiricism, Hegel wrote, is that it makes sense-perception [Wahrnehmung] the form in which fact is apprehended, but the process of knowledgeproceeds to find out the universal and permanent element in the individual apprehended by sense. This is the process leading from simple perception to experience [Erfahrung]. For, in propositions where the subject-matter is due to the senses e.g. This leaf is green we have such categories introduced, as being and individuality. More than a century and a half later, Wilfrid Sellars argued against the myth of the given, and held that all cognition presupposes econceptualization, that sense perception as such grasps no facts. To say that perception involves conception need not mean that perception is propositional, that statements of the form S perceives x entail statements of the form S perceives that p. It need only mean that statements of the form S perceives x entail statements of the form S perceives x as F. But surely this is false of neonates, and perceiving-as must not be confused with perceiving-that. A neonate first just perceives (sees, feels) Mother, and neither perceives Mother as the neonates mother nor that Mother is the neonates mother. The latter are later stages of the chilkds cognitive development. Presumably perceiving Mother is first followed by perceiving Mother as the chilss mother, and perceiving that Mother is the childs mother comes later, perhaps never if it presupposes the acquisition of language. 345 Antirealism faces the paradox of seeming to say that the world is dependent, at least for its nature if not also existence, on some of its zoological parts whether just on me or on us. I have argued that there is no special entity self, ego, subject denoted by the first-person singular pronoun. I have also argued that to avoid the paradox we must understand first-person singular pronouns (I, me, mine), when used in philosophical contexts such as Cartesian doubt and the realism/antirealism debate, as impersonal. I shall argue now that in such contexts the first-person plural pronouns should also be understood as impersonal. 348 5

The choice between I and we is often stylistic. A politician may use the latter in order to avoid showing conceit. It is often stylistic even in philosophy. In epistemology we speak of skepticism regarding the limits and extent of our knowledge, though what we actually consider is usually the extent and limits of my knowledge. But the choice is not stylistic in ethics, which from Plato to Hobbes to Sidgwick to Rawls has been preoccupied with egoism, the view that advocates the pursuit of my own good rather than of our good. Nor is the choice stylistic in stating the antirealist thesis and understanding the metaphysical doctrine of antirealism. 348 I have argued that in philosophical contexts such as Cartesian epistemology and the realism/antirealism debate the first-person singular pronoun should be understood as indicating a perspective, a view, a cognition, not a person. Would that cognition be my cognition? In such contexts, indeed in any advanced context, a cognition is inherently social. My worldview is the same as our worldview, just as my view of Manhattan is the same as our view of Manhattan when both are the view of Manhattan from the Empire State Building, the one that, say, a tourist guide describes. In the philosophical contexts just mentioned, there is no difference between the role of first-person singular pronouns and the role of first-person plural pronouns. The distinction between me and us lacks significance in such contexts. 348 Needless to say, this is not true in most nonphilosophical contexts. In everyday life as well as in science, usually I indicates, say, P.B. and we indicates P.B. and you. The difference is obvious. P.B. weighs 160 pounds, P.B. and you together weigh 300 pounds. P.B. visited Charleston last year, we did not. But in doubting the existence of the world or considering the question of its mind-independence, there is no relevant difference between me and us, between my doubt and our doubt, between my stand and our stand on the question. There are such differences only when we assume that P.B. and you are parts of the world, a natural assumption in other contexts but one that cannot be made when doubting the existence of that world or considering whether it is mind-independent. 349 In the philosophical contexts in which the use of the first-person singular pronoun I is impersonal, the use of the first-person plural pronoun we is also impersonal, and for the same reason. But, though required in those contexts, the impersonal use of we is not limited to philosophy. It is common in science, as in We know that the speed of light cannot be exceeded, the impersonal sense of which becomes explicit when restated as It is known that the speed of light cannot be exceeded, Physics has found (established, discovered, confirmed) that the speed of light cannot be exceeded, or According to physics, the speed of light cannot be exceeded. An impersonal use of we is also common in moral contexts, as in some occurrences of We dont torture prisoners, the impersonal sense of which becomes explicit when restated as Torturing prisoners is unacceptable in this country or Morality prohibits the torture of prisoners. It is common also in etiquette, as in We dont wear flip-flops at weddings, which means that wearing flip-flops at a wedding is a violation of proper dress rules, custom, or expectation. Clearly, in such cases, we does not refer to any particular persons knowledge, moral attitude, or dress rules. This is made explicit when the speaker is asked Who knows that the speed of light cannot be exceeded? or Who doesnt torture prisoners? or Who does not wear flip-flops at weddings? and replies: I didnt mean anyone in particular. 349 350 We usually acknowledge that in any cognition above that of infants one is dependent on others: ones parents, teachers, the authors of the books and articles in magazines and newspapers one has read, the people one has talked with at home, in school, or in the street, or has heard in the classroom or on radio or television. If one has knowledge that is strictly ones own, underived from and unaffected by others, it is of the sort enjoyed by neonates and qualifies as knowledge only barely, namely, mere sensation. Educated persons explicitly or implicitly take physical things to be as physicists say they are, the past to be as historians say it was, and mathematical truths to be those vouchsafed by mathematicians. And physics, history, and mathematics themselves are inherently social. This is especially evident today. The requirement that experiments be repeatable by others as well as by oneself is sacrosanct. The 2009 status report by the Large Hadron Collider listed 2900 authors. The Wikipedia on the Internet is familiar to most of us. Even in everyday life, individual claims to knowledge count for little if they do not survive scrutiny by others. 351 The cognitive dependence of oneself on others is not only causal. It is also logical. It was an essential thesis of Wittgensteins later philosophy, as well as Sartres. Much of the history of 20th century AngloAmerican philosophy was epitomized by Wittgensteins move from the thesis in the Tractatus that the world is my world to the thesis in the Philosophical Investigations that that a private language is 6

logically impossible and that understanding a public language presupposes agreement in judgment. And many consider the highlight of 20th century continental philosophy to be Sartres argument that only through the Others look can I see myself as an object, as the entity in the world that I am that it is the Other who defines me as a person. 351 But it is to Hegel that we must go for the original account of the logical dependence of oneself on others. He insisted on the need to move from the I (das Ich) of the colourful show of the sensuous here-andnow to the I that is We and We that is I (Ich, das Wir, und Wir, das Ich ist). He called the latter Spirit, and described it as this absolute substance which is the unity of the different independent selfconsciousnesses which, in their opposition, enjoy perfect freedom and independence. This was a description of what we may call society. As Terry Pinkard puts it, Hegel reconceptualized the unity of thought and being as an intersubjective unity. And J. N. Findlay explains: Hegel holds that the understanding of other minds, far from being more obscure than the understanding of things, is the model and paradigm in terms of which intercourse with things can assume a limited clarity. In all intercourse with things we are striving towards the complete penetration and lucidity of social intercourse. But Pinkard and Findlay are downplaying the metaphysical significance of Hegels view. Human cognition may be social, but to say this of reality, of the Absolute, would be nonsense. Recognition of the primacy of society over the individual is only one step, however necessary, to grasping the Absolute. 352 Hegels central argument was that A self-consciousness exists for a self-consciousness. Only so is it in fact self-consciousness; for only in this way does the unity of itself in its otherness become explicit for it.A self-consciousness, in being an object, is just as much I as object. To exist is to be an object, and self-consciousness must earn its objecthood and therefore existence by being an object of another self-consciousness. The two self-consciousnesses recognize themselves as mutually recognizing one another. At first, of course, such recognition appears as a split between two extremes, one self-consciousness being only recognized, the other only recognizing. Each is indeed certain only of its own self, but not of the other. But its own self-certainty would have truth only if its own being-forself had confronted it as an independent object.[T]his is possible only when each is for the other what the other is for it. That self-consciousness is a social phenomenon had been suggested by Fichte a decade earlier in the Foundations of Natural Right. 353 According to Hegel, Spirit (mind, Geist) develops from the primitive stage of subjective spirit sensuous cognition to its second stage of objective spirit, by which he meant the normative customs and traditions of everyday life [Sittlichkeit], the family, the state, and institutions such as corporations and professional guilds, which would include what today we call the academic disciplines, and culminates in its third stage of absolute spirit, which includes art, religion, and the complete, perfect, knowledge Hegel called philosophy. If by mind we mean cognition, then at the stage of sensation the mind is primitive, undeveloped, and becomes developed only at the social stage. Society can be said to be a stage in the development of mind in the sense that it would not exist if the perceptions, thoughts, and judgments of its individual members did not. Indeed, academic disciplines, including philosophy, can be said to be mental in the sense that they would not exist if the perceptions, thoughts, and judgments of their individual members did not. 353 It may still seem that we are left with anthropocentrism. Wittgensteins public language, after all, is a human language, Sartres Other is a human being, and Hegels objective spirit is a society of humans, not of ants or Martians. (Hegel explicitly held even that the Absolute achieves self-knowledge in human beings, those he regarded as philosophers, his position thus appearing to be not just anthropocentric but anthropomorphic.) However, though all this is true, it is not the whole truth. Statements about a public language, Wittgensteins agreement in judgment, or Hegels social institutions are not reducible to statements about human beings, even though the language, agreement, or institutions would not exist if human beings did not. The public that speaks the public language, the parties to the agreement in judgment, and the institutions of society are not mere collections of human beings. Nor, of course, are they themselves human beings. In an important sense, they are impersonal. They are not mere collections of the particular individuals they include, just as universal statements are not mere conjunctions of the singular statements that instantiate them even though they would not be true if any of those singular statements was not. Hegel and Wittgenstein could still be charged with anthropocentrism in an extended and rarified sense, but their views also suggest how anthropocentrism can be avoided, which will be the topic of the next chapter. 354 7

Hegel famously thought that the discipline of philosophy itself was an example of the priority of society over the individual. Philosophical views are historical achievements. They are inseparable from the history of philosophy. Earlier philosophical systems, despite their diversity, are preserved in those developed later. [T]he diversity of philosophical systems is the progressive unfolding of truth, not simple disagreements. The history of philosophy is either one philosophy at different degrees of maturity; or the particular principle, which is the groundwork of each system, is but a branch of one and the same universe of thought. In philosophy the latest birth of time is the result of all the systems that preceded it. Hegel was fully aware that a philosophers thought, however original, is rooted in the thought of other philosophers, past and present, read or heard, as well as in the culture to which it belongs and the language ii employs. The content of a philosophical work is incomprehensible in abstraction from the content of previous works. The same is true, of course, of the other disciplines. Einsteins work was strikingly original but inseparable from that of his predecessors. Instruction in philosophy has always been mainly instruction in the history of philosophy, even if only the history of its latest period or in a single country. 354 Commentators sometimes say that when Hegel described philosophy as the most advanced stage of the development of spirit he meant his own philosophy. Indeed, he did hold that the Absolute achieves selfknowledge through humans. But Hegel would have insisted that no individual human beings knowledge could be identified with the Absolutes self-knowledge. Discussing Spinoza, he wrote, Such a standpoint [as Spinozas] is not to be regarded as an opinion, a subsjective, arbitrary way of thinking of an individual, as an aberration of speculation; on the contrary, speculative thinking in the course of its progress finds itself necessarily occupying that standpoint and to that extent the system is perfectly true; but it is not the highest standpoint. Perhaps Hegel thought, or hoped, that the standpoint of his syetem was the highest, though I doubt that he did. If he did, he would hardly have been unique among philosophers in this respect. 355 That advanced cognition is social is obvious in the case of cognition that necessarily involves the use of language. Some cognition is possible without language. A neonates and a dogs sense perception are examples. But distinctively human cognition, like that exemplified by the sciences and mathematics, obviously is not. Perhaps neonates and dogs have an innate language of thought that is needed even for sense perception, and perhaps God could not know the world without employing a divine language, but neither possibility will be discussed here. Our concern is with distinctively human cognition, the sort of cognition neonates and dogs clearly lack and God is believed to infinitely surpass. Wittgensteins perhaps most persuasive reason for holding that language is necessarily public, social, was his description of an imaginary private language for keeping track of ones sensations. Even if possible, such a language would be so impoverished as to be pitiful. A public language is indispensable to any cognition above that of infants. And it does presuppose agreement in judgment at least regarding which sounds or marks to count as words and which words they happen to be. 356 Until Hegel, epistemologists from Plato and Aristotle to Descartes, Hume, and Kant took for granted that cognition is individual, personal. In the Introduction I called this view cognitive individualism and contrasted it with what I called cognitive socialism. Hegel saw that it is true only on the primitive, subjective, level, and that advanced, truly objective, cognition is necessarily social. It is the nature of humanity, he wrote, to press onward to agreement with others; human nature only really exists in an achieved community with others. Elsewhere he described civil society (brgerliche Geselschaft) as a universal family. Self-sacrifice in defense of ones community, whether country or infantry squad, is not an empty ideal but a fact. It is vividly present among front-line combat soldiers: the survival of the squad, of ones buddies, is a goal overriding any other goal. Hegel became famous partly because of the moral and political implications these views were taken to have. Cognitive socialism, however, by itself has no such implications. One of its elementary tenets is that advanced cognition requires a public language. This is hardly a moral or political tenet. Cognitive socialism has been acknowledged in the sociology of knowledge and in social epistemology, though not by appealing to metaphysical considerations like Hegels. 356 Initially, as in Kants version, antirealism insists on the dependence of the world on my cognition. In Hegels version, it insists on its dependence on our cognition. In Goodmans, Dummetts, and Putnams versions, it insists primarily on the dependence of the world on our language. Heidegger assertion that it is language, not we, that speaks, was not a continental idiosyncrasy. 357

To say that the world depends on me seems absurd. To say that it depends on us may appear to multiply the absurdity. But what is at issue is dependence on cognition, and the dependence of the world on our cognition is less jarring. Its dependence on me would be abhorrent to common sense because it makes the world, the paradigm of what is objective, appear subjective. Its dependence on us would be less alarming. Not only would it be not dependence on a solitary person, it would not be dependence even on a mere collection of persons. It would be dependence on a cognition that is irreducibly institutional. There is no mystery about what such a cognition might be. An example is the cognition, knowledge, embodied in any developed discipline, say, physics. As cognition, a body of presumed knowledge, physics is not reducible to the collection of individual physicists cognitions. And while it would be glaringly absurd to hold that the physical world is what I say it is, it would not be absurd to hold that the physical world is what physics says it is. The move from I to we does not signify total abandonment of anthropocentricity. But it does signify abandonment of subjective anthropocentricity. 357 Berkeleys paradigm of cognition was perception, not conception. (He argued vigorously against abstract ideas.) Indeed, this is the immediate, natural, way to think of cognition. It explains why we also think of our cognitive faculties as personal, not social, and of knowledge as a personal, not social, achievement. This is why we find cognitive individualism more plausible than cognitive socialism and why antirealism usually has been understood as subjective. Indeed, it is more plausible in the case of perception. Cognitive socialism comes into its own when we recognize the role in cognition of conception, the understanding, and especially of description. While perception is inherently individual, personal, and subjective, conception and especially description are inherently social, public, and objective. A neonates cognition of Mother is entirely perceptual and dependent on no one elses perception of Mother. But the childs later conception of Mother as mother is totally dependent on others, obviously so when expressed in language. Even if Berkeleys idealism could be defended on the level of perception, it was Kants transcendental idealism and Hegels absolute idealism that were plausible on the level of conception. If we accept Kants and Hegels view that perception presupposes conception, then conception is primary. 358 Hegel did not mean by the Absolute some mysterious object or part of reality. He meant reality itself, which of course must be understood as relative to nothing else, as unqualified, unconditional, allencompassing. M. J. Inwood, one of Hegels most astute commentators, remarks, this expression [the absolute] is meant to convey no more than reality as a whole and even a materialist, for example, might agree that reality as a whole cannot be dependent on, or conditioned by, anything distinct from itself. The Absolute is infinite in the sense of being unlimited. But to understand this sense we must resist thoughts of infinite space and infinite time, or appeals to mathematical truths such as there is no greatest number. We must think rather of the infinite in the literal sense as what is limited by nothing (unlimited, indefinite, indeterminate), self-contained, complete, and of the finite as what is limited (definite, determinate), incomplete, a part of something else. And the image of infinity on which we must rely would not be that of the straight line, as it might be when thinking of infinite space or time or of the infinity of the series of natural numbers, but rather that of the circle. 368 Chapter Thirteen: MIND AND THE WOR 373 1. The Ghost in the Machine. 373 The antirealist thesis seems paradoxical, absurd, because it seems to say that the whole world from the page you are reading now to the remotest known galaxies, and since the Big Bang to the farthest conceivable future depends for its existence and nature on the minds, cognitive capacities, of humans, a species in one of its planets fauna. This would be a zoological thesis, and as such has no place in philosophy. Of course, it would also be absurd zoology, as well as absurd physics and astronomy. But antirealism is not a theory in zoology, nor in physics or astronomy. As we saw in the preceding two chapters, the mind relevant to the philosophical topic of the dependence or independence of the world on the mind is not a thing in the world, human or nonhuman, nor is it a part or property of such a thing. This is why the antirealist does not say that the world is dependent on the human mind. There is no human mind on which the world might depend. There is only the world. It is cognized or at least cognizable perceivable, conceivable, describable but there is no other world that is relevant, whether in philosophy or in everyday life, even though there might be a world that is neither cognized nor cognizable. Of course, there are humans in the world, just as there also are whales and chimpanzees. But there is nothing in or about humans, just as there is nothing in or about whales or chimpanzees, on which the world might depend. 373 9

Hume denied that there is a philosophical self, a metaphysical subject, in or outside the world. If he was right, it would follow that there is nothing with which the world might be contrasted, and on which it might depend. It would also follow that there is no distinction relevant to the realism/antirealism issue between me and the world. Antirealism would become what in the Tractatus Wittgenstein called pure realism. 374 Hegel denied that individual, personal cognition, if at all advanced, is independent of societal, public cognition. If he was right, and if there is no distinction relevant to the realism/antirealism issue between me and the world, there is also no distinction relevant to the realism/antirealism issue between us and the world. If by we or us is meant societal, public, cognition, and by world its object, the distinction between us and the world is rather like the distinction between feeling a headache and the headache felt. We attach little sense to talk about unfelt headaches or to talk about feelings that are feelings of nothing. If we bypass our obsession with the mind-body problem, we can say that headaches are nothing but feelings and that feelings of a headache are nothing but headaches. Yet we have little inclination to reject the distinction and to speak only of headaches. Headaches are the same as feelings of headache, but in the sense in which the Evening Star is the same as th Morning Star, not in the sense in which the Evening Star is the same as the Evening Star. Hegel explained the two senses by saying that identity is always in unity with difference, in contrast with the abstract idenity asserted in silly statements like mind is mind or planet is planet. Frege explained them by distinhguishing between the sense and the reference of an expression, holding that senses belong in a third ralm of reality, neither mental nor physical. I have offered an explanation by distinguishing between formal identity and material identity, arguing in detail that the difference between them is metaphysical, not just semantical or grammatical, as Russell held in his theory of definite descriptions. It is due to the difference between objects and entities, an object being whatever happens t be perceived or thought and an entity an object that survives the application to it of the concept of existence. 375 376 The relation between a headache and feeling that headache is not just analogous to the relation between the world and cognition. It is a familiar and reasonably uncontroversial example, instance, of that relation. Another but very different, unfamiliar but perhaps uncontroversial, instance might be the relation between an esoteric elementary particle in physics and the theory that is the only reason for accepting its reality. The instances that antirealists usually consider, such as the relation between material things or causality and the cognition of them, are seldom familiar and never uncontroversial. So are also the instances from logical cognition, especially generic statements, to which we have devoted much of this book. To be sure, few think that logical cognition has objects, but few also see that if it did not there would not be a world, just as if there were no feelings of headache there would be no headaches. But, as we have seen, logical antirealism entails at least the cosmological version of metaphysical antirealism. And what metaphysical antirealism says about the relation between the world and cognition is not fundamentally different from what common sense says about the relation between a headache and the feeling of that headache. Hegel called his system absolute idealism> He would have gladly agreed that it is pure idealism or pure realism, had he confronted Wittgensteins reasoning that led to the latter. 376 Berkeleys lasting contribution was to point out that the world is dependent on the mind insofar as the world is perceived. Kants central contribution was to point out that it is dependent on the mind also insofar as it is understood, conceptualized. Contemporary antirealism has added that it is dependent on the mind insofar as it is described. Berkeley could not consistently mean by perceiving a relation that a human animal bears to an object, and of course he did not. Kant could not consistently mean by concepts inhabitants of human skulls, and he did not: he regarded concepts as what transforms empirical intuitions into objects and thus is responsible for the objective character (essence, nature) of the things to which they apply; in a reasonably clear sense, Kantian concepts, even though imposed on things by us, can be said to be in those things. And contemporary antirealists, who appeal instead to meanings or uses of words, cannot consistently appeal to human activities such as speaking and writing. Gilbert Ryle was the first to see this when he distinguished between the usage and the use of a word, making clear that that philosophers had no business appealing to usage. 377 Ryle wrote, Hume's question was not about the word 'cause'; it was about the use of 'cause'. It was just as much about the use of 'Ursache'. For the use of 'cause' is the same as the use of 'Ursache', though 'cause' is not the same word as 'Ursache'. Hume's question was not a question about a bit of the German 10

language. The job done with the English word 'cause' is not an English job, or a continental job. Ryle went on to castigate the confusion between a 'use', i.e., a way of operating with something, and a 'usage'. A usage is a custom, practice, fashion or vogue. The method of discovering linguistic usages are the methods of philologists. In a symposium with J. N. Findlay on the topic, he remarked, The famous saying: Don't ask for the meaning; ask for the use, might have been and I hope was a piece of advice to philosophers, and not to lexicographers or translators. 377 Findlay concurred: the notion of use, as it ordinarily exists and is used, presupposes the notion of meaning (in its central and paradigmatic sense), and it cannot therefore be used to elucidate the latter, and much less to replace or to do duty for it . [We] cannot fully say, in a great many cases, how an expression is used, without saying what sort of things it is intended to refer to, or to bring to mind What Ryle meant by use of a word was not significantly different from what philosophers have meant by concept. But Findlays important addition that the use of a word involves the things to which it is applied suggests Hegels view of concepts as sunk in the things to which they apply, as permeating them, as Strawson put it. Hegel wrote, It is not we who frame the concepts [die Begriffe]Rather the concept is the genuine first; and things are what they are through the action of the concept, immanent in them, and revealing itself in them. 378 A nonzoological view of mind demands a nonzoological view of perception and conception, as well as of language. If perceiving consists, not in a minds relation to an object, but in that objects possession of certain properties, if, as we shall shortly find Strawson ssaying, concepts permeate, soak, the things conceived, and if, as we just found Findlay saying, to speak of the use of a word is to speak of the things to which it refers, then the mind itself, insofar as it perceives and employs concepts and words, cannot be just in the world, it must be the world,. This, of course, was Wittgenstein pure realism. It was also Hegels absolute idealism, which he proposed after rejecting Kants transcendental idealism as unacceptably psychological and thus subjective. 378 Berkeley began with the simplistic notions of a self (mind) confronted with an object (material or mental) and of consciousness as a relation between the two (perceiving, imagining), questioned the independent reality of the object, and arrived at idealism. Kant reduced the self to a unified consciousness. Hegel identified that consciousness with its objects. Wittgenstein rejected the self altogether, thereby removing also the motivation for a consciousness that the self aims at objects, and allowed only for the world as reality. Hegel was philosophically far apart from Berkeley, and Wittgenstein even more so. But it is unclear that on our topic Hegel and Wittgenstein were far apart from each other. It is unclear that absolute idealism and pure realism differ in more than literary and philosophical style or approach, striking though these differences are. Wittgensteins pure realism is a realism that has earned its keep, unlike the pre-Kantian realism (nave realism), which, as Descartes, Berkeley, and Hume saw, lay helpless before the skeptic. Hegels absolute idealism held that reality is thought, but it did not hold that vegetables, animals, and minerals are mental images, much less feelings or sensations. It held only that there is no cashable difference between what they are and how we perceive and think of them. There is a world and it is the only knowable world, but its nature and contents can be known only as what perception reveals and conception allows. Justice is thus done to both realism and to antirealism. 379 If perceiving and understanding are not relations between a subject and the object perceived or understood, be it external to or in the mind, then they can belong only to the category of monadic, nonrelational, properties. They can only be properties of that object, though not properties like its color or shape. Indeed, in the case of perception and its species, e.g., seeing and hearing, and imagination, everyday language does contain adjectives, e.g., visual, auditory, and imaginary, which apply to the object, not to the subject or to a relation between them. That there are such properties in the case of understanding or conceptualizing is held by most metaphysical systems: they are the properties of an object we call its essence. Little effort should be needed to convince one that understanding the French Revolution includes knowing what a revolution is. 380 2. Concepts, Properties, and Universals. 382 It is convenient here to use the terminology H.H. Price proposed in his classic work Thinking and Experience. Universalia post rem the British empiricists position are subjective states, ideas, perhaps images in the mind, possibly just words. Universalia ante rem are universals as understood by Plato, entities that exist independently of both subjective states and spatiotemporal objects. Universalia in rebus are, roughly, universals as they were understood by Aristotle, entities that are in things as their 11

properties. Berkeley and Hume denied, of course, that there are universalia ante rem. But they seemed to deny also that there are universalia in rebus, common properties, presumably on the grounds that the properties of particulars are themselves particulars. This position, known as trope theory, was developed and defended two centuries later by D. C. Williams. 387 The central question, according to trope theorists, is whether, e.g., the color of this page is identical with the color of the next page, or only resembles it, whether exactly or inexactly. Thus the appeal to tropes is an attempt to avoid commitment to universals by appealing to the presence instead of a relation of resemblance. But there is no fact of the matter here. If there were, we should be able to tell, in the case of the two pages, which is the right view by just looking at them after all, we see them and we also see their color. We would expect to see whether the relation between the color of this page and the color of that page is identity, i.e., that a certain color is a property common to both, or resemblance, i.e., that they are particular properties related by resemblance. But how would these alleged relations of identity and resemblance differ? The truth is that neither the philosophers who say that properties are universals nor the philosophers who say that there they are particulars, tropes, are reporting anything to be settled by experience or inference from experience, that neither are really reporting a discovery of some facts about the world. They can claim only to be offering a more illuminating description of a familiar and uncontroversial fact about the properties of individual things. The problem of universals is whether this fact is more like a paradigmatic informative identity, say, that of the Morning Star and the Evening Star, or like an ordinary relation, say, the spatial relation between the two pages. Both science and common sense hold that it is the former. Both would choose the conjunction of the subject-predicate statements a is F and b is F to describe the color of the two pages, rather than a relational statements of the form aRb. In fact, there are no relevant statements of the latter sort because there are no names of particular properties, of tropes. In aRb, where R stands for resemblance, a and b could not be names, respectively, of the color of this page and the color of the next page since there are no such names. There are the definite descripions the color of this page and the color of the next page, but The color of this page resembles the color of the next page is not a statement of the form aRb. 388 Universalia in rebus are what in current philosophy are just called properties. They are therefore taken to be present in everything that has properties. Hegel rejected Platos universalia ante rem, and had no patience with the empiricists universalia post rem. At first glance, therefore, he seemed to accept Aristotles universalia in rebus. Indeed, we find him writing, Animal, qua animal, does not exist: it is merely the universal nature of the individual animalsBut to be an animal is the property of the particular animal, and constitutes its definite essence. Take away from the dog its animality, and it becomes impossible to say what it is. 389 This remark about animal is best understood in its original context of Aristotles Categories. It concerns substance universals, what Aristotle called secondary substances. Such universals are the genera and species of what Aristotle called primary substances, that is, individual things, which are neither present in nor said of (i.e., predicable of) anything. For centuries, animality (the Animal) served as a standard example of such a universal, but so would have been caninity (the Dog). The latter is a species of the former, which is its genus. In Aristotelian terminology, both are said of, but are never present in, individual things. Hegels just quoted remark applies also to property-universals, what Aristotle called accidents, say, round shape and brown color. Their instances, strictly speaking, are not the individual things in which they are present and of which they are said (predicated), but rather certain properties of those things, say, the shape of this ball and the color of that dog, which are present in but not said of those things. A brown dog is brown, but it is not its brown color. 389 We may perceive a dog as well as its brown color, and thus can be said to perceive also the universals of which they are instances, namely, caninity and the color brown. If we see the dogs brown color, we may be aware also of its genus, the generic property color, though not by seeing it as a separate entity, in addition to seeing the dogs brown color. But we can think of the generic property, whether or not we perceive an instance of it. We can think of it even without employing a suitable mental image, as we do when trying to imagine it. We may be said, then, to have the relevant concept, but this need mean no more than that we are capable of thinking of the universal and recognizing it in its instances. We do not have the concept therefore in the way we might have a mental image we would have the concept, but not the image, even when asleep. This is a sufficient reason for denying that concepts are mental images. Another reason is that we need concepts even to produce and then recognize the images we want. 390

12

Cognition, according to Hegel, necessarily involves concepts (Begriffe), understood as comprehensions, cognitions, of the things in which they are sunk. Reality necessarily involves universals, but in Hegels sense of concrete universals sunk in things, not mere abstractions. Hegel explained the plausibility of Platos theory of universalia ante rem as follows: When the universal is made a mere form and coordinated with particular, it sinks into a particular itself and [T]he opposition [of the universal] to the particular is so rigorously maintained, that it is at the same time also reduced to a particular again. Universals, for Hegel, are not denizens of a Platonic realm. They are not separate from the world. They are not entities that individual things merely imitate, as Plato supposedly believed. They are in things, but not as parts of things. They are not spatiotemporal parts of the world but they are properties of its spatiotemporal parts. They are concrete, not abstract, universals. 390 When Hegel says that universals particularize themselves, that they self-particularize, this need mean no more than that there is nothing in individual things that might connect them to their properties, that there are no Lockean I know not what substrata or Bergmannian bare particulars. When Hegel says that the Absolute develops from the Logical Idea to Nature, this need mean no more than that concepts have no reality apart from individual things. When he says that the Absolute develops from Nature to Spirit, this need mean no more than that individual things have no reality apart from concepts. 391 3. Conclusion. 391 Hegels account of concepts was echoed, in an Anglicized but clearer way, by H.H. Price and P. F. Strawson, despite their great differences with him in other respects. Prices account of concepts and universals in Thinking and Experience was not unlike Hegels. Cognition as understanding, thinking, necessarily involves actual or potential classification and thus concepts, which are best understood, Price suggested, as principles of classification. This is why sometimes we use the clumsy word conceptualization instead of understanding. Insofar as the principles of classification are objective, they are indistinguishable from the classes of the things classified. Concepts thus are not words or mental images, but rather recognitional capacities. The account of classification Price preferred was the philosophy of universals, essentially Aristotles view of universals as properties of things, universalia in rebus. The properties of a thing are the natural ground of its objective classification. We may go a small distance beyond Price and say that as principles of classification concepts are not clearly distinguishable from the classes of individual things that they yield, and therefore from their corresponding common properties. For the distinction between class and property is murky. It may be worth recalling that Goodman and Quine argued at one time for nominalism regarding both properties and classes, on the grounds that both are abstract entities: We do not believe in abstract entities. No one supposes that abstract entities -- classes, relations, properties, etc. -exist in space-time; but we mean more than this. We renounce them altogether. Later, at least Quine significantly modified this rather extreme stand. 391 Price was not an antirealist, but he might have agreed that we cannot actually distinguish between what the world really is and how we understand it. It is commonplace to speak of the child's world, the soldier's world, the scientist's world, and so forth. Such descriptions do not refer to different portions of the world. They refer to the world itself, but as understood or conceptualized in a certain particular way: the childs, the adults, or the scientists way. The phrase conceptual scheme is often applied to such understanding. It suggests the possibility of there being many conceptual schemes incompatible with each other. This possibility was explored by Strawson. 392 The metaphysical picture of the world we have drawn shares with Hegels the absence of the usual subject-object dualism. Is it then idealist? If it is, the idealism certainly is not Berkeleys. Is it Hegels idealism? It depicts no Absolute that develops from Logical Idea to Nature and from Nature to Spirit, nor does it employ any dialectical method. Is our metaphysical picture then Sartres? They have in common the explicit rejection not only of subject-object dualism but also of the act-object dualism that considers consciousness as an entity. But our metaphysical picture bypasses Sartres existentialist concerns, e.g., with freedom, bad faith, sadism and masochism, which surely are irremediably anthropocentric. Is our metaphysical picture Wittgensteins, in which subject-object and act-object dualisms are also absent? If Wittgenstein had a metaphysical picture at all, it was so sketchy that the question is unanswerable. Had he had one, what our picture and his would share is the rejection of the philosophical self and thus of subjectivism in all its forms. In this respect both would be descendants of Hegels metaphysical picture. Wittgensteins assertion that The self of solipsism shrinks to a point 13

without extension, and there remains the reality co-ordinated with it and Hegels assertion that the thoughts, far from being merely ours, must at the same time be the real essence of the things need not convey different metaphysical pictures, despite the great differences in terminology, style, and historical context. 398 Our picture of the world is compatible with other philosophical pictures, as long as they avoid divorcing the world from cognition and including a subject in the world (Wittgenstein). If consciousness has no inhabitants, not even an ego, then all things are outside consciousness, and there is just the world. Hence the plausibility of the physicalist picture of the world as matter. But, unlike it, ours does not exclude consciousness it merely does not include it, much as a group portrait of a family might not include the photographer. 398 Our metaphysical picture is also compatible with that of a phenomenology that remains faithful to Husserls motto We must go back to the things themselves. Consciousness is not in our picture because it is not among the things themselves, an additional item in the world competing with them for a place in reality. It has no inhabitants or qualitative nature. There are only its objects. In itself it is nothing. Nonetheless, consciousness is the revelation of things and the regulator of which are admissible into the world. It is a revelation on the level of perception (sensibility, outer or inner; intuition, sensory of intellectual). It is a regulator on the level of conception (understanding, thought, reason, language). Perception opens the entrance to the world, but conception is the gate-keeper. So understood, consciousness is what we have called cognition, sometimes even knowledge, the outcome of the successful employment of cognition. We can responsibly regard as objects in the world only those revealed and then allowed by consciousness. To this extent our metaphysical picture is compatible also with that of pure idealism: the world as mind. It does justice to what makes idealism plausible. But though our picture contains only what consciousness reveals and accepts, it does not exclude Kantian things-in themselves. It merely does not include them, much as the group portrait of a family might not include the grandparents. 399 In The Vocation of Man, Fichte offered a characteristically dramatic, but also unhappy, view of consciousness: All that I know is my consciousness itself..I know of no being, not even of my own. There is no being.Pictures [Bilde] are: - they are the only things which exist Fichte thought the moral was that we must rely on faith, rather than knowledge. The views of consciousness we find in Hegel, Wittgenstein, and Sartre, however, suggest that we should rather say: All that I know is the world itself. The things in it are: - they are the only things which exist. But we might be wiser if we refused to choose between the metaphysical picture Wittgensteins pure realism conveys and that of Fichtes pure idealism, and welcome both. Our philosophical understanding would then be richer. Of course, the welcome need not be unreserved. Pure realism does not entail physicalism, and therefore it is free of the latters rashness and implausibility. Pure idealism does not entail doctrines about an absolute or a dialectical development, and it need not be expressed in the jargon of German Idealism or of Husserlian phenomenology. As to which is preferable, suffice it to note the advantage of the close proximity of pure realism to both common sense and science. Pure idealism and Husserlian phenomenology lack that advantage. 399 A metaphysical picture, of course, is not a metaphysical system. Our project has been to free philosophy first epistemology and ethics, then metaphysics of anthropocentrism, not to construct ethical, epistemological, and metaphysical systems. But even if we distrust the claims to philosophical knowledge that such systems make, we have left room for them. Our metaphysical picture allows for the investigation of numerous ordinary, traditional, philosophical topics, to which, as I pointed out in the Introduction, the realism/antirealism debate is irrelevant because they are affected neither by the presence nor by the absence of anthropocentrism. The dehumanization of philosophy proposed in this book is compatible with any standard theory about such a topic. 400 For example, much of the history of metaphysics has consisted of arguments for the existence of Platonic Forms and of God. The arguments are not as bad as many take them to be. The usual objections to Platos Forms are cases of ignoratio elenchi: that the Forms play no role as efficient causes and thus are irrelevant to our knowledge of the world, or that we are not acquainted with universals that are not exemplified. If the Forms were proposed by Plato as causes, obviously they would have been formal, not efficient causes, in Aristotles terminology. And if by acquaintance is meant sense-perception, the Forms were explicitly proposed as examples of entities that cannot be perceived through the senses. Much of the history of metaphysics has also consisted of arguments for the existence of God. The 14

objections to these arguments have been less than compelling. The standard complaint about the ontological argument rests on an assumption less plausible than any premise of the argument, namely, that the concept of existence does not stand for a property (perfection) but is properly expressed by the particular (existential) quantifier. The cosmological argument from the need for a first cause has unimpeachable motivation that even scientists share. The cosmological argument from design, however, is on shakier grounds because of the anthropocentrism latent in its use of the concept of design. 401 Our picture leaves room for much else in the history of metaphysics that is less unfashionable today than Platonic Forms and the existence of God. A prime example is the project of ontology: the listing and description of the most general kinds (categories) of entities and the relations among those kinds. Other examples are the questions whether the properties of individual things are universals or particulars (tropes), and whether individual things are merely bundles of their properties. 401 There is also the novel metaphysical topic that emerged from our rejection not only of the metaphysical self but also of consciousness as ordinarily understood. I suggested that the familiar distinctions among kinds of consciousness, such as vision, audition, memory, imagination, and conception, should be understood as distinctions among monadic properties of the putative objects of the consciousness, perhaps already expressed by the adjectives visual, auditory, and imaginary. Related to this topic is the question of how the thesis of antirealism should be understood in the case of cognition that is not advanced, e.g., that of neonates and adults who never acquired a language. 402 There is also the question whether what the metaphysical pictures drawn earlier, realist and idealist, do not include nonetheless ought to be in some fashion included: consciousness in the case of the realist picture, Kantian things-in-themselves in the case of the idealist picture. No science can answer such a question. If it can be answered at all, only metaphysics can do it. If it cannot be answered, only metaphysics can explain why it cannot. 402

Notes 401

Chapter One: INTRODUCTION

1. Anthropocentrism. I am unable to believe that, in the world as known, there is anything that I can value outside human beings, and, to a much lesser extent, animals, wrote Bertrand Russell.1 Many think that such anthropocentrism mars our relationship to other animals and the environment generally, just as egocentrism mars our relationship to other humans; speciesism, they would say, is no more acceptable than is egoism, androcentrism, or
15

ethnocentrism. Many think also that anthropocentrism mars our conception of God. They are inclined to agree with Spinoza that, contrary to standard religious doctrine, neither intellect nor will pertain to the nature of God,2 and that God is free from passions, nor is He affected with any emotion of joy or sorrow.3 To attribute to God human characteristics such as intellect, will, joy, or sorrow, they would say, is to think of God as a sort of superhuman.

These cases of anthropocentrism are well-known, and have been amply discussed for centuries. They are not the topic of this book. Its topic lies deeper: the anthropocentrism present in our most fundamental thinking about knowledge (in epistemology), goodnessw (in ethics), and the world itself (in metaphysics). Its importance is obvious, but it has seldom been discussed.

The first entry in the philosophy curriculum of a distinguished American university bears the title Philosophy and Human Nature. The curriculum, like those of other universities and colleges, includes also courses on human knowledge and sense perception, happiness and pleasure, even the world itself as a human construct. Those who teach these courses probably do not explain how they acquired expertise in such human matters. Concern with human beings, of course, is natural and morally expected of all. However, it is a professional concern only for some neuroscientists, psychologists, psychiatrists, economists, sociologists, anthropologists, demographers, linguists, lexicographers, physicians. Aristotle did engage in biological investigations, but biology was hardly a developed science at the time. Today, it is. For most of the history of philosophy and science, if a topic did not belong in mathematics, astronomy, or theology, it was
16

dispatched to philosophy there seemed to be no other place to put it. Even today, in some institutions, psychology is called mental philosophy, and physics natural philosophy. But neither is considered part of philosophy.

Philosophers willingness to assume authoritative stands on matters concerning human beings is not due to hubris or academic gerrymandering. Its roots are deeper. Anthropocentrism has been present in philosophy throughout its history, but it became especially incongruous after the emergence of experimental sciences devoted expressly to the study of humans. It has often swayed the very conception of philosophy and driven its choice of topics. And because philosophy influences indirectly and silently, but lastingly all other advanced thought, the harm anthropocentrism has done in philosophy extends further. This book is concerned with the ways it has affected the three central branches of philosophy metaphysics, epistemology, and ethics. I shall argue that it has no place in any of them. I shall also argue, however, that this does not mean that they should be abandoned which would amount to abandoning philosophy but that they should be radically rethought.

My argument will rest not on abstract philosophical premises or assumptions but on specific and readily understandable truths. Whatever the nature of the world may be, humans are only inhabitants of it. Knowledge of them, like knowledge of any other inhabitants of the world, is most credibly sought by empirical means. But philosophy is not an empirical discipline. Philosophers perform no experiments, have no labs, use neither telescopes nor microscopes, embark on no field trips nor do they as a rule have the inclination, training, or talent needed for doing any of these. One of Hegels
17

criticisms of Kant was that Kant has simply no ground but experience and empirical psychology for holding that the human cognitive faculty essentially consists in the way it appears.4 But what other kind of ground can there be for holding anything about a human faculty? The moral to be drawn, however, is not that philosophers are experts in nonempirical, a priori, things or facts. If numbers are such things, it is mathematicians, not philosophers, who are the experts about them.

Some may say that not all of human nature is empirical, that humans also have immortal souls. But this is a matter of faith, not investigation, empirical or nor, and even if true it is seldom relevant to the concerns of philosophers. Others may say that even if humans have no immortal souls they have immaterial minds, which are entirely distinct from both their brains and their behavior. But there has been an empirical science investigating such minds: the introspective psychology of James, Wundt, Titchener, and many others. To be sure, it was largely unsuccessful, though not because its subject matter called for nonempirical investigation; the introspective psychologists explicitly relied on experience. Still others may follow Kant, Heidegger, Sartre, or Nelson Goodman and say that the empirical world itself is in some sense human, made by us, as Goodman put it. This is essentially the thesis of antirealism, in the broad sense of the term that applies to Berkeleys, Kants, Hegels, and 20th century views such as Hilary Putnams, Michael Dummetts, and Nelson Goodmans. Much of this book will be devoted to that thesis. If taken literally, however, surely it is absurd.

More likely today is to be told that philosophical inquiries are not really about human beings, that they really are conceptual or linguistic. They are about concepts or words,
18

not about the things or facts, human or nonhuman, those concepts or words stand for. For example, it would be said, in ethics philosophers investigate the concept of happiness or the meaning or use of the word happiness, not any facts about happiness, which indeed should be left to psychiatrists and pharmacologists to investigate, and in epistemology philosophers investigate the concept of perception or the meaning or use of perceive, not the facts about perception, which should be left to psychologists and neuroscientists to investigate. But surely the concepts and words in question are human, not platonic or divine, and constitute an empirical subject matter themselves. The investigation of them calls for empirical methods those of psychology, linguistics, and lexicography not for philosophical speculations, intuitions, or a priori arguments.

The traditional claim of philosophy to a distinct place among the cognitive disciplines has rested on its absolute fundamentality, supreme abstraction, and unlimited scope. In these respects it surpasses even mathematics: one of its topics are the nature and status of mathematics itself. Its scope includes that of physics and astronomy space, time, and everything in them but philosophy is also concerned with anything that is not or might not be in space and time. Philosophy presupposes nothing and conceals nothing. This is why philosophers court paradox when preoccupied with things as concrete, literally down to earth, as the humans in a planets fauna. The paradox is no less glaring than it would be if they were preoccupied with the cetaceans in that fauna. If philosophers do not see the paradox, the reason presumably is that they are human. Had they been cetacean, they might have been preoccupied with cetaceans.

19

The concern in philosophy with humans is not a consequence of its unlimited scope, its interest in all time and existence. It is not the trivial application to humans of general philosophical propositions, like the application to humans of arithmetic by the Census Bureau or of physics by a pilot worried about takeoff weight. It is a substantive concern. Anthropocentrism in philosophy may be woefully misguided, but it is natural. The reason for it is obvious. Plumbers or philosophers, we all are humans. We are deeply interested in ourselves and other humans. We see ourselves as the center of the universe even when we know that probably we are at its periphery. To suggest that philosophy should not be about humans, that it ought to be dehumanized, may seem even offensive.5

A common thesis in moral and political philosophy is that one ought to treat others humanely, as ends, not just as means, perhaps even love them. The claims of kings and barons to special dignity were rejected in the past by declaring the dignity of all men. Inhuman and inhumanity are standard terms of condemnation, humane and humanitarian of approbation. Politicians handlers try to humanize their clients in order to get them elected. The human condition is a perennial object of marvel or despair. Works of art are often praised for their human quality. A favorable review of a recent novel emphasizes its being deeply human. Many demand that space exploration be funded only if it might lead to cures of human diseases. A recent television commercial announces, The human element, nothing is more fundamental, nothing more elemental, and advocates adding it to the periodic table. A distinguished contemporary philosopher writes of the heart-breaking specialness of the human.6

20

Indeed, we all think humans are special and even feel their specialness. But the reason is not that we think we are angelic and thus obviously special in a way that, say, cetaceans or extraterrestrials would not be, but that we are aware of the unsurprising fact that we all are human. We all desire and seek pleasure, happiness, well-being for ourselves, those we love, often for strangers. So, ethics has devoted itself to the investigation of the human good well-being, flourishing, happiness, pleasure and the habits, actions, and institutions conducive to it, rather than to the good of cetaceans, extraterrestrials, or angels. We all treasure our ability to see and hear, and to remember and think. So, epistemology has devoted itself to the investigation of the nature and sources of human knowledge, not the nature and sources of cetacean, extraterrestrial, or angelic knowledge. But human happiness and human knowledge are obviously empirical matters, belonging today in the subject matter of developed sciences.

This, of course, has been pointed out repeatedly and eloquently by the proponents of naturalism, the dominant orientation in current philosophy. W.V. Quine was perhaps its most prominent defender and accordingly worked mainly in logic. Unlike Quine, however, most naturalists have lacked the courage of their conviction and continued to engage in analyzing concepts or describing the workings of our language, as if concepts and language were not themselves parts of nature and thus in the province of empirical science.

Russell was unable to value anything outside human beings, and, to a much lesser extent, animals, but elsewhere he did write that man does not have the cosmic significance assigned to him in traditional theology.7 We see ourselves as the center of
21

the universe, even though most of us know that we are not. We have no cosmic importance. Our vanity, however, is cosmic. We crave center stage. The existence of Homo sapiens perhaps began about half a million years ago and, because of global warming or global cooling, cosmic collision or catastrophic earthquake, worldwide nuclear war or devastating global epidemic, or some other disaster, it is unlikely to continue for more than another half a million years. It is no more than a flicker in the four or five billion years of the history of the Earth, and the Earth itself is no more than a speck in the vast expanse of the universe. All this has been a familiar cause of despondency and pessimism, a standard theme in gloom-and-doom preaching, and a ground for skepticism about the ultimate meaning of human life.

To object by saying that this presupposes the naturalist and ignores the religious standpoint would be a misunderstanding. The issue here is not the merits of naturalism or of religion. Whichever standpoint we adopt, the idea that philosophy ought to be about us has no more merit than the idea that astronomy or mathematics ought to be about us.

This is why philosophy ought to be freed from its anthropocentrism, dehumanized. Dehumanized is an ugly word and the phrase dehumanizing philosophy may sound alarming, but they do capture literally and briefly the needed drastic change in philosophical focus. Of course, humans do deserve investigation: misanthropy is not advocated here. My aim is not to stoke despondency or pessimism. Nor is it to advocate rigid naturalism, let alone scientism. It is to draw attention to facts that ought to be uncontroversial, as uncontroversial as the fact that humans are not part of the subject matter of mathematics or astronomy. To say that philosophy should not be about us is no
22

more an expression of a religious or antireligious standpoint, or of misanthropy, pessimism, naturalism, or scientism, than saying that humans are not part of the subject matter of mathematics or astronomy would be such an expression.

Our cosmic vanity is fueled by various assumptions, some true and some false. We may not think that we are angels but we do assume that with the possible exception of gods, angels, or extraterrestrials humans alone are rational, capable of reasoning. We assume also that only humans are capable of moral and aesthetic judgment, and perhaps that only humans enjoy the moral and political status of possessing rights. The monotheistic religions, which were the home of medieval philosophy and profoundly influenced early modern philosophy, assure us of humans unique origin and special place in nature. Some of these assumptions and assurances are matters of faith, e.g., that, though all things were created by God, only man was created in Gods image. Some are essentially scientific but now abandoned, e.g., geocentrism, the Ptolemaic view that the Earth is the center of the universe. Other assumptions are also essentially scientific but of unknown and perhaps unknowable truth-value, e.g., that there is no intelligent life elsewhere in the Universe. And many do rest on scientific fact, e.g., that no other terrestrial animals equal humans in intelligence. All these assumptions not only encourage our special interest in ourselves but seem to justify it.

Anthropocentrism is especially evident and easiest to understand in epistemology and ethics. But it is present also in metaphysics, in the form of antirealism, the metaphysical orientation that distinguishes post-Berkeleyan and especially post-Kantian philosophy most sharply from ancient, medieval, and early modern philosophy. According to
23

metaphysical realism, reality, the world, is mind-independent, in particular, it is independent of our knowledge of it. Antirealism stands for the opposite view, exemplified by Berkeleys subjective idealism, Kants transcendental idealism, Hegels absolute idealism, and recent positions such as Michael Dummetts antirealism, Nelson Goodmans irrealism, and Hilary Putnams internal realism. According to the metaphysical antirealist, reality depends, at least insofar as it is known or knowable, on our ways of knowing it, on our cognitive capacities sense perception, introspection, intellectual intuition, imagination, memory, recognition, conceptualization, inductive and deductive reasoning, use of language and other symbolisms. Cognition is the employment of these capacities. It leads to knowledge when successful, and to error when unsuccessful. So understood, antirealism allows for the possibility of an unknowable reality (Kants things-in-themselves), which is independent of our cognitive capacities, even if, as Goodman claimed, it might not be worth fighting for or against.8

The antirealist claims that the world is shaped or sculpted by our cognitive faculties, and thus that it depends on us, who of course are humans. This, essentially Kantian, claim amounts to the humanization of metaphysics. It rests on the tautology that we have no cognitive access to the world except through our cognition of it, that whatever we know, perceive, understand, believe, imagine, or say about the world depends on our cognitive capacities. The world is cognized also by nonhuman animals, as well as, perhaps, by extraterrestrials and angels, but in order to know or even understand and say this we must rely on our own cognitive capacities, even if only our imagination and our language.
24

Anthropocentrism is paradoxical in all branches of philosophy, though for different reasons. Epistemology and ethics claim expertise about what they must and usually do regard as certain animals. But animals are part of the subject matter of the empirical sciences, not philosophy. Metaphysical antirealism holds, however tacitly, that reality, the whole world at least insofar as known or knowable is dependent on the cognitive faculties of those animals. Common sense the mature and thoughtful judgment we all share and all theorizing, scientific or philosophical, begins with and must respect even if not accept finds such cosmological humanism, indeed human creationism, absurd.9 How could the whole world depend on just some members of one of its planets fauna? In epistemology and ethics, philosophers seem to play the role of zoologists. In metaphysics, they seem to announce that the whole world is zoological.

Although in epistemology and ethics anthropocentrism is natural and understandable, though indefensible, in metaphysics it is unnatural and almost incomprehensible, but at least as defensible as Kants transcendental idealism and 20th century antirealism. In epistemology and ethics, we can reject anthropocentrism unqualifiedly, however painful this might be. In metaphysics, however, such rejection would be a blunder. The reason is that it is present there in the form of antirealism, which is made plausible even if not not entailed by a tautology and therefore indisputable trith. The tautology is that whatever we know, perceive, understand, believe, imagine, or say about the world depends on our cognitive capacities. So, in metaphysics we face the challenge of finding a way to avoid anthropocentrism without rejecting this tautology. We must not just return to preKantian metaphysics, yet we must avoid the absurdity of human creationism. We must
25

interpret antirealism as making no reference to humans. Hence, the unusual dialectical structure of this book. Part One is devoted to defense of the dehumanization of epistemology and ethics, Part Two to explanation and provisional defense of the antirealist humanization of metaphysics, and Part Three to the dehumanization of antirealist metaphysics.

2. A Glance at History. The history of anthropocentrism in philosophy is illuminating and helps us understand it, though here we can devote to it no more than a passing glance. It was preceded and encouraged by the anthropocentrism already present in religion. According to the Book of Genesis, in the beginning God created the heaven and the earth, including its flora and fauna, and then created man in the image of God. But it also attributed to God actions such as resting from work, speaking, and inflicting punishment. Thus it appeared also to depict God in the image of man. Its anthropocentrism with respect to Creation seemed accompanied by anthropomorphism with respect to God, the personification of God that became essential to Judaism, Christianity, and Islam, but was deplored by Spinoza.

Pre-Socratic philosophy focused on the cosmos the heavens and the earth, the four elements, ensouled animals. The Socratic revolution shifted the focus dramatically. It moved philosophy from concern with all things to concern with just one. Its slogan became Know thyself! But it did not go unchallenged. The stand most distinctive of Platos philosophy was both anti-preSocratic and anti-Socratic. The chief concern of philosophy, Plato taught, is neither the cosmos nor oneself, but rather what lies beyond both: the abstract entities he called Forms. Almost immediately, however, Aristotle
26

urged a return to the cosmos, away from both Socrates preoccupation with the innermost and Platos preoccupation with the outermost.

Medieval philosophers Christian, Jewish, Muslim sometimes followed Plato, sometimes Aristotle, but all took for granted the declaration in Genesis that humans were made in the image of God. Their anthropocentrism, however, was subordinate to theocentrism. And the strictures on anthropocentrism in philosophy do not apply to theocentrism in philosophy. For instance, Thomas Aquinas theocentrism, according to which God was the source and ground of all being, exemplified unambiguously the abstraction and generality distinctive of philosophy. And his credentials as a theologian were unquestionable.

Following the scientific revolution in the 16th and 17th century, when Copernicus, Galileo, Boyle, and Newton became the authorities on space, time, and matter, philosophers came to hold, in sharp contrast to both their ancient and their medieval predecessors, that they were concerned only with minds and ideas. The new physics compelled them to adopt the way of ideas. Anthropocentrism thus became firmly established in philosophy, since it was not cetacean, extraterrestrial, or angelic minds and ideas that attracted the attention of the early modern philosophers. But their anthropocentrism consisted in preoccupation with just one part or aspect of human beings: their minds and the ideas in those minds. It eventually led many to idealism, the view that minds and ideas were all that there is. To be consistent, however, idealists ought to have denied that those minds or ideas were human. For humans have material bodies, even if also immaterial minds,

27

and to allow that there are human minds would be to allow that there are also human bodies. Few appreciated the severity of the problem this fact posed for them.

Early in the 17th century, Descartes began his Meditations by arguing that I exist was the only truth he could not doubt, presumably referring, though inconsistently, by I to himself, a certain Frenchman. But after his propaedeutic dalliance with epistemology, and thus with human matters like the possibility and extent of human knowledge, Descartes focused on the existence of God and the nature of mind and matter. This focus was metaphysical and nonanthropocentric. It was shared later in the century by Spinoza and Leibniz. But it was vigorously opposed by the British empiricists, who returned to Descartes epistemology and thus to anthropocentrism. The titles of their chief works speak volumes: An Essay Concerning the Human Understanding (Locke), The Principles of Human Knowledge (Berkeley), A Treatise of Human Nature (Hume).

In 1781, less than half-a-century after the publication of Humes Treatise, Kant rejected in his Critique of Pure Reason both continental metaphysics, which he called dogmatism, and British epistemology, which he called empirical idealism. He described his own philosophy as transcendental idealism, arguing, for example, that space and time are not actual entities because they belong only to the subjective constitution of the mind as forms of appearances.10 But Kants version of antirealism allowed that there are things-in-themselves, entities independent of our cognitive faculties.

Transcendental idealism is often described as a manifestation of the humanism characteristic of the Enlightenment. But this is misleading, and so is the vague and
28

overused term humanism. Like any antirealist position, transcendental idealism does rest on the tautology that our knowledge of the world depends on our cognitive faculties. On the basis of that tautology, Kant went on to assert also that the world itself as known or knowable by us depends on our cognitive faculties. Whether there is a world that is not known or even knowable by us, or whether there are at least parts or aspects of the known or knowable world that are not known or knowable by us, is a further question. The weaker (at least) formulation of the question is needed because it accords with how sometimes Kant formulated his distinction between things-in-themselves and things-forus.11 It also makes the distinction less unusual and Kants antirealism more plausible by allowing that, even if there is only one world or set of things, namely the world or things we know or can know, this world or these things may have parts or aspects that are not known or knowable by us.

When Kant wrote that space and time belong only to the subjective constitution of the mind, he was referring to humans explicitly. We can speak of space, extended things, and so on, only from the human standpoint, he wrote, and explained that this is why space is transcendentally ideal, though it is also empirically real in the sense that it is not an illusion, or mere fancy.12 But when he wrote that space and time belong only to the subjective constitution of the mind, he couldn't have meant Kant's mind or any other human's mind, and we know he did not mean Gods mind. What did he mean then? Humans themselves, including Kant, are inhabitants of the transcendentally ideal space and thus are transcendentally ideal. To be sure, they also are empirically real: although Kant held that our knowledge is limited to appearances, he made a sharp distinction between appearance (Erscheinung) and illusion (Schein).13 Transcendental idealism was
29

thus anthropocentric insofar as humans are empirically real, but it was not anthropocentric insofar as humans are transcendentally ideal. Nonetheless, it usually has been understood as unqualifiedly anthropocentric. So have its immediate successors: Fichtes absolute idealism, who titled his best known book The Vocation of Man, and Hegels absolute idealism, according to which the Absolute achieves self-knowledge through the human mind. We shall find that a nonanthropocentric reading of all three Kant, Fichte, and Hegel is not only possible but plausible. To explain this reading would be the special challenge of dehumanizing antirealist metaphysics, which will be attempted in Part Three.

Standard ethics and epistemology regard humans, and their happiness, sense perception, etc., only as empirically real. Kant called such ethics practical anthropology, which he described as an empirical discipline, and reserved the term metaphysics of morals for what he considered the properly philosophical, a priori, inquiry into morality. The philosophical inquiry into our cognitive faculties was exemplified by his transcendental aesthetic and analytic. He did not have a special name for their empirical study, but presumably would have agreed that it is the task of what we now call cognitive psychology and neuroscience.

Nevertheless, Kant certainly thought, as everyone does, that all disciplines, the physics of space and time as well as the metaphysics of morals, are anthropocentric, indeed literally human, in the straightforward sense that they employ human cognitive powers and thus are constrained by the demands and limitations of those powers. Whether we contemplate all time and existence or just snails and tomatoes, we cannot transcend our
30

sense organs and brains, just as we cannot get out of our skins. Understood as the tautology that our knowledge of the world depends on our cognitive faculties, this truth is not grounded in facts about humans beings, just as the truth of All bachelors are unmarried is not grounded in facts about bachelors. Both enjoy the certainty of logic.

Antirealism became dominant after Kant. Early in the 19th century, Hegel declared that Spirit (mind, Geist) develops or unfolds logically (dialectically) from subjective spirit (individual human mental states like sensations) to objective spirit (society as exemplified by the family, customs and traditions [Sittlichkeit], the state, and institutions such as corporations and guilds, among which would be todays academic disciplines), and reaches its fulfillment in absolute spirit, the three stages of which are art, religion, and philosophy (in Hegels sense of philosophy as the perfect system of knowledge). Thus, seen superficially, Hegel appeared to endorse not only metaphysical anthropocentrism but also a sort of metaphysical anthropomorphism. But his anthropocentrism involved a major epistemological innovation: a dramatic move from earlier philosophers cognitive individualism, the view of knowledge as a personal achievement, to what, for the lack of a better term, I shall call cognitive socialism: the view of knowledge as a social, disciplinary, often literally collaborative achievement. It is especially evident today that the cognitive disciplines are inherently social and, at least to users of Wikipedia, that so is virtually all cognition beyond the infantile stage. Cognitive socialism, of course, need have no political or economic implications.

In the 20th century, Wittgenstein declared in Tractatus Logico-Philosophicus, I am my world, seemingly referring, though inconsistently, to a human being, namely himself.14
31

Heidegger focused his Being and Time on human being, the inquirer, using Dasein as a technical term for it,15 though two decades later he insisted that [e]very kind of anthropology and all subjectivity should be left behind.16 Sartre referred in Being and Nothingness to Being-for-itself as realit humaine, which he contrasted with Being-initself, and wrote the article Existentialism is a Humanism, to which Heidegger responded in a paper titled Letter on Humanism.17

While the scientific revolution of the 16th and 177h century motivated philosophers to confine their subject matter to minds and ideas by taking the way of ideas, in the 20th century minds and ideas themselves became the subject matter of flourishing empirical sciences psychology and later neuroscience and philosophers retreated further. They took the way of words, the linguistic turn, claiming now that their subject matter is language. Sometimes, they described their inquiries as conceptual, not factual, but by concepts they did not mean the ideas or mental images of early modern philosophy. They meant meanings or uses of words and syntactic structures, the workings of our language.

But language and words are matters no less empirical than space, time, and matter, or human minds and ideas, and they are investigated today by linguists, lexicographers, and even computer scientists. Research in them requires meticulous empirical description and fruitful, empirically verifiable hypotheses, not definitions or if and only if statements as in philosophical writings. Philosophers have no more special insight into the workings of language than they had had into the workings of matter in space and time or into the workings of the human mind. Instead of investigating space, time, and matter,
32

the 17th and 18th century philosophers settled for investigating mind. Instead of investigating mind, the 20th century philosophers settled for investigating talk. They remained open to the charge of rummaging in empirical matters. To be sure, philosophers have a good ear for the nuances of some segments of speech, as J. L. Austin famously did, but Austin insisted that such an ear was not a substitute for empirical research.18

When combined with antirealism, as it often was, the linguistic turn had a startling implication: If the world is sculpted by our cognition of it and our cognition is sculpted by our language, then the world is sculpted by our language. Indeed, we do find Wittgenstein saying in Tractatus Logico-Philosophicus, the limits of my language mean the limits of my world.19 We also find Heidegger writing in Letter on Humanism that language is the house of being,20 and in The Origin of the Work of Art that language is the happening in which for man beings first disclose themselves to him each time as beings.21 These assertions, of course, require detailed discussion, which will be attempted later. Suffice it here to note that the language to which Wittgenstein and Heidegger were referring presumably was not cetacean, extraterrestrial, or angelic.

3. Antirealism and its Varieties. The antirealist thesis is not entirely unfamiliar or even unacceptable to science or even common sense. We know that dogs can hear sounds we cannot, eagles see things too small for us to see, and computers calculute much faster than we can. We take for granted that things look very different to a fly, to say nothing of how they might look to and be understood by an extraterrestrial life-form with sense organs, concepts, and modes of
33

reasoning completely different from ours; space scientists puzzle over what pictures or symbols to engrave on space probes that might reach intelligent beings. Russell explained Kants transcendental idealism by saying: "If you always wore blue spectacles, you could be sure of seeing everything blue.22 Neuroscience unambiguously holds that our brain is not a tabula rasa on which the senses just write information. Educated people are likely to find the idea that the physical world is totally unlike what physics and astronomy say idle, if not empty. Before the insurance payout to the leaseholder of the two World Trade Center Towers, it was necessary to decide whether their destruction on 9/11/2001 counted as one or two events, a decision obviously not dictated by any facts in the world.23 Our beliefs about the history of the United States depend on what history teachers have told us and what history books we have read. The difficulty, sometimes impossibility, of faithful translation from one language to another is notorious, especially in the case of poetry. What is expressed in mathematical symbolism may be inexpressible in a natural language, and vice versa. We often find a picture of a happy childs face or of a city devastated by a hurricane worth more than a thousand words, and a caricature sometimes tells us more than any text could. What the world might be in itself, i.e., as neither known nor knowable, does seem to many of us, as it did to Goodman, not worth fighting for or against.

Of course, as a philosophical doctrine, antirealism is worthy of interest only if argued. The most impressive traditional arguments can be found in Kants defense of transcendental idealism. Arguments for antirealism are offered also in its contemporary versions, such as Dummetts, Goodmans, and Putnams, as well as in Heideggers ontology of being-in-the-world and Sartres existentialism.
34

But the standard arguments have been densely abstract, often enigmatic, and thus seldom persuasive. Perhaps, as Berkeley held, we cannot know the objects we perceive unless we perceive them, but it hardly follows that they cannot exist unperceived. Perhaps, as Kant argued, we can know things only as they are for us, not as they are in themselves, but his premise that for knowledge to be possible, the objects of knowledge must conform to knowledge, rather than knowledge to its objects is hardly less obscure or more plausible than the conclusion. Perhaps, as Michael Dummett argued two centuries later, a realist interpretation of a sentence requires understanding what would count as its conclusive verification or its conclusive falsification, and such understanding is possible in the case of few if any sentences. But this is comprehensible only to a few professional philosophers, and even they seldom find it persuasive. Hilary Putnam argued for a version of antirealism by saying that it does not require us to find mysterious and supersensible objects behind our language games that we actually play when language is working.24 But even if there are objects behind our language games, they need not be supersensible, and supersensible objects need not be mysterious (love and hatred are familiar but they are not objects of the senses, whether as events or as dispositions). A clear and plausible defense of antirealism, I suggest, must bypass the standard arguments. It must start afresh, from specific and readily understandable truths, not abstract and vague philosophical assumptions.

The reasoning underlying the standard arguments for antirealism, in Berkeleys, Kants, as well as its recent versions, may be sketched as follows: (1) We cognize only what we can, i.e., have the capacity to, cognize. This is a tautology. Therefore, (2) there is no
35

reality, no world, that is independent of our cognitive capacities. Of course, (2) does not follow from (1). What follows is another tautology: (3) we cannot cognize reality independently of our cognitive capacities. Contemporary antirealists often argue on the basis of (1) for (2), not (3), probably because the negation of (2), namely, Kants view that (4) there is a reality, things-in-themselves, that is independent of our cognitive capacities, seems to them idle. But there is at least one very good reason for (4), namely, that (2) implies an absurd sort of cosmic humanism, perhaps human creationism, the proposition, presumably held by no one, that the whole world from the page you are reading now to the outermost known galaxies, and since the Big Bang to the farthest conceivable future depends for its existence and nature on the minds, cognitive capacities, of humans, certain inhabitants of one of its planets fauna. Because of their forbidding level of abstraction, the standard arguments leave unclear both what they claim and what motivates them.

I shall not ignore these arguments, but my focus in this book will be on specific and readily understandable truths that lead to antirealism. Arguments for antirealism from such truths are very different from the standard arguments. They have the following form: (1) we cognize (perceive, understand, describe) the world as necessarily having a certain uncontroversial and familiar specific feature. But it is obvious (2) that the world does not, perhaps cannot, have that feature. Therefore, (3) the world as we cognize it, as it is for us, is not as it is in itself.

The major defenders of antirealism, from Kant to Goodman and Putnam, also offered arguments of this second sort. In defense of his obscure thesis of the ideality of space,
36

Kant pointed out that we can imagine only one space, and that we can imagine it as empty but not as absent. Regarding the ideality of time, he noted that all objects of sense, outer and inner, are necessarily in time, and that time is necessarily one-dimensional. Regarding the ideality of causality, Kant argued that we necessarily conceive of the objects of sense perception as causally related but do not perceive causal relations between them. Goodman dazzled his readers with examples of features of the world that are best understood as made by us, as how we perceive, conceive, or represent them in language or in art, not as how they are in themselves.25 We see the sun rising in the east, majestically moving overhead, and setting in the west, but if educated we know that it is we, not the sun, that is moving. The fairness of samples is a sacrosanct requirement both in science and business, but there are no objective criteria for it. We often see the world, at least briefly, as radically different after we watch a film, read a novel, or hear music. We conclude at time t that all emeralds are green because we have observed only green emeralds, but the same observations support with equal logical legitimacy also the conclusion that all emeralds are grue, if grue applies to all things observed before t just in case they are green, and to other things just in case they are blue. We reach the former conclusion because green, not grue, is entrenched in our linguistic practice. Putnam pointed out that we can count the objects in a room (a lamp, a table, a chair, a ballpoint pen, and notebook) and come up with the answer five, but that if we also count their mereological sums and ignore the null object then we come up with the answer 31.26 Such examples were often these philosophers most persuasive arguments for antirealism.

The specific and readily understandable truths on which I shall focus in this book are quite different and less impressionistic. The first is that, as Wittgenstein held in the
37

Tractatus, the world, if there is one, is the totality not of things but of facts, in the robust Russellian sense of fact.27 (Its ordinary sense, in which many speak even of false facts, is too vague to be of philosophical value.) Most philosophers today would deny that there are such entities as Russellian facts, and I would too. So would common sense, since they are supposed to be entities categorially different from the individual objects, properties, or relations familiar to common sense. But the category of fact is essential to realism regarding the world, which following etymology I shall call cosmological realism, even if not for understanding realism regarding individual things, which, again following etymology, I shall call ontological realism. The reason is simple, obvious, and independent of Wittgensteins views. If Jack admires Jill but Jill does not admire Jack, what would distinguish the world in which this is so from the world in which Jill admires Jack but Jack does not admire Jill, the world in which they admire each other, and the world in which neither admires the other, if these worlds differed in no other respect? There would be no answer if we supposed that there are only individuals, properties, and relations. Only the fact that Jack admires Jill but Jill does not admire Jack, not their mere presence in the world, as well as the relation of admiration, would distinguish that world from the other three. Perhaps there are no such entities as facts, but then there is also no world, and cosmological antirealism wins by default. Realists cannot consistently hold both that there is a world and that there are no facts robust, brute facts.

The second truth I shall focus on is the nonexistence of logical objects, in Wittgensteins sense of entities to which words like if, is, or all correspond. (There is reason, to be explained in the next section, to count facts also as logical objects.) Common sense would readily agree that there are no such entities as if or all in the world,
38

though it would be surprised that anyone should think otherwise. Very few philosophers would disagree.

The third truth I shall focus on is the obvious absence from the world of generic facts, that is, facts that would correspond to what linguists call generic statements, usually of the form Fs are G, as contrasted with universal statements, usually of the form All Fs are G. Physicians believe that patients with prior strokes benefit from taking Lipitor, but they do not believe that all patients with prior strokes benefit from taking Lipitor, nor do they confuse what they do believe with their evidence for it, say, that in a clinical trial 265, or 11.2%, of the patients who took Lipitor suffered a stroke over five years, while 795, or 37%, of those who took a placebo did. Indeed, much of what we think we know about the world is expressed only in generic statements. But we do not need abstruse argument or analysis to see that no distinctive facts correspond to them, even if there were facts that correspond to the related singular and universal statements.

4. Logical Antirealism. Antirealism is not limited to metaphysics. There is also moral antirealism, aesthetic antirealism, scientific antirealism, mathematical antirealism, and so on. Even in the case of metaphysical antirealism, numerous qualifications, distinctions, and explanations are needed. No metaphysical antirealist denies the reality of everything, just as no metaphysical realist asserts the reality of everything, including the Easter Bunny. The solipsist says, Only I exist, not Nothing exists. Berkeley denied that there are material objects, those stupid material substances, but he insisted on the existence of minds and their ideas. According to Kant, as we saw earlier, material objects are
39

transcendentally ideal, dependent on our cognitive faculties, but they are also empirically real, not just present in the mind or mere fancy. Bertrand Russell distinguished between existence and subsistence: some things do not exist, yet they are not just nothing they subsist; for example, material objects exist but universals only subsist. According to Wittgenstein, some things cannot be said, i.e., represented in language, but they show themselves in what can be said. Among them are those that matter most in logic, ethics, aesthetics, and religion.

When a position unqualifiedly asserts the reality of certain items, I shall call it unqualified antirealism with respect to those items, or just antirealism if the context makes the adjective unnecessary. I shall use antirealism as an umbrella word for all positions that are not unqualifiedly realist, however different they may be in other respects. Antirealism may be too strong and perhaps misleading term to apply to all of them, but there is no convenient alternative to it in the current philosophical lexicon. Nonrealism might be such an alternative, but it is not obviously less strong than antirealism.

Stated briefly and crudely, the antirealist thesis that the world, at least insofar as it is known or knowable, depends on us would convince no one. It gains plausibility, however, if we agree with Wittgenstein that logic pervades the world,28 that the world must have a logical, not just spatiotemporal, physical, and causal, structure, and then agree with him also that there are no logical objects.29 He wrote, My fundamental idea is that the 'logical constants' are not representatives [nicht vertreten]; that there can be no representatives of the logic of facts.30 Wittgenstein meant that logical expressions,
40

such as the sentential connectives (not, and, or, if...then) and the quantifiers (all, some), which are needed for expressing the logical structure of the world in language, stand for nothing in the world. I shall call cognition that requires their use logical. Logical cognition is indispensable for any cognition beyond that of babes. It includes but should not be confused with the sort of cognition pursued by the branch of philosophy called logic. Statements expressing logical cognition usually contain also nonlogical expressions, while those in logic contain only logical expressions. I shall call the former logical statements and the latter statements of logic. All men are mortal is a logical statement because it includes the logical expression all, but it is not a statement of logic. Few, if any, sstatements are not logical in this sense. Hence the power of what I shall call logical antirealism.31

We arrive at logical antirealism if we agree with that logical expressions stand for nothing in the world. It is a far-reaching version of metaphysical antirealism, yet almost everyone would find it plausible. If a realist interpretation of a true statement is one that pairs the statement with a fact in the Russellian sense of fact, i.e., an entity that makes the statement true but is categorially different from anything mentioned in the statement, then almost everyone would find plausible an antirealist interpretation of general statements, whether universal (e.g., All men are mortal), particular (e.g., Some men are mortal), or generic (e.g., Men are mortal), and of compound or molecular statements, whether negative (This page is not blue), conditional (If this page is white then the next page is white), disjunctive (Either this page is white or it is blue), or conjunctive (This page is both white and rectangular). Few believe that there are in the world irreducibly universal, particular, generic, negative, conditional, disjunctive, or
41

conjunctive facts, though they may believe that there are atomic facts, e.g., the fact to which the statement This page is white might correspond. Even fewer are those who believe that there are in the world entities such as all, not, if, or, and, which would be the defining constituents of such facts. Yet, as Sartre eloquently argued, negation plays an essential role in cognition; his striking example was seeing that the person you expected to meet at a caf is not there. And Gustav Bergmann pointed out that there could be no laws of nature if generality, meaning general universal facts and the quantifier all that makes them general, were not in the world: most laws of nature are universal statements.32

Spatiotemporal/physical structures and even individual things may also involve logical structure, in which case they too would not exist if logical structure does not, but whether they do involve logical structure is a question that goes beyond the realism/antirealism issue. It belongs in the philosophy of space and time and in general ontology. Logical antirealism as such does not deny the reality of spatiotemporal/physical structures or of individual things, it denies only the reality of logical objects. This is why it is more plausible than, say, Kants or Goodmans antirealism. Even though it has much the same metaphysical bite, it does so indirectly and in a principled fashion.

If facts are logical objects, then declarative sentences themselves, not just the sentential connectives and the quantifiers, would be logical expressions. We do not learn sentences as we learn names of things, we learn to make sentences. The prelinguistic child does not find facts under the Christmas tree, it finds toys, and later learns to talk about them. Facts may be counted as logical objects because, like the sentential connectives and the
42

quantifiers, they are required by logical, not empirical, considerations. This is why Wittgenstein counted the concept of fact as formal, like the concept of object, thus unsuited for literal application.33 This is why he denied that there can be representation in language of the logic of facts. We may doubt the existence of such entities as facts. But if the word is the totality of facts, as Wittgenstein held, then if there no facts there is also no world.

Logical antirealism is not the only species of metaphysical antirealism. There is one corresponding to each species of cognition. The ancient and most familiar version is perceptual antirealism. It denies that there really are objects corresponding to what we seem to perceive through the senses. Berkeleys to be is to be perceived was its brief but memorable slogan. But perceptual antirealism need not be so drastic and implausible. It can be limited to only some features of perceived objects, to their secondary qualities, such as color, and then it would represent the view held by contemporary science and most educated people: we seem to see colors but in the world there are only the light waves that initiate vision.

More innovative is the species of metaphysical antirealism we may call conceptual objects, perceived or not, depend on our understanding of them, on the concepts we possess. Conceptual antirealism was Kants major contribution to philosophy. It can be limited to just some concepts and thus gain greater plausibility. It can be limited to what Kant called the pure concepts of the understanding, such as causality. It can be limited even further to the concept of identity, which is presupposed by the acquisition and application of all other concepts. The acquisition of language presupposes possession of
43

that concept: to learn words one must be able to recognize sounds or marks. There could not be anything in the world that corresponds to the concept of identity and provides objective standards for its application. Even if identity were a relation in the world and we were aware, perceptually or intellectually, of that relation, we must be able to recognize it, to see it as the same again.34 To function as a standard of correctness in applying a concept or word, this relation would need to be recognized itself as identical with our original paradigm of identity. And there would be nothing that would determine the correctness of that recognition.

The kind of conceptual antirealism that became characteristic of 20th century philosophy after it took the linguistic turn is linguistic antirealism, according to which our cognition of the world is dependent on language, if not on the particular characteristics of the language we speak, as Benjamin Whorf thought, then on the universal characteristics of all languages, in Noam Chomskys sense. Conceptual antirealism would entail linguistic antirealism if concepts are words or syntactic structures, rather than ideas in the mind. And if limited to logical expressions and syntactic structures, it would be also a version of logical antirealism. But not all versions of logical antirealism need be linguistic. We may allow, as Kant did, for the possibility of logical concepts that are purely psychological, not linguistic. This would be the case, for example, with a logical antirealism that is limited to the concept of identity.

Logical antirealism is the most plausible version of metaphysical antirealism. Few would disagree with Wittgenstein that there are no logical objects no items, no fragments of the world, that correspond to the propositional connectives and the quantifiers, and thus
44

no distinctive facts that correspond to irreducibly compound and general statements. No such items and facts belong in what can be said, and so they could not count as ordinary denizens of reality. But, Wittgenstein also held, they can be shown. Wittgensteins distinction between saying and showing thus introduced a position that is neither unqualified logical realism, like that of Frege and Russell, nor unqualified logical antirealism. We may call it semirealism.35 In chapter 4 I shall argue that the distinction between saying and showing has a clear and noncontroversial application to ordinary, not just logical, pictures, and that Wittgensteins picture theory of meaning on which the distinction rested was no more than a sophisticated version of the traditional view that meaning and thought involve representations: ideas, even mental images.

Before Kant, the tautology on which antirealism rests that our knowledge of the world depends on our cognitive faculties led to skepticism. But antirealism should not be confused with skepticism, though they are similar in some respects. Both are rooted in the commonsense belief that things may not really be what they seem (appear) to us to be. Both appeal to the proposition that we can never discover what things really are because any discovery is still a discovery of what some things seem to be. The skeptic concludes that we can never know what things really are. The antirealist concludes that there is no distinction between what things really are and what they seem to us to be. This is why, left unexplained, antirealism is paradoxical, while skepticism is at most outrageous. At the risk of gross oversimplification and trivialization, let us take feeling a headache as an example of a seeming, appearance, i.e., of a cognition, and the felt headache as an example of the thing, res, reality, that is cognized. The antirealist would say about this simple, protocol-Berkeleyan, case that there is no difference between the feeling of a
45

headache and the felt headache, and the skeptic would say that we can never know whether we have a headache without feeling it. The realist presumably would say that there are, or at least could be, headaches no one feels.

Antirealism is a metaphysical view, skepticism belongs in epistemology. We shall see in chapter 2, however, that much of traditional epistemology can be understood really as logic, since its chief concern has been the validity of certain inferences. Insofar as the home of skepticism is such epistemology-as-logic, it is no more anthropocentric than logic is. But in its usual formulations skepticism is overtly and unabashedly anthropocentric. It concerns the limits of human perception and understanding. And today the proper study of these limits belongs in psychology and especially neuroscience. Whether certain brain states are the outcome of external stimulation, and if they are to what extent and in what way they represent anything external, is hardly to be determined by armchair speculation. Moreover, if what I called cognitive socialism is accepted, the epistemological question What do I know? would be replaced by the question What do we know? and traditional skepticism would lose much of the purchase it has had. We might conclude that there are some things that, say, physics does not know, perhaps cannot know, but our reasons would rest on certain facts about the scientific discipline of physics and bear little resemblance to Cartesian skeptical concerns.

Metaphysical antirealism enjoys little initial plausibility even among philosophers, but its indirect and usually unnoticed influence on nonphilosophical thought has been enormous. A noteworthy example is Kuhns important account of the history of science as involving relativity to paradigms. Less admirable are the popular, but careless and
46

unphilosophical, relativisms that insist that truth is relative to era, culture, race, gender, ethnic origin, or even just personal belief system. Indeed, metaphysical antirealism is a form of relativism, but it must not be confused with any of these. It acknowledges only relativity to being human. This is why metaphysical antirealism is not a sort of subjectivism. What our cognitive faculties deliver can count as objective in the straightforward sense that, in principle, it can be and often is the same for all humans, and personal divergence from it is what counts as subjective. This is why metaphysical antirealism can allow a sharp distinction between objective truth and personal opinion or fancy.

We know that there are fairly advanced nonhuman cognizers, e.g., chimpanzees and whales. There may be also extraterrestrial cognizers who are far more advanced than humans. The world cognized by whales is relative to cetacean cognition. The world cognized by chimpanzees is relative to simian cognition. Only the world cognized by humans is relative to human cognition. That relativity is biologically inescapable for us. By contrast, relativity to era, culture, race, gender, ethnic origin, or personal belief is not. We cannot literally transport ourselves to the past, but we can and often do transcend the present by viewing what past architects designed and past builders built, we read what past authors wrote, and we even hear recordings of what past speakers said. We cannot change our ethnic origin but we can transcend it by communicating, and often agreeing, with people of a different ethnic origin. We can do so because, since we belong to the same species, we share roughly the same cognitive faculties. This is why relativity to era, culture, race, gender, ethnic origin, or personal belief system is quite unlike relativity to being human.
47

In the case of epistemology and ethics, anthropocentrism faces only the paradox of implying that philosophers, the spectators of all time and existence, engage in zoological investigations. In the case of metaphysics, it faces the paradox of implying that the world itself is zoological. To avoid the former paradox, we need only redirect our efforts in epistemology and ethics. To avoid the latter paradox, we must do much more. We must understand the first-person pronouns, when used in the formulation of antirealism, as impersonal and thus as not referring to humans. This would require a radical rethinking of their role. When used in philosophical contexts, like Cartesian doubt or the realism/antirealism debate, which question the existence of the world itself, I and we cannot be taken without question-begging to refer to parts of that world. The rethinking of the role of these pronouns would thus require also a radical rethinking of the distinction between oneself and the world.

I mention in this book, and sometimes discuss in detail, the views of various philosophers, from Plato, Kant, and Hegel to Moore, Wittgenstein, Sartre, Strawson, Putnam, Goodman, and Dummett. But no exegesis or exhaustive exposition will be attempted for its own sake. Some readers may be surprised by the considerable attention accorded to Hegel, Moore, and the early Wittgenstein. But Hegels insistence on the cognitive priority of society over the individual suggests, when generalized, the needed radical rethinking of the distinction between oneself and others. Moores ethics remains the modern paradigm of nonanthropocentric ethics, and his account of the nature of consciousness is invaluable for the formulation of a defensible antirealism in metaphysics. Wittgensteins distinction between saying and showing provided a
48

strikingly original third alternative in the realism/antirealism debate. It has not been appreciated sufficiently perhaps because, like Hegels and Moores views, it is found obscure and difficult.

Several chapters are related to previously published articles: Metaphysical Realism and Logical Antirealism, in Richard Gale, ed., Guide to Metaphysics (Blackwell, 2002); Saying and Showing the Good, in Heather Dyke, ed., Time and Ethics: Essays at the Intersection (Kluwer Academic Publishers, 2003); Ethics Dehumanized, in Mark Timmons and Terry Horgan, eds., Metaethics After Moore (Oxford University Press, 2006); Bergmann and Wittgenstein on Generality, Metaphysica, Vol. 7, No. 1, April 2006; Epistemology Dehumanized, in Quentin Smith, ed., Epistemology: New Essays (Oxford University Press, 2008); Facts, in Studies in the Ontology of Reinhardt Grossmann, Javier C. Arteseros (ed.), (Frankfurt: Ontos Verlag, 2010); and Generic Statements and Antirealism, in Logos & Episteme, I, 1 (2010).

Part One: EPISTEMOLOGY AND ETHICS DEHUMANIZED

How

can he who has magnificence of mind and is

the spectator of all time and existence, think much of human life? (Plato)

Chapter Two: THREE VARIETIES OF EPISTEMOLOGY

1. Naturalistic Epistemology.
49

The subject matter of both epistemology and ethics traditionally has been considered human the knowledge and the good, respectively, that humans, not cetaceans or angels, seek and enjoy or lack. This, I have argued, is a mistake. Its correction calls for drastic rethinking of both branches of philosophy, for their dehumanization.

This is more easily accomplished in the case of epistemology. Throughout its history, it has wrestled mainly with issues concerning the validity of certain inferences, hardly a matter to be settled by zoological considerations. In effect, epistemology has often been a sort of logic. In this chapter I shall attempt to make clear how epistemology as logic differs from naturalistic epistemology, which celebrates the primacy of zoological considerations, as well as from subjective, Cartesian, epistemology, which is logically incompatible with zoological considerations but thereby also lacks subject matter altogether.

The case for dehumanizing epistemology is better understood in the context of the important differences among the several varieties of epistemology. Fundamental disagreements in epistemology, as elsewhere in philosophy, often arise from differences of interest, not genuine conflict. It is because of such differences that there are three main varieties of epistemology: naturalistic, subjective (or Cartesian), and what I shall call epistemology-as-logic. Naturalistic epistemology is explicily anthropocentric, humanized. To be consistent, subjective epistemology cannot be anthropocentric, though its practitioners are seldom aware that this is so. Epistemology-as-logic is as nonathropocentric as logic. All three have been with us at least since Socrates. My chief
50

concern in this chapter will be with epistemology-as-logic, but I must begin with the first and second, because they constitute standard epistemology.

That epistemology should be naturalistic may seem obvious. Its name is a synonym of theory of knowledge, the knowledge in question surely is that of humans, and humans are parts of nature, of its fauna. Epistemology naturalized is thus epistemology humanized: it is about humans. Not only does it ignore gods, angels, and extraterrestrials, it ignores also chimpanzees, whales, and bats. But thereby it also lacks the generality and abstraction distinctive of philosophy. Humans already belong in the subject matter of several special sciences that seek detailed information about such traditional epistemological topics as perception, conceptualization, and reasoning. This is why naturalistic epistemology is largely programmatic. The substantive work is done by biology and psychology. Quine, who took up the case for epistemology naturalized, often mentioned the role in cognition of what he called surface irritations, but wisely left the study of those irritations to neuroscience. Naturalistic epistemology remains focused on human matters even when straying into talk about biological but nonhuman computational, hence nonbiological, states. The intrinsic interest of such states is indisputable, but the naturalistic epistemologist considers them mainly for the light they may cast on the epistemic faculties and states of humans.

Naturalistic epistemology may be only programmatic, but the pedigree of the program is impressive. Much of Aristotles epistemology was naturalistic. When he described the parts and functions of the soul, he was doing nothing in principle different from what biologists and psychologists do today. And the rationale of the program of naturalistic
51

epistemology is impeccable. Surely, human beings are parts of nature, not heavenly spirits. But this is also why the investigation of them and of their epistemic faculties and states belongs today in the natural sciences and is empirical. Today, it would be strange to propose to investigate any part of nature nonempirically, in an a priori and armchair philosophical way. Even if the human epistemic faculties and states were faculties and states of immortal souls, a genuine investigation of them would be empirical. For a human immortal soul is still the soul of a human being, a certain animal. Both Plato and Aquinas would have agreed.

If human beings, including their epistemic faculties and states, belong in the subject matter of disciplines other than philosophy, the obvious question is what room is left for naturalistic epistemology. We saw in the Introduction that concern over this question may explain the shift to a conception of philosophy as just conceptual, not factual neither about natural facts nor about nonnatural facts, but about concepts or words. Hence its preoccupation with definitions, analyses, and elucidations of the workings of our language. But if concepts and words are in nature presumably in human brains or languages they, too, lie outside philosophers professional competence: there is neuroscience, as well as linguistics and lexicography. If concepts are not in nature, e.g., if they are Platonic Forms, then they should be of no concern to an epistemology that claims to be naturalistic. The investigation of brain-states and words calls not for definitions, tested by subjective intuitions, but for meticulous empirical descriptions and fruitful hypotheses, to be tested by standard scientific methods. The very idea of aiming at definitions of brain-states is foreign to neuroscience. It may offer some occasionally, but its interest obviously is elsewhere. As to words, more than half a
52

century ago Wittgenstein pointed out that they are not used in accordance with necessary and sufficient conditions, and thus their uses cannot be captured in definitions.

Perhaps the most familiar project in recent epistemology was the search for a definition of knowledge, which preoccupied it from the 1960s through the 1980s. It was born in the late 1950s, when A. J. Ayers Problem of Knowledge and R. M. Chisholms Perceiving appeared. But the project was out-of-date even at its birth. Thirty years earlier Wittgenstein had written: If I was asked what knowledge is, I would list items of knowledge and add and suchlike. There is no common element to be found in all of them, because there isnt one.36 Linguists and lexicoghraphers of course agreed. A famous paper by Edmund Gettier, a student of Wittgensteins disciple Norman Malcolm, argued the point in the early 60s. But few of those who wrote the thousands of pages devoted to discussion of his paper seemed aware that, whatever its authors intentions might have been, it called not for greater diligence, sophistication, or imagination in the project of defining knowledge but for its abandonment.

Epistemology is the theory of knowledge, but the word knowledge stands either for a disciplinary, essentially social achievement, such as grammar (Aristotles favorite example), astronomy, and arithmetic, or for a personal achievement. The study of knowledge as a social achievement belongs in the history of science and the sociology of knowledge. Investigation of it would be, of course, naturalistic, essentially historical and sociological. But epistemologists usually have been interest in knowledge only as a personal achievement.

53

This interest may assume one of two very different forms. I may ask whether, how, and what knowledge is possible for me a human being, taking into account also what knowledge is possible for your or Jack. If so, my epistemological endeavor would remain explicitly anthropocentric and therefore would count as naturalistic. It would be objective, even if rather narrow in subject matter. But I may ask instead whether, how, and what knowledge is possible for me in abstraction from the fact that I am human and thus ignoring what knowledge is possible for your or Jack. This would be the question that the philosophical skeptic who respects consistency would ask, especially if it concerns the philosophical topic of knowledge of the existence of an external world, things outside, among which would be rocks, trees, stars, but also human bodies, such as yours, Jacks, and mine. We may properly describe an epistemology limited to this question and to such concerns as subjective.

2. Subjective Epistemology While naturalistic epistemology lacks sufficient generality to be philosophical, subjective epistemology initially seems to lack generality altogether. It seems to be baldly and bleakly about only one person, myself, which is hardly a topic of scientific or philosophical interest, whoever and whatever I might be. But it cannot presuppose, explicitly or implicitly, my humanity, including the existence of my body, my language, and my place on earth or in history, for then it would be naturalistic, though also rather narcissistic one. It would not be subjective. The subjectivity of subjective epistemology does not consist in its being about a solitary human being.

54

The term subjective is ambiguous. It should not be understood as a synonym of mental or mentalistic. If minds are immaterial, your mind would be no less objective than would mine. And if I could refer to my brain only as my brain, my brain would be something subjective. Quines rejection of Cartesian epistemology on naturalistic grounds was both too narrow and too wide. It was too narrow because what was characteristic of Descartes epistemology was not its subject matter, presumably a thinking thing, but the exclusive use of first-person indexicals in its defining initial stages. In his argument I think, therefore I am Descartes did not argue for the existence of Descartes, a Frenchman who of course possessed a material body even if also an immaterial soul. To be consistent, he could only use I, as of course he did. And one need not be naturalistically, scientifically, minded in order to reject a theory that is dependent on the exclusive use of first-person indexicals. Quines rejection of Cartesian epistemology was also too wide because, as R. M. Chisholm pointed out, in its initial stages, including the proof of I exist, Cartesian epistemology was consistent with a naturalistic, even materialist ontology.37 For all Descartes knew when stating the proof, the pronoun I in it might have referred to a material, not a thinking, thing. The truth is that, pace Descartes and almost all other philosophers, his cogito had no ontological content at all.

Subjective epistemology is not a capricious narrowing of the subject matter of naturalistic epistemology from all humans to just one. If it were, it would be of no philosophical interest and also would not be subjective. It arose, most notably with Descartes, Berkeley, and Hume, as a distinct variety of epistemology in order to face the skeptical challenge. The skeptic cannot assume that he or she is human, since being human
55

involves having a body, which is a part of the material world the existence of which the skeptic doubts. Therefore, the subjective epistemologist also cannot make this assumption, at least when pursuing the Cartesian project of answering the skeptic. Subjective epistemologists cannot consistently write, as the greatest of them did, books titled An Essay Concerning Human Understanding, A Treatise Concerning the Principles of Human Knowledge, or A Treatise of Human Nature. If true to their titles, such books would beg the question against the skeptic.

Subjective epistemology is essentially Cartesian, though it was anticipated by the Greek skeptics, especially Sextus Empiricus. Its raison dtre was the project of answering the skeptic. Had it succeeded, its mission would have been accomplished, and there would have been room left only for naturalistic epistemology and epistemology-as-logic. Naturalistic epistemology does not beg the question against the skeptic by taking its subject matter to be humans because it is not concerned with the skeptics question. Indeed, though a theory of knowledge, it need have little concern with knowledge itself. It is best taken to be concerned with cognition, as this is understood in cognitive psychology and the other cognitive sciences, i.e., as the faculty the successful employment of which leads to knowledge. That faculty is of interest even when not employed successfully, when it leads to false belief sense perception, imagination, memory, and intelligence are worthy of investigation for their own sake. For subjective epistemology, however, only success counts, since as an attempt to answer the skeptic its concern is with the veridicality of cognitive states. Its focus, therefore, must be on knowledge. Alleged cognitive states such as rational, justified, probable, or warranted belief or opinion are at best images of knowledge, and are sought only as a consolation
56

prize. We seek them when knowledge is absent or impossible, in the hope of finding something still worth having. Let us consider these topics in greater detail.

As used in epistemology, rational, justified, probable, and warranted are technical terms, often explicitly introduced as primitive, of obscure meaning and uncertain reference. For example, in everyday usage justified is a deontic term, and thus justified belief, the central phrase in recent epistemological discussion, is a solecism. Actions that are justified or unjustified, and beliefs are not actions. If told that the phrase stands for belief resulting from reliable processes, this would be merely a verbal stipulation, far removed from common usage, and appealing to yet another technical term, reliable, which promptly required extensive explanation.

Indeed, even the word belief, whether or not adorned by adjectives likejustified, is seldom used in epistemology with clear sense and reference. If we adopt behaviorism, belief would be a behavioral disposition, moreover a multi-track one, i.e., it manifests itself in a great variety of events, e.g., what one says, what one does to oneself, or to ones spouse, children, colleagues, friends, or strangers, what one buys, sells, or reads, what one eats or drinks, etc. A translation then of S believes that p into logical notation would include all sorts of counterfactuals (which neither Russell nor Quine allowed) and would be so cumbersome in other respects that no one would bother to attempt it. It certainly would have neither the form B(S,p) nor the form B(S,x,y, R) where x, y, and R stand for constituents of p. If we take belief to be a brain state, process, or condition, then the translation of S believes that p into logical notation would need to be provided by neurologists, though they have little use for such a vague sentence. And if
57

we took belief to be an irreducibly mental state or condition, plain respect for the phenomenological facts would tell us that beliefs are as rare as experiences of faith or strong conviction. For example, it would be absurd for me to say that I believe, justifiedly or not, that I am typing now.

The truth is that in ordinary discourse S believes that p (e.g., S believes that Jones owns a Ford, when uttered as a contribution to office chatter) functions roughly as a synonym of the colloquial S thinks that p, and is no more enlightening or needed in philosophy and psychology than is the latter. It is not used to express religious faith or some other commitment that may have psychological reality and thus desrve the interest of psychology and, perhaps, philosophy. To say that I am justified in believing that I am typing now would therefore be grossly misleading.

The word justified, whether applied to such a phantom belief or to the sentences, statements, assertions, judgments, hypotheses, etc., supposedly expressing it, is often prefaced by the adverb epistemically. But this renders the use of believe even less clear because the noun corresponding to that adverb in Greek is episteme and in English knowledge, exactly the word that the phrase justified true belief is usually intended in current epistemology to replace. Much the same can be said about the adjectives rational, probable, and warranted. We resort to them when aware that we have no knowledge but also hope that we are not entirely ignorant. But none is the natural and traditional word for describing what we might have in such cases. It is the word evidence that is the natural and traditional word for that in science, courts of law, and careful everyday discourse.
58

The root of evidence, however, is the same as that of evident. And to be evident is to be seen or at least to be readily visible, whether literally or metaphorically, and therefore to be true if what is evident is a proposition or a state of affairs. Seeing is believing, the saying goes, but what it usually means is that seeing is knowing. Hence, the traditional account of knowledge as apprehension, intuition, awareness, or acquaintance, and the existence of the so-called strong sense of know, roughly that of certaionty. But for our present purposes it suffices that being evident entails being known in some sense of know, and we can bracket the question whether being known in that sense entails being evident. There is also the so-called weak sense of know, roughly that of true belief that is merely justified, and of course there are still other senses.38

A proposition is sometimes evident because it is seen to be true by itself, i.e., to be self-evident. More often, however, it is evident because it not only follows but also is seen to follow from one more other propositions that are evident.39 We seldom say, however, that the latter propositions are evidence for the former. We seldom call the premises of a valid deductive argument evidence for its conclusion, even if the premises are evident and the validity of the argument is itself evident, as it would be if its form were as simple as, say, that of modus ponens. Rather, we speak of evidence when what we want to know is neither self-evident nor seen to follow from anything evident, yet we think or hope that something else supports the proposition in some other manner.

J. L Austin wrote, The situation in which I would properly be said to have evidence for the statement that some animal is a pig is that, for example, in which the beast itself is not
59

actually on view, but I can see plenty of pig-like marks on the ground outside its retreat. If I find a few buckets of pig-food, thats more evidence, and the noises and the smell may provide better evidence still. But if the animal then emerges and stands there plainly in view, there is no longer any question of collecting evidence; its coming into view doesn't provide me with more evidence that its a pig, I can now just see that it is, the question is settled.40 Many might have think, however, that when the animal comes in view another event also occurs, that it becomes evident that the animal is a pig. They might even add that it is even self-evident, since it would not owe its being evident to something else that is evident. Taken literally, of course, the term self-evident is a pleonasm, as self-seen and self-visible would be. But the term does serve to mark the important difference between what is evident in virtue of being seen to be true by itself and what is evident in virtue of being seen to follow from one or more other propositions that are seen to be true in themselves. Even in a modus ponens argument with self-evident premises, the conclusion would not be evident unless seen to follow from the premises, i.e., unless the validity of the argument is self-evident. The conclusion would be evident only if its relation to the premises were evident.

Skepticism begins by noting that usually what we want to know is neither self-evident nor made evident by anything that is self-evident. And subjective epistemology usually tries to avoid the disconcerting implications of this fact by appealing to something else we hope is relevant to what we want to know. We may call it evidence, although, even if itself self-evident, it does not make what we want to know evident. This is how Austin used evidence in his example. Religion and the law are the noteworthy home of such uses of the word, which are often exquisitely circumspect. They are also common, even
60

if less circumspect, in the lab and the street. The notion they express is a degenerate offspring of the notion of the evident. But, though degenerate, it is usually understandable and also harmless. We may need to know, not merely believe, even in the proper sense of faith, that God exists, yet we may be aware that we do not. So we look for evidences of his existence. In the courtroom, a verdict of guilt or innocence is mandatory, though neither guilt nor innocence is self-evident or made evident by anything that is self-evident. So we look for something else we hope is relevant to guilt or innocence, and call it evidence, whether beyond reasonable doubt or not, and whether just circumstantial or not.

If we think that such degenerate evidence is strong enough, we may even say that we know that for which we take it to be evidence. Saying this would make explicit the reason we appealed to it in the first place, i.e., our desire for truth, and may seem to anoint the appeal. How strong the evidence must be to justify the anointment, however, is never made clear because, given the sort of reasons that lead us to employ that notion of evidence, it cannot be made clear. As cognitive beings, we seek knowledge because it is truth we want, not such evidence, even if we were to dress it up with phrases like epistemic probability and epistemic justification. The idea that evidence comes in degrees and that possession of it yields an approximation to knowledge, something worth having even though knowledge is absent or impossible, may suggest to some that there can be such a thing as an approximation also of truth. But while truth may be incomplete, irrelevant, or misleading, there cannot be two-thirds or 86% truth.

61

Nonetheless, though as cognitive beings it is truth and therefore knowledge that we seek, we are not purely cognitive beings. The degenerate uses of evidence and know in religion and the courts of law are defensible. They are often needed even in the lab and the street. To go about our business we may, perhaps must, think of certain judgments as final, settled, even if we soon revisit them. There are practical reasons in religion, the court of law, the lab, or the street for resorting to a degenerate notion of evidence. But no such reasons exist in epistemology, which is neither a religion or courtroom nor a lab or the street. Our concerns in it are purely cognitive. This is why the degenerate notion of evidence is not harmless in epistemology. It gives rise to the illusion that knowledge is relatively easy to achieve, or at best that what knowledge requires is merely the limit, perhaps only ideal, of a range of degrees of evidence, of epistemic probability, or of epistemic justification, and that what falls short of that limit would nonetheless suffice. But it is never made clear for what it might suffice, given that it does not suffice for truth, since practical considerations are now irrelevant. Unsurprisingly, it has not sufficed for the skeptic.

In everyday life and thought, the degenerate notion of evidence provides a way of clearing our epistemic conscience. It is analogous in this respect to the degenerate notion in ethics of subjective duty, which provides a way of clearing our moral conscience. The weak sense of know is analogous to the weak sense of ought introduced by the notion of subjective duty. If ignorant, as we often are, of our objective duty, of what we ought to do, we may settle for our subjective duty, for doing what we think, perhaps just feel, we ought to do. We may even insist that one always ought to do what one thinks or feels one ought to do. However, just as our real concern as cognitive beings is with truth
62

and therefore knowledge, our real concern as moral beings is with doing what we really ought to do and therefore with objective duty. The weak senses of know and ought are natural, in view of the scarcity of cases in which we can use know and ought in their proper, strong senses. There is no need for legislation against them. But we are deeply aware of the difference between the weak and the strong sense when facing important matters, and then we stay faithful to the strong senses. We do not say we know we will be alive tomorrow and thus that we need no life insurance today, regardless of how healthy and safe we think or feel we are today. And we do not say that our children ought to sacrifice their lives whenever they think or feel they ought to do it.

Neither the weak nor the strong sense of knowledge is clear. The attempts in the 1960s and 1970s to define the weak sense were notoriously unsuccessful. A definition of the strong sense was not even attempted, perhaps because of the widespread opinion that all knowledge is fallible. Nonetheless, the difference between, say, knowing that Jones owns a Ford and knowing that 2 + 2 = 4 is obvious and readily acknowledged by common sense, just as is the difference between knowing that if one wants to be trusted one ought to tell the truth and knowing that one ought to tell the truth. We dont need formal definitions to see the difference.

We can now understand why the strong sense of know, which requires that what we say we know is self-evident or seen to follow from something self-evident, has been central in subjective epistemology. Self-evidence and what it makes evident are the core of what we have in mind when seeking knowledge about important matters. The attraction of religion is that it promises certainty about matters of ultimate concern, not
63

mere probability. The attraction of subjective epistemology is similar. Its main topic is also a matter of ultimate concern, namely, whether material things the earth and the sun, your body and mine really exist. The existence of other minds, another of its topics, is also a matter of ultimate concern, though it is somewhat recherch. So are the validity of induction and the reliability of memory. The focus on epistemic probability or epistemic justification, rather than knowledge, came about when it became obvious that we cannot have genuine knowledge of such matters, yet we remain unwilling to accept skepticism. The epistemological works of Roderick M. Chisholm were influential in this shift of focus because they exemplified that motivation especially clearly.

I have briefly dwelt on the concept of knowledge because it plays a central role in subjective epistemology. But Descartess principal question was not What is knowledge? This question had been asked before and usually answered, briefly and informally but sufficiently, in the same way knowledge is seeing, apprehension, grasping, getting hold of the truth and then steadfastly holding on to it. The Cartesian question that inaugurated subjective epistemology was rather whether we have knowledge of anything, especially of an external material world. Cartesian epistemology began by taking skepticism seriously, hoping to refute it. And for this reason it was essential that it ask whether I have knowledge, that is, to employ an indexical, the firstperson pronoun I, rather than a name like Descartes or a definite description like the author of the Meditations.41 For had it done the latter, the skeptic would have accused it it of question begging. The skeptic, who questions the existence of the external material world, questions also the existence of human beings, including Descartes or the author of the Meditations, since they, or at least their bodies, are parts of the material world.
64

In the First Meditation, Descartes could not have used I to refer to the Frenchman named Descartes, since that Frenchman had a body, part of the material world the existence of which Descartes was to prove later, in the Second Meditation. In the First Meditation, where he employed the argument I think, therefore I am, Descartes could not have been referring even to the thinking thing that in the Second Meditation he concluded he was, or had. He could not have said he was referring to his thinking thing, for how would he have answered the question, which thinking thing that was, whose thinking thing was it? Louis XIII also was, or had, a thinking thing, but Descartes did not think he was proving the existence of that thinking thing and thus of Louis XIII. Nor could Descartes infer from I think only There is a thinking, as some have suggested. Which thinking would that be? Descartess or Louis XIIIs? If it is no ones, there might be thousands of such orphaned thinkings. The existence of which one would Descartes be inferring? Epistemological ventures seldom benefit from ontological adventures.42 To confront the skeptic without begging the question Descartes needed to begin his inquiry by renouncing claims to any subject matter. He could refer to nothing when using I. Subjective epistemology must lay claim to no subject matter at all when facing skepticism.

Descartes probably did not see that he faced these difficulties because all along he thought he was directly aware of a thinking thing and its states. But even if we ignore the question about which thinking thing or states those were, his or Louis XIIIs, yet another question, also fundamental but ignored by Descartes, can be asked. If a necessary condition of awareness is that its object exist, then the skeptic would ask
65

whether Descartes was really aware of a thinking thing or its states, just as the skeptic asks whether we really perceive bodies when we think we perceive them. On the other hand, if the existence of its object is not a necessary condition of awareness, then the skeptic would question the cogency of any inference from the occurrence of the awareness to the existence of the thinking thing or its states. Descartes thought he might be deceived by God or an evil demon regarding 3+2=5, but did not see that if this is so then he might be deceived also about being aware of something, about what he thought he was aware of, or found, in his mind.

This, we may note, vitiates also Descartes several inferences from the existence of his idea of God to Gods existence. Couldnt God, or an evil demon, deceive him into thinking that he had that idea? Perhaps he did not really have it. The failure to see that in this way the skeptic could question an appeal to awareness vitiates also the familiar appeals in post-Cartesian subjective epistemology to intuition, immediate experience, or direct acquaintance. Plato pointed out the poverty of such appeals in the Theatetus, and so did Hegel,43 but they remain common in philosophy. At most the subjective epistemologist could reply that it is impossible to be mistaken, whether due to Divine deception or not, about what one is directly aware of, or is acquainted with. But this would be a technical philosophical answer to a technical philosophical question. No wonder Descartes did not appeal to it when considering what was the first truth he could know.

While naturalistic epistemology has a subject matter too limited to be philosophical, subjective epistemology thus appears to have no subject matter at all. Its raison dtre is
66

to meet the challenge of skepticism. Otherwise, there would be no rationale for distinguishing it from naturalistic epistemology, albeit it could be a naturalistic epistemology concerned, inexplicably, with just one natural object, only one human being oneself . To remain subjective, subjective epistemology must refer to the self only by means of indexical expressions such as I. To both have a subject matter and not beg the question against the skeptic, it must be satisfied with a subject matter that is an entity that can be referred to only with indexicals like I. And it must allow for the possibility that only that entity exists. But would anything be an entity if it could be referred to only with an indexical, if as Frege might have said it could present itself in only one mode and thus be referred to through only one Fregean sense? To suppose that there could be thinkers who are only Is borders on incoherence, just as to suppose that there could be times and places that are only nows and heres borders on incoherence. Even to say this has required use of the grammatical monstrosities Is, nows and heres.

Before his optimistic inferences to the existence of God, Descartes seemed to have only himself, to be in a state of absolute solitude. But in fact he did not have even himself, whether as the writer of the Meditations or just as a certain thinking thing. He had nothing. Subjective epistemology is essentially a search for an answer to the skeptic. It must not presuppose that the indexicals it employs have reference because doing so would beg the question against the skeptic. It is like a geography of here or a history of now that in principle are unable to say where here is or when is now.

Subjective epistemology is thus dependent on the use of I as a dangling pronoun, a pronoun without an antecedent noun. This is why it is often described as epistemology
67

from the first-person perspective. There are no such entities as first persons, second persons, or third persons, but there are first-person, second-person, and third-person pronouns. The vague and much abused term perspective can be misleading, but it is helpful because there are no obvious alternatives to the adjective perspectival. There is an alternative to the noun perspective, however, which is less pretentious, namely, view, as long as we understand this term not in the optical but in its ordinary broad sense. So understood, view would be a synonym of cognition. The first person, we may then say, is only a perspective in the sense that it is only a view, a cognition. It is in this sense that subjective epistemology is only perspectival. There is no entity that is just the first person, and so the subject matter of subjective epistemology is not that privileged entity. The first person is not a person. There is only a view, a cognition. I shall have much more to say on this topic in chapter 11.

Nonetheless, though subjective epistemology is only perspectival and thus lacks a subject matter, the rationale for the perspective is impeccable. Lack of subject matter does not imply unimportance. The idea of a geography of here or a history of now that are unable to say where here is or when is now does seem absurd. But the idea of traveling somewhere without being able to say where or when does not. The first-person pronoun is indispensable, not because of what it refers to but because of the role it serves. It is essential to all talk and thought, and thus to all inquiry. Subjective epistemology may be an epistemology of pronouns without nouns, but to get nouns we must, so to speak, begin with pronouns. Of course, it is not the word I that is essential, but rather its use, to which other expressions can be put. In some languages, first-person reference is achieved with a form of the verb, not a separate word. Sometimes, ordinary nouns or other words
68

that are destined to be nouns serve the purpose of achieving it. Baby might not yet be able to say I cry, Baby might just say Baby cries, but Baby then would not function as it does in Baby cries when later said by Baby about another baby.

In normal contexts, to heedfully assert a sentence p one must be willing to assert It is true that p. But to heedfully assert It is true that p, one must be willing to assert I know that p (rather than the very different I think that p or even It is probably true that p). Indeed, one must be willing to assert I know that p in order to heedfully assert He (she, Jack, the expert) knows that p. I can say that Jack knows when the train will leave even if I cannot say that I know, but I cannot say he knows that it will leave at 5 p.m. if I cannot say that I also know this. In any inquiry, one must begin with the first-person view, with the use, however implicit, of I, even if only in judgments, implicit or not, such as Ill look for it in the bush, Ill ask Jill, or Ill check the dictionary. This is a proposition neither of physics nor of metaphysics. Its like Every journey must begin somewhere, not like Every journey must begin in Iowa City. Yet the proposition enjoys the abstraction characteristic of philosophy and bequeaths it to subjective epistemology. I think must be able to accompany all of our representations, Kant held, even though, as Sartre later argued, it seldom actually does. Russell wrote, When you are considering any sort of theory of knowledge, you are more or less tied to a certain unavoidable subjectivity, because you are not concerned simply with the question what is true of the world, but What can I know of the world?You cannot go outside yourself and consider abstractly whether the things that appear to you to be true are true.44 Russell may have been wrong in thinking that there is an inside to be contrasted with an outside, but his grasp of the rationale for subjective epistemology
69

was firm. Unless one opens ones eyes and looks, one does not see. Unless there is a view, nothing is seen.

This is why the allure of the subjective turn that led to Cartesian epistemology is everpresent. It would be sad if subjective epistemology were all there is to epistemology, but outrageous to deny its essential place in thought. As a theory it is futile and usually misguided, yet it is as indispensable and unavoidable as ones awareness that to get anywhere one must start somewhere and that to see anything one must look. The mistake is to suppose that subjective epistemology is about me, even if there is such an entity as me, whether a human being or a mere thinking thing. It is the mistake of supposing that subjective epistemology has a subject matter and thus that it is a cognitive discipline, a theory of something, presumably knowledge or cognition, when in fact it only draws attention to the necessary entry into any subject matter and serves as the prelude to any discipline. Subjective epistemology must use I, or a synonym of it, yet it can refer with it to nothing, neither to Descartes nor to a thinking thing, not because there is nothing to refer to or because of the ordinary semantics of I, but because of the very nature of its project.45 That project was not a mistake. Sometimes only I (or a synonym of it) can be used, even though its use cannot be captured by mentioning what it refers to.

Antirealism is a metaphysical, not epistemological, theory, but it shares with subjective epistemology the peculiar feature we have just described. Its thesis is that the world, at least insofar as it is perceived and understood, depends on our powers of perception and conceptualization. This is not a zoological proposition, however, it is not about humans even though we are humans. We can now see better how it should be understood.
70

Indeed, the proposition is not about humans, but neither is it about nonhumans. It is not about entities at all. Rather, it is about the necessary conditions of all thought, talk, and inquiry about entities. As such, it is intimately related to subjective epistemology, as intimately as Kants philosophy was related to Humes, and can cast further light on it. That Humes skepticism led to Kants transcendental idealism was not just an event in Kants life. It manifested a turning-point in the history of philosophy, just as more than a century earlier Descartes epistemology had manifested its turn to epistemology. But Hegels absolute idealism manifested a third turning-point. The subjective epistemologist makes essential use of the first-person singular pronoun, I, but Hegel saw that it is the first-person plural pronoun, we, that is essential to full-fledged cognition. The self-centered focus on the conditions that my cognition and my heedful talk, thought, and inquiry must satisfy was broadened as well as deepened by Hegel as a focus on the conditions that our cognition and our heedful talk, thought, and inquiry must satisfy, somewhat as youth broadens and deepens into adulthood. Nevertheless, adulthood is not a substitute for youth. To heedfully assert p, one must still be willing to assert I know that p. We can now see this, however, as a prelude to full-fledged cognition, not as the barrier to cognition that the subjective epistemologist feared and the skeptic exploited.

3. Epistemology-as-Logic. Subjective epistemology has no subject matter. Naturalistic epistemology does, but its subject matter is human and thus lacking the abstraction and generality needed to be philosophical. The third variety of epistemology, epistemology-as-logic, has a subject
71

matter that exceeds the bounds of the special sciences. Like formal logic, it is unambiguously dehumanized and belongs in philosophy. All three varieties of epistemology, however, are defensible within the limits of their very different yet not incompatible concerns. Their differences call for mindful distinctions, not mindless quarrel.

In the Introduction I noted that in ethics some have eschewed the need for empirical investigation by taking the deontic way, focusing on what we ought to or at least are morally permitted to do. Some have taken the deontic way also in epistemology, by focusing on what we ought to or at least are justified in believing. They have thus avoided competing with the empirical sciences. Whatever its merits in ethics, the deontic way is natural in epistemology insofar as its concerns have to do with the validity of certain inferences and thus are essentially logical. Logic, the art of reasoning, is commonly said to tell us what may and what must not to be inferred. The inferences of interest in epistemology are seldom deductive, and for this reason the need for grounding them is especially evident. Much of epistemological research has consisted in the search for such grounds.

Contrary to what textbooks sometimes say, formal logic is concerned not with inferences as activities, presumably human and thus properly studied by psychology, but with their formal validity, the relation of the truth-value of the premises to the truth-value of the conclusion, in particular, the formal consistency of the conjunction of the former and the negation of the latter. Its general subject matter thus consists of alethic relations, in the broad and etymologically proper sense of relations between propositions with respect to
72

truth-value. If some propositions, or at least sentences, are neither true nor false, as a consequence of the truth-value of other propositions or sentences, this fact too would belong in its subject matter. Formal logic exemplifies the generality and abstraction definitory of philosophy to the highest and purest degree. This is why Aristotle assigned the principles of the syllogism, especially that of noncontradiction, to the science of being qua being. This is why Frege wrote, Just as beautiful points the way for aesthetics and good for ethics, so do words like true for logic...[I]t falls to logic to discern the laws of truth...The Bedeutung [reference, meaning] of the word true is spelled out in the laws of truth.46 Elsewhere, Frege explained: What is distinctive about my conception of logic is that I begin by giving pride of place to the content of the word true ....47 If metaphysics is the science of being qua being, logic may be said to be the science of being qua truth, ethics of being qua goodness, and aesthetics of being qua beauty. Indeed, all four belonged among what the medievals called transcendentalia.

Epistemology-as-logic differs from formal logic by focusing on the validity legitimacy, cogency, worth of certain nonformal inferences, but its subject matter, like that of formal logic, consists of alethic relations, in particular, the relation of the truth-value of the premises of the nonformal inference to the truth-value of its conclusion. It enjoys the level of generality and abstraction characteristic of philosophy. Like formal logic, it is concerned not with inferences as human activities but with the alethic relations they exemplify. Unlike subjective epistemology, it does not lack subject matter, it is not just perspectival. And unlike naturalistic epistemology, which does have a subject matter, it is not just programmatic. Of course, epistemology-as-logic does apply to human matters, just as formal logic does. But it is not about them. There is nothing puzzling about this.
73

Arithmetic is applicable to humans, as well as to bats and to stars, but it is not about humans nor about bats and stars. It is about numbers.

In attempting to answer the skeptic, subjective epistemology hoped to find cogent inferences, formal or nonformal, from premises it deemed known to be true, even if as minimal as I think. Epistemology-as-logic, however, does not ask whether the premises of such an inference are true, nor does it agonize if the inference is not deductive. It is free from obsession with skepticism, just as naturalistic epistemology enjoys such freedom. In this it follows the lead of its older and more mature sibling, formal logic. In evaluating a deductive argument, formal logic is not concerned with the truth of its premises. And, we may note in passing, it also does not fret that, even if they are true, the formal validity of the argument might not suffice for the truth of its conclusion because God might deceive us about logic just as he might deceive us about arithmetic, a possibility Descartes did worry about.

Epistemology-as-logic may aim at a general theory of the alethic relations exemplified in nonformal inferences. But it began by examining particular kinds of such inference, just as formal logic began with the examination of Aristotelian syllogisms and only later, mainly through Freges work, offered a general theory of the alethic relations they exemplify. Inferences involving probability, induction, and abduction are standard topics in subjective epistemology. They would also be topics in epistemology-as-logic, though in abstraction from possible use against skepticism, which was the reason for subjective epistemologists interest in them. Appeals to probability, of course, failed to answer the skeptic, who either denied that mere probability suffices (would a religious person be
74

satisfied if told that God only probably exists, and would anyone be satisfied if told that other people only probably exist?), or questioned the truth of one or more premises (does the universe really display design? does other peoples behavior really display thought or feeling?) Nonetheless, the calculus of probability remains an established discipline of some distinction. Appeals to induction and abduction, notoriously, also have failed to answer the skeptic. But they remain standard topics in the philosophy of science, which seldom strays into Cartesian doubts.

Of special interest to epistemology-as-logic would be relations of nonformal entailment. That there are such relations is usually acknowledged independently of epistemological concerns. A standard example is the entailment of being colored by being red, and anyone who, like Kant, regards mathematical truths as necessary but synthetic, i.e., not reducible to logical truths, allows also for nonformal entailments in mathematics. Neither mathematicians nor philosophers of mathematics worry that God might be deceiving us about 3+2=5.

An inventory, much less detailed discussion, of all nonformal alethic relations is neither possible nor needed here. I shall limit myself to an especially important one: the relation of presupposition. Bringing to light and focusing on that relation was the turning-point in the development of epistemology beyond its subjective stage. If we call subjective epistemology Cartesian, then epistemology-as-logic, insofar it focuses on the relation of presupposition, may be called Kantian. Hence the application of Kants term transcendental to recent arguments from presupposition. Instead of presupposition Kants term was condition, and his project in the Critique of Pure Reason can be
75

described as discovering the conditions that the several levels of cognition must satisfy. But epistemology-as-logic need not adopt Kants essentially mentalistic approach to epistemology or any of his specific doctrines. It is not what Kant meant by transcendental logic.

The relation of presupposition became a major topic in 20th century philosophy because of Strawsons criticism of Russells theory of descriptions.48 Strawson gave as examples the presupposition of There is a king of France by The king of France is wise, and the presupposition of He is not dead by He cares about it.49 A proposition p presupposes a proposition q, according to Strawson, when both If p then q and If not-p then q are true, in other words, when q is a necessary condition of both the truth and the falsity of p, thus of ps being a coherent proposition at all, though p does not formally entail q and not-p does not formally entail q.50 In the example of the king of France, the presupposition manifests itself in our dismissing as confused, not as inconsistent, both anyone who said that the king of France is wise but denied that France has a king and anyone who said that the king of France is not wise but denied that France has a king. Our judgment of anyone who said, He cares about it but he is dead, would be similar.

Presupposition is neither entailment, formal or nonformal, nor a probabilistic, inductive, or abductive relation. This is why appeals to it have seemed to provide answers to skepticism that are entirely different from the usual answers. The latter are almost certainly either formally invalid or contain premises the skeptic finds to be as questionable as the conclusion. The anti-skeptics predicament has been that to answer the skeptic one must assume more than the skeptic would allow, but if one assumes less
76

then the answer does not follow. In appeals to presupposition, however, the consequence of denying the presupposed proposition is not the falsity but the incoherence of the proposition that presupposes it, whether a trifling incoherence as in the example of the king of France, or a deep one as in the examples I shall give shortly.

Some have said that presupposition is merely a feature of language, just internal or pragmatic, not logical or semantic, as if pervasive features of language are ever merely features of language. Aristotle defended the principle of noncontradiction not by trying to infer it from more certain propositions, but by showing that it is presupposed even by reasoning intended to cast doubt on that very same principle. Russell repeatedly pointed out that the proposition what follows from a true proposition is true is primitive and presupposed by all deductive reasoning.51 The complaint that presupposition has no place in logic is not just false, it shows misunderstanding of the very nature of logic. Of course, that the natural sciences are rife with presuppositions has always been evident. Physics presupposes, does not discover, the existence of a spatiotemporal world. Psychiatrists presuppose, do not discover, that it is not evil spirits that cause mental illness.

The examples from Strawson I have mentioned may be of little intrinsic interest. This cannot be said of those in his major metaphysical work, Individuals, or his book on Kant, The Bounds of Sense. Certainly, it cannot be said of the examples in Kants own works. Kant defended important but controversial philosophical propositions on the ground that they are presupposed by other propositions that are not controversial. A simple example is the presupposition that the objects of sense perception (outer sense) are in space.
77

This page is white would not be true if this page were not in space, but neither would This page is not white be true if this page were not in space. Kants argument that morality presupposes freedom is another, more familiar example. Freedom seems to be presupposed by all genuine actions, moral, immoral, or nonmoral. It is what seems to distinguish actions from mere movements. A no less famous but more difficult example is Kants argument that objective order in time presupposes causal necessity. It is complex and not to be dealt with lightly, whether in agreement or disagreement, but we need not go into its details to get a glimpse of it. If we ask whether Jack met Jill before or after she moved to town, the answer would depend in part on reasoning about when and where he could have met her. If it really mattered (as it might in a court of law), we would be foolhardy to rely on memory impressions, what Kant called a subjective play of fancy. As this example shows, what is presupposed need not be a single proposition, just as a deductive proof ordinarily does not rest on a single premise. It might even be a system of propositions, and what presupposes it might itself be such a system. This is why the philosophically interesting examples of presupposition seldom have the simple structure of the examples about the present king of France and the man who is dead. Here are three other examples.

The first is a presupposition especially relevant to traditional epistemology: the existence of a material world. It assumes many forms, and is neither simple nor obvious. In doubting the existence of the material world, Descartes presupposed its existence by presupposing the existence of certain of its parts. Therefore, his doubt was incoherent, like doubting that France has a king both when asserting and when that the king of France is wise. As G. E. Moore noted, in doubting the existence of the material world Descartes
78

would have had to doubt the existence of other philosophers, past and present, including those he had read, heard, argued with, and whose works and views were the context of his doubt, whether through agreement or disagreement.52 But philosophers are human beings, therefore inhabitants, parts, of the material world. The history of philosophy is no more a history of philosopher-angels than of philosopher-kings. Descartes could not have taken his doubt seriously as a philosophical doubt if he had considered the implications of the doubt, the details about what he doubted, what would be the case with respect to his own doubt if there were not a material world. Philosophical skepticism about the material world questions its own existence. Let me explain this further.

The philosophical context of philosophical thinking, such as Descartes doubt, is essential to it, however original the thinking may be. It is essential to it even more obviously than, as contemporary essentialists have argued, the biological origin of an organism is essential, metaphysically necessary, to that organism. The historicity of a philosophical view is no more a contingent fact than the historicity of a political event. Both bear necessary relations to their past. Neither Cartesian epistemology, nor Democratic or Republican politics in the 21st century, would be comprehensible if stripped of those relations. The skeptic therefore questions what makes it possible for skepticism to be the philosophical view it is: its roots in what some other philosophers have held. It would not exist if those philosophers had not existed. and thus if the material world did not exist. Descartes methodological doubt would not have occurred if the proposition the truth of which he doubted had not been true.

79

Indeed, the very language in which Descartes developed and explained his doubt would not have existed. Surely, employment of language is essential to philosophical thought, even if rudimentary thoughts are possible without language. Philosophical thought, whether superior or mediocre, involves argumentation, good or bad, which has a fairly complex structure, distinct premises and conclusions, each with its own structure, and logical connections rooted directly or indirectly in that structure. The terms employed in the argumentation are chosen, usually with deliberation and discretion, from a fairly extensive and often highly technical lexicon. But any language actually employed in philosophy, say, Descartes French or Latin, involves phonemes and inscriptions, and has been shaped by a human community. All three phonemes, inscriptions, and the human community are parts of the material world.

Could the argumentation be stated and explained in a private language, at least in the minimal, not necessarily Wittgensteins, sense of a language created by the philosopher alone without reliance (as in devising a secret code) on a public language? Surely such a private language, even if possible, would be too primitive for epistemology and philosophy generally. The reader is invited to attempt constructing a fragment of the language and then translating into it a page from Descartes Meditations. Writing philosophy is not like recording ones sensations. A private language for the latter might be possible, perhaps invented by hypochondriacs, but to say that the sort of rich and sophisticated language that the philosophical skeptics argument requires could be such as language would be mere posturing. Might the argument take place just in the skeptics thought, without use of language, public or private? Even if some thought without language, e.g., recalling an unusual sensation, were possible, to suppose that
80

philosophical thought might be such thought would be like supposing that we can understand differential equations, or what it is for a car to travel at 59, rather than 58, miles an hour, without using symbols. The skepticism considered in Descartes first meditation was not a tipsy sailors declaration, Maybe I know nothing. It was a professional, serious and informed, philosophical view. This is why we still take it seriously. Of course, that philosophical skepticism about the material world questions its own existence does not entail that it is false. It does not make it self-contradictory. But it does make it deeply incoherent. If the material world did not exist, then philosophical skepticism about it would not exist.

My second example of philosophically interesting presupposition is Sartres strikingly original defense of the existence of other minds. Ones acceptance of the Other is not discursive, he pointed out. It is presupposed by various psychological states of oneself, it is essential to them. Sartre dwelt at length on the experience of shame, say, when looking through a keyhole. It is an immediate shudder that runs through me from head to foot without any discursive preparation. But it is shame of oneself before the Other, even if I know that no one is actually looking at me.53 To say that what is presupposed here is only the possibility of being the object of anothers look would be to deny that the experience is genuine shame. There is that possibility almost always, At any rate, the interesting sort of skepticism about the existence of other minds, which is not just a trivial consequence of skepticism about the existence of other human bodies, questions even its possibility. It questions the very intelligibility of there being anyone other than myself. It resembles the sort of skepticism about the existence of bodies that denies, as Berkeley did, that we can even conceive of unperceived bodies.
81

My third example of presupposition belongs to an even deeper level. We may call it conceptual presupposition. What is presupposed is a particular understanding of a certain concept, not some proposition. All discussions, skeptical and anti-skeptical, of the existence of bodies, other minds, the past, etc., presuppose a particular understanding of the concept of existence. Therefore, to ask whether bodies exist presupposes some answer to the question of what it is for a body to exist, and ultimately of what it is for something to exist. Is existence a property? If it is not, then what are skepticism and anti-skepticism about? Would it be intelligible to speak of the truth or falsity of any sentences asserting or denying existence, and thus of knowledge or ignorance of their truth? Would such sentences express genuine propositions?

Standard epistemology, whether naturalistic or subjective, provides little guidance on these questions. It usually takes for granted Kants view that existence is not a real predicate, meaning that it is not a property, a determination, of a thing, res, but fails to consider the epistemological implications of the view.54 Kant wrote that the existence of a thing has to do only with the question whether [the] thing is given to us in such a way that the perception of it could in any case precede the concept.55 This was not Berkeleys to be is to be perceived. Nor was it Mills matter is a permanent possibility of sensations, though it resembled it. Kant held that to be actual (wirklich) a thing must standin accordance with the laws of empirical progression.56 But these laws, like any other laws, involve applications of the pure concepts of the understanding. If so, the actuality or existence of a thing, like its causality, was for Kant not so much a matter of discovery of empirical fact as the imposition of a concept. Exists expresses
82

not a property we find in a thing, but rather a conception of the thing that we contribute.57 This does not mean that how we apply the concept of existence is mere caprice, any more than that how we apply the concept of causality is one. What is meant is closer to what Nelson Goodman meant by the entrenchment of the predicate green, which the predicate grue lacks, though for Kant the application of the concept of existence was grounded in the activity of our faculty of the understanding, not in our linguistic practices. Clearly, Kants view of existence required that skepticism and subjective epistemology, insofar as they concern the existence of bodies, be drastically rethought or just altogether bypassed by him, as in fact they were.

Might existence be just what the existential quantifier expresses, the satisfaction of a propositional function, as Russell argued and most contemporary epistemologists take for granted?58 To suggest that it might would be a nonstarter. Whether the propositional function x is a horse is satisfied depends on what we allow as values of the variable x. Is x is a horse satisfied by both Secretariat and Pegasus, or only by Secretariat? If we say the latter, our reason would be that Pegasus does not exist, but in a sense of exist now obviously other than yet presupposed by Russells. It is its ordinary sense, which we employ in saying, for example, that the Loch Ness monster and a childs imaginary friend Jack do not exist.

In any genuine case of inference involving presupposition, there is a natural desire to think of it as a case of formal entailment, since this is the relevant alethic relation we find most familiar and understand best. And when we see that the conditional corresponding to the inference is not a tautology we are tempted to declare the inference invalid. Or, if
83

we find it compelling, we are tempted to change the conditional, perhaps by adding a premise or modifying the conclusion, so that it becomes a tautology, and we may even bless the change by calling it an analysis or translation. This is what Russell did in his theory of definite descriptions. He saw that if the present king of France is wise then, of course, France has a king. He also saw, however, that this is not a tautology. So, he proceeded to translate it into one. In the philosophically substantive cases, however, like those Kant and Sartre made familiar, no such analyses or translations are plausible. The proposition If there is objective order in time, then there is causal necessity, for example, obviously is not a tautology, and to try to change it into one by analysis would hardly be a task worth attempting. So, the remaining option is to just deny the proposition.

If the presupposition is conceptual, however, there is no such third option. For example, the skeptic can deny, however implausibly, that skepticism, like all philosophical thought, presupposes the existence of the material world, that objective order in time presupposes causal necessity, or that the phenomenon of shame presupposes the Others look. But the skeptic is not likely to question the presupposition of the concept of existence. An answer, whether affirmative or negative, to a question obviously presupposes understanding the question. The skeptic cannot deny the central place of the concept of existence in any discussion of what does or does not exist, or of what we can or cannot know to exist.

Would these examples of presupposition count as answers to the skeptic, whom naturalistic epistemology ignores and subjective epistemology cannot answer? They do
84

not refute skepticism they are not proofs that we do know what according to it we do not know but they might silence the skeptic. Whether they do or do not, however, epistemology-as-logic is interested in them independently of their relevance to skepticism. Skepticism is a problem for subjective epistemology. The question How do I know that p? that the skeptic asks is addressed to me. It has no theoretical substance, since the pronoun I in them must refer to nothing, neither to the skeptic nor to anyone else. The question that does have theoretical substance is How we know that p? where we of course refers to human beings, not gods or extraterrestrials. It belongs in naturalistic epistemology, but today it calls for answers from the natural sciences specializing in human cognition, not from philosophy. Epistemology-as-logic, of course, asks neither question, just as formal logic does not.

Chapter Three: THE PROPERTY GOOD 1.Anthropology and Conceptual Analysis in Ethics. Although ethics usually has concerned itself explicitly with human matters, there have been noteworthy exceptions. Platos theory of the Form of the Good in The Republic is a familiar example. In the 18th century, Kant distinguished sharply between the metaphysics of morals and the empirical discipline of practical anthropology. Early in the 20th century, G.E. Moore offered in his Principia Ethica a detailed exposition and defense of an ethics concerned with the nonnatural property good. Shortly thereafter, Wittgenstein outlined in Tractatus Logico-Philosophicus a subtle conception of an ethics concerned with the world, not with oneself or other humans. Plato, Kant, Moore, and
85

Wittgenstein provided both the rationale and the outline of a dehumanized ethics, however much they differed in other respects. In this chapter I shall consider Moores ethics, and Wittgensteins in chapter 4. My aim in neither case, however, will be to offer exegesis of what they say but rather to learn from it.

Moore's Principia Ethica was published in 1903, and soon became one of the signposts to the philosophy of the following 100 years. It is too early to judge how 20th century philosophy ended, but its beginning was remarkable. Both Moore's Principia Ethica and Russell's Principles of Mathematics appeared in 1903, the first volume of Husserl's Logical Investigations in 1900-01, and four of William Jamess major philosophical books in 1902-09. Wittgensteins Tractatus Logico-Philosophicus was written between 1914 and 1918. There was not a significant difference, except in style and temperament, between Anglo-American and European philosophers. The analytic/continental schism came much later. Both Russell and Husserl began as mathematicians. Frege was the German philosopher-mathematician to whom, by their own admission, Russell and Wittgenstein were both heavily indebted. In the Preface to Principia Moore wrote that his ethics was closest to the Brentanos. Russell studied and discussed Meinong in detail. James was much admired in Britain and in Europe, influenced Husserl and Wittgenstein, and was the subject of articles by Moore and Russell.

Principia Ethica remains the best example in 20th century philosophy of the sort of metaphysical ethics that is free of anthropocentrism. Moore dehumanized ethics even more clearly than Plato had done. But his book also inaugurated what has come to be known as analytic ethics, the sort of ethics characteristic of 20th century analytic
86

philosophy. The relation of analytic ethics to Moores book rested largely on his thesis in Chapter I of Principia Ethica that the property good is indefinable. But Moore made clear he had no interest in what he called verbal and the tradition calls nominal definitions. They are the business of lexicography, he wrote. Yet it was just such definitions that analytic philosophers sought, sometimes calling them analyses. The most familiar example comes from analytic epistemology, not ethics: the definitions of S knows that p, which preoccupied epistemologists from the late 1960s to the early 1980s. The definitions sought and offered by analytic philosophers were not even lexicographic definitions, which record lexical facts and are tested by empirical investigation of speech and writing. Rather, they recorded their authors impressions of lexical facts, and were tested by the authors intuition of what would in some hypothetical situation, called a counterexample if it did not fit the intuition. The question How do I know what one would say in that situation, given that I am not in it? was usually ignored. For it could be answered properly only by appealing to what I or others have said in similar situations, and this would be to make an appeal, however amateurish, to lexical fact. Even the Oxford English Dictionary is valuable mainly for the examples of usage it lists, not the definitions it distills from them.

Indeed, searches for definitions are alien even to current philosophy of language, because of three trailblazing developments in it half a century ago. The first was Quines attack on appeals to meanings generally. It is widely accepted today, but usually only pro forma. Phrases such as conceptual question, conceptual content, and conceptual connection still abound in the literature, with the noun concept explicitly used for the meaning or use of a word, not for a mental or a brain state. The second development, also
87

widely accepted but just pro forma, was Wittgensteins relentless argument in the Philosophical Investigations, posthumously published shortly after Quines article, that words are not used in accordance with necessary and sufficient conditions. He gave game as an example, but his argument applies also to good, right, reason, know, exist, and other denizens of the philosophers lexicon, which, like game, are everyday words, not technical terms introduced as abbreviations of multi-clause descriptions. The third development was Chomskys linguistics, announced four years later. It marked a striking advance by stressing the biological, largely inherited, core of linguistic competence and urging the use in the study of language of the standard methods of scientific research.

The kind of definition Moore sought was an account of the constitution of the thing, res, that was defined. It was closer to what the tradition calls real definition, though it gave not the genus and differentia but the parts of the thing. Such a definition can be called an analysis, in a sense reasonably similar to that employed in chemistry. In later years analyses were offered mainly of facts and propositions, which were taken to be nonlinguistic entities categorially different from those chemistry analyzes. Their analysis was intended to reveal logical form, and for this reason was called logical analysis. It was in such analyses that analytic philosophy took root, beginning in 1905 with Russells theory of definite descriptions and culminating in Moores claim two decades later, in A Defense of Common Sense, that he knew the proposition This is a hand to be true but did not know how to analyze it. In Principia, however, his example was the definition of a horse and consisted of an anatomical inventory. Our example might be the account of water as H2O. Moore in effect agreed with Kant that in matters of morality it is always real definitions that must be sought.59
88

Despite its inattention to what he meant by definition, analytic ethics did begin and develop in relation to Moores ethics, though by way of sustained disagreement, not agreement. Discussions of Principia seldom ventured beyond Chapter I, which alone was included in most anthologies. Usually ignored were the crucial Preface, where Moore explained what he meant by intuition and self-evidence, and thus what anyone calling him an intuitionist and foundationalist ought to mean. Also usually ignored were the beginning of Chapter II, where he explained what he meant by natural and nonnatural, thus what anyone calling his ethics nonnaturalist ought to mean, and Chapter V, where he explained his theory of right on the basis of the theory offered in Chapter I.

By intuitions, Moore wrote, he meant self-evident propositions, and nothing whatever as to the manner or origin of our cognition of them.60 And a self-evident proposition, he explained by plainly following the etymology of the word, is one that is evident but not by virtue of inference from other propositions.61 Moore did not say what he meant by evident, perhaps thinking it unnecessary. A proposition is evident, of course, if it is, or can readily be, seen to be true, either literally or metaphorically. Therefore, it may also be said to be known, in the serious and traditional sense of know, which sharply distinguished between knowledge and belief or opinion. The noun evidence, as used in court or in the lab, has a wider meaning, but the same root. Moore used self-evident for the propositions in ethics that state what kind of things ought to exist for their own sakes, i.e., are intrinsically good. He avoided the term justification, which became common in later analytic epistemology as a way of avoiding skepticism but

89

acknowledging that little of what we say we know passes any traditional test of knowledge.

As to the meaning he attached to calling something natural, Moore wrote he meant that the thing is in time. Thus a paradigm of a nonnatural thing would be a number, an abstract entity. The mere fact that a thing falls outside the subject matter of physics does not make it nonnatural. For example, irreducibly mental states would be in time and thus natural. If we said that a natural thing is one belonging in the province of the natural sciences, as Moore himself did on occasion, we would need a noncircular account of what is meant by calling a science natural, as he doubtless was aware and for this reason did not offer it as his definition of natural. The fact is that the distinction between the natural and the nonnatural did not play a central role in his book, though the phrase naturalistic fallacy did. As Moore made clear in the also ignored Chapter IV, which was devoted to what he called metaphysical ethics, even ethical theories concerned with the supersensible committed the fallacy. The fallacy was merely a case of confusing two things: the property good and some other property.

In Chapter I Moore held that the property good is nonnatural and simple, therefore indefinable, that almost all earlier ethical theories had committed the naturalistic fallacy of confusing it with another property, and that they could be refuted with what later was called the open-question argument, which in effect encouraged the reader to pay close attention to the property such a theory confuses with the property good in order to see that they are two properties, not one. But his contemporaries in the Society of Apostles and the Bloomsbury Circle, who included Russell, Keynes, and Virginia Woolf, found more important not these metaphilosophical generalities but the substantive view,
90

defended in Chapter VI, that personal affection (love, friendship) and aesthetic appreciation (contemplation of beauty, in art and in natural objects, human and nonhuman) are the greatest goods. In contrast with Mills view that pleasure alone is good or Kants that a good will is the only thing that is unconditionally good, for Moore it is personal affection and aesthetic appreciation that were the Ideal. And it is the place accorded to them in Moores ethics that prompted Keynes to rate Moore higher even than Plato. That chapter of Principia Ethica, too, has been ignored in analytic ethics, which has focused instead on the preliminary discussions in Chapter I, especially the objectivity of value it took Moore to be defending there. But, in a recent book, Brian Hutchinson points out that Moore never even entertained doubts about the objectivity of value. Hutchinson acknowledges that for us this may be a mystery difficult to fathom, but wisely suggests that the mystery is to be savored rather than solved.62

While the central tenet of Moores theory of good was that good is a simple, indefinable, and nonnatural property, the central tenet of his theory of right was that the right action, i.e., duty, in a particular situation is the action that will cause more good to exist in the Universe than any possible alternative,63 that is the best thing to do, that together with its consequences presents a greater sum of intrinsic value than any possible alternatives, either because it itself has greater intrinsic value than any alternative or because the balance of intrinsic value of its consequences does, so that more good or less evil will exist in the world if it is adopted.64 Of course, the action need not do so on a grand scale. To think that it must, or even could, would be human conceit of cosmic proportion. And cause or produce are used in the broad sense of contribute, since the action might be the best thing to do because of its own goodness or its organic relationships. Moores was an ideal utilitarianism, which unlike Benthams, Mills, and
91

Sidgwicks presupposed a theory of good that placed no limits on what items might enjoy intrinsic goodness, thus allowing that some may be actions.

Moores theory of right may thus be called cosmological. It tells us that we ought to do what would be best, all things in the Universe considered. It accords with Aquinass first principle of natural law: Good is to be done and promoted, and evil is to be avoided.65 It does imply, as Moore noted, that justice is not to be done if the heavens should fall unless, he wryly added, by the doing of justice the Universe gains more than it loses by the falling of the heavens.66 The ethical views of Russell and Wittgenstein, the other two founders of analytic philosophy, were also nonnaturalist and cosmological. But, with the exception of H. A. Prichard, a philosopher of unsurpassed acuity, and W. D. Ross, whose terminology and distinctions are still found indispensable, later Anglo-American ethics diverged in both respects. They are related. If ethics is naturalistic, then it is not likely to be cosmological. And if it is cosmological, then it is not likely to be naturalistic.

Naturalistic ethics is almost certainly ethics humanized: it is about humans, not cats or bats. So it is not cosmological. Not only does it ignore the good of the universe, it ignores that of extraterrestrials, angels, and gods, if there are any, and usually also that of nonhuman animals, plants, and rivers. Thus it lacks the supreme generality and abstraction distinctive of philosophy and alone justifying its existence alongside the other cognitive disciplines. A cosmological ethics can be expected, of course, to have application to humans, just as chemistry and mathematics can. But this makes none of them about humans.

92

We all are human, thus interested in ourselves and in other humans. But this is not a reason for making ethics to be about us, any more than it is a reason for making astronomy to be about us. As I argued in the Introduction, there is a special, deep, and often misunderstood sense in which humans may be cosmically central - namely, that leading to Kants transcendental idealism and its recent versions in Goodman and Putnam. In that special sense, Kant, Putnam, and Goodman may be said to have humanized even astronomy. But they did not hold that astronomy is about humans. Although for Kant the ultimate end of the pure use of our reason was ethical, he resolved to [keep] as close as possible to the transcendental and [to set] aside entirely what mightbe psychological, i.e., empirical,67 since the metaphysic of ethics is really the pure morality, which is not grounded on any anthropology.68

Ethics humanized is unphilosophical. Its practitioners also lacks the necessary competence. If humans are natural objects, a species of animal, we can hardly expect to have special philosophical knowledge of them, just as we can hardly expect to have special philosophical knowledge of stars or bats. I argued in the Introduction that accounts of human well-being and searches for the best explanation of human conduct do not belong in philosophy, just as accounts of human anatomy and human evolution do not. My point does not depend on a narrow use of the words natural and science. If mental states are not reducible to physical states, there could still be a natural science of them, in Moores sense of natural and the traditional sense of science in which history and political geography are social sciences. In fact there was such a science in Moores time - namely, the largely introspective psychology of James, Wundt, and Titchener. My point does depend, however, on taking competence seriously, whether in forensic pathology and medieval history or in epistemology and ethics. Genuine
93

competence requires serious training, for example, in chemical analysis or parsing Latin. Nothing analogous with respect to humans occurs in philosophy seminar rooms. If employed in hospitals, medical ethicists are expected to learn some medicine. They can be invaluable, not because they know something physicians do not, but because they are Socratic they ask questions physicians do not.

We need not go to hospitals for examples. How to achieve happiness, in the ordinary sense, recognized by both Kant and Mill, of enjoyment or satisfaction of our needs and desires, has been a stock question in ethics, with Epicurus and even Plato offering advice, but arguably the invention of aspirin and contraceptives, tractors and pesticides, air conditioning and spreadsheets, answered it better. This is especially evident in politics. In Buddhist ethics, sadly but realistically, suffering seems the primary concern, not pleasure, as in Western ethics. But Indian Benthamites hoping to learn from Americans how to reduce suffering presumably go to American colleges of agriculture and schools of public health, not to American philosophy departments.

One may ask, indignantly, what about loftier goods, not Benthams but certainly Platos and Kants, such as justice, authenticity, salvation? Especially in India, a deeply religious country, they are often thought far more important. But these loftier goods call for nonzoological considerations. Philosophers who avow allegiance to naturalistic ethics do write about some of them, at least about justice, and seldom if ever about gustatory delights. Do they think they have access to human nature that biologists and psychologists lack? Of course, they do not. As we saw in the Introduction, they are likely instead to have adopted a conception of ethics far removed from both naturalism and nonnaturalism: ethics as a conceptual, not factual, discipline. This allows them to
94

avoid both commitment to nonnatural facts and responsibility for competence regarding natural facts. Such ethics may indeed be called analytic because there is a metaphorical sense in which we do speak of analyzing concepts. But if concepts are in nature presumably in human brains or languages they too are outside philosophers competence: there is neuroscience, as well as linguistics and scholarly lexicography. If they are not in nature, Moores venture into the nonnatural was at least straightforward.

It also, unlike conceptual analysis, was not dated, though its critics relish calling it obsolete. Like the 17th century way of ideas, conceptual analysis has been out of date ever since Kant pointed out in 1787 that our business is not merely to analyze concepts but to extend our knowledge.69 It certainly has been out of date since 1951 when Quine pointed out that meaning is what essence becomes when it is divorced from the object of reference and wedded to the word.70 Like the 17th century way of ideas, the 20th century way of concepts inherited its rationale from the medievals concern with essences, but the medievals did not exclude from their concern the things in which essences were grounded. Without such grounding, the rationale for the way of concepts becomes opaque, even if we say, imitating Wittgenstein, that our concern is with how language or discourse works. What special qualifications do philosophers have for research in the workings of language?

Some have eschewed the need for empirical investigation by taking the deontic way, focusing on what we ought, or at least are morally permitted, to do. They have thus avoided competing with the empirical sciences, which are interested at most in what we do, not what we ought to do. The deontic way seems natural in ethics, where imperatives are common. But a deontological ethics must ground, not just announce, its imperatives.
95

Kant held that the ground of obligation ... must not be sought in the nature of man or in the circumstances in which he is placed, and urged that it is a matter of the utmost necessity to work out for once a pure moral philosophy completely cleansed of everything that can only be empirical and appropriate to anthropology.71 He gave Thou shalt not lie as an example of an imperative of duty, but promptly explained that it does not apply to men only, as if other rational beings had no need to observe it. Kant appealed to the nonsensible world of noumena as the ground of the commands of practical reason.

Resorting to a nonsensible ground, as Kant did, is uncommon in contemporary ethics. It is more likely to ground their pronouncements in what philosophers call intuitions, but meaning usually no more than their felt inclinations about what they would or would not say is morally right in various, usually just imagined, circumstances. Appeals to such intuitions are made also by philosophers who shun the deontic way, and they are common even in current metaphysics. Their success obviously depends on the worth of the intuitions. If the appeals are made merely as a way to bypass empirical investigation of the linguistic facts, their worth is likely to be minimal.

Like post-Gettier analytic epistemology, post-Moorean analytic ethics was unfazed by misgivings such as Quines about concepts and meanings. It clung to conceptual analysis, and thus was really neither naturalistic nor nonnaturalistic. It went through several stages. The first began in Vienna, soon after the publication of Wittgensteins Tractatus LogicoPhilosophicus and with some personal involvement by him. Ethical statements were rejected as nonsense, or at least as lacking cognitive sense. The subtlety of Wittgensteins verbally similar position, however, was missed altogether. The second was to offer a
96

positive characterization: they are expressions of emotion. But the rich literature already in existence on the emotions in psychology (from James to Arnold) and phenomenology (from Meinong to Sartre) was ignored, though it seemed to show that the emotions are not, as the emotivists thought, self-contained subjective episodes, Humean impressions of reflexion, but rather intentional states, directed upon objects, with character dependent on that of their objects, and thus in principle cognitive. The third stage, probably motivated by the experience of the Second World War, which made both the outright rejection and the emotivist interpretation of moral statements appear jejune, was to suppose that they express a special moral point of view, something psychologically no less genuine than the emotions but supposed to be less subjective, and that their function is to guide, not goad. In effect, this third stage consisted in denying moral statements a full-fledged, unqualifiedly cognitive status, yet conceding that their function is not merely that of imperatives or exclamations. Taken for granted in all three stages was that the job of ethics is to describe the meanings or uses of moral words, or the content of moral concepts, or the features and workings of moral discourse. We cannot give people what really interests them namely, an ethics that says what they should do, Moores heirs held, but we can give them an ethics that says what they mean a metaethics. This was the message even of the more recent fallback positions of projectionist antirealism and supervenience realism, where the focus remained metaethical, not substantive. Few worried that the very idea of telling people what they mean seemed paradoxical except to psychoanalysts.

97

2. The Good and the World. By taking this conceptual turn analytic ethics did not provide a genuine alternative to ethics humanized. Its new, conceptual, subject matter was either specious, or genuine but beyond its competence. The alternative provided by Moore remained. Let us return to some its details. I suggested that the role of the indefinability and nonnaturalness of the property good was relatively minor, given Moores explanation of what he meant by them. Less familiar is that Moore proposed a criterion, a test, for determining whether a thing exemplifies that property, whether it is intrinsically good. Moral thought is usually concerned with what things are good, not with the property good itself.

The criterion was the method of isolation. It consisted in asking whether a world, a whole world, that contains the thing but otherwise is just like a world that lacks it, would be better.72 The two worlds might be wholly inanimate, and even considered apart from any possible contemplationby human beings. Moore applied the criterion to the intrinsic goodness of beauty, in opposition to Sidgwicks contention that nothing appears to possess this quality of goodness out of relation to human existence, or at least to some consciousness or feeling.73

In addition to (1) being good intrinsically, in itself, independently of anything else, by just exemplifying that property, a thing (object, action, state, property) may be said to be good also (2) because it has a totality of consequences that exemplify it, which are determined empirically, or (3) because it noncausally enhances the goodness of the organic unities or wholes of which it is an element which also are determined
98

empirically. The overall goodness of a particular item thus depends on the actual or possible goodness of the world, the Universe. Pleasure is a familiar example of (1), surgery of (2), and the presence in a philosophy department of its mediocre but only aesthetician might be an example of (3). A thing may be said to be good overall when considered in all three respects. In the case of actions, consequentialism ignores (1), deontological ethics ignores (2), and (3) is ignored by both. Acknowledging all three, as Moore did, gives both consequentialism and deontological ethics their due. In particular, consequentialism is given its due because the goodness or badness of an action overall (in W. D. Rosss terminology, if not meaning, its being actually, not just prima facie, right) would depend, in part, on the goodness or badness of its consequences all of them, for anything less would be morally unacceptable.

An explicit provision of Moores method of isolation was that we could ignore anthropological, indeed anthropocentric, facts about the two hypothetical worlds. They might be supposed to be unpopulated, and even considered apart from any possible contemplationby human beings, as Moore did when applying his method to beauty. This is why all the consequences of an action and all the organic wholes to which it belongs are relevant to its being good or bad overall. For the action will have consequences even after the human species has become extinct, a fact at least the thoughtful environmentalist should not find irrelevant (would species that survive cease to matter then?) And, presumably, the action belongs to organic wholes possessing parts both past and future, with an indeterminate range, and thus requiring reference to all time the whole world.

99

The method of isolation implies important similarities of Moores views to Kants and Wittgensteins, which seldom are noticed. I am not suggesting historical connections, though as a student Moore did study Kant assiduously, even attending a course on Kant in Germany, and enjoyed a close relationship with Wittgenstein. During his stay at Cambridge in 1912-14 Wittgenstein attended Moores lectures, they camped together in Norway, with Wittgenstein dictating philosophical notes to Moore but presumably discussing not just logic and the Norwegian landscape. Moore might well have agreed with Wittgenstein that by requiring reference to the world as a whole ethics is in a certain sense transcendental, that it concerns the limits of the world, rather than just some of its contents (see the next chapter). It would be surprising if they did not notice the similarity of their views on this deepest level, despite their many differences, both philosophical and personal.

Both belonged to the society of Apostles, together with Russell, Whitehead, and Keynes. Moores ethics was a central topic of discussion in Britain at the time, not only among philosophers but intellectuals generally, especially in the Bloomsbury Circle, which included Virginia Woolf and other literary figures. There should be no question that Wittgenstein was fully aware of Principia Ethica. Indeed, in a 1912 letter to Russell, he wrote that he had been reading it, though also that he hadnt liked it, because Moore repeated himself dozens of times.74 Similarities between major philosophers are especially enlightening when their views are reached independently. Kant, Moore, and Wittgenstein shared a conception of ethics as dehumanized despite their fundamental differences in most other respects.75

I believe that the usual objections to Moores theory of a nonnatural property good are no
100

more properly motivated or philosophically astute than the objections to Platos theory of forms as a commitment to a heaven. But more to the point here is that even if good were a natural property, right would remain nonnatural if it is understood in terms of good. For it would involve reference to all the consequences of an action and all the organic wholes of which it is a part to all space and time, to the whole world. And this is not unsupported by scrupulous moral thought, which sets no time or place beyond which it cares not what happens. Some Americans do care about the floods in Bangladesh, and many people, wherever they may be, care about the climate on earth a century from now. Authentic environmentalists do not say that when humans become extinct, whales and prairie grass might as well. Many believe honesty would be owed to, and expected of, even extraterrestrials, angels, and gods, should they exist. But these totalities of consequences and organic wholes, indeed the world itself, would not be natural objects in Moores precise sense. They would not be natural because they would not be in time, even if they consist only of things that are in time. But they would not be natural also in a larger sense. Wittgenstein held that, although sentences about such totalities show something, indeed that they show the higher (das Hhere), they say nothing.

At a meeting of the Apostles in 1912 Wittgenstein heard Moores paper Is Conversion Possible? which Moore had first read to the Apostles in 1900 while working on Principia. That Moore read the paper again suggests he had not abandoned its ideas. Moral conversion, he said in the paper, is not unlike religious conversion, even though it is not necessarily connected with any religious ideas. It is both a great good in itself and it secures all other goods which depend on ones own mind alone.You see life steadily and whole and can feel neither desire nor fear of what you see to be bad in it.76
101

It is worth noting that in 1903, when Principia was published, Bertrand Russell wrote, Mans true freedom [lies] in the determination to worship only the God created by our own love of the good,77 and that in 1914, after two years of intense discussions with Wittgenstein, he attributed to the ethical work of Spinozathe very highest significance, as an indication of some new way of feeling towards life and the world.78 This new way of feeling, Russell added, lay outside the scope of the scientific method.

At about the same time, Wittgenstein wrote in his Notebooks: To believe in a God means to understand the question about the meaning of lifeto see that the facts of the world are not the end of the matterto see that life has a meaning,79 and later, in the Tractatus: The sense [Sinn] of the world must lie outside the world. In the worldno value exists.... (6.41). The sense (or meaning) of the world is not something in the world because it is the sense of the whole world. (6.4312).

Even if we refrain from calling the world mystical, we should acknowledge that it is mysterious. (It was one of the three spurious objects of knowledge Kant thought required treatment in the transcendental dialectic, the other two being the self and God.)The reason the world is mysterious is logical, not mawkish or cabalistic. It is not that the world is too big or too unlike what we take it to be. Not its size or content, but its logical/ontological category, or rather its failing to fall in any category, is what makes it mysterious. This is why genuine propositions about it, and thus ethical propositions, are impossible. We may say that the world is everything, but this would only acknowledge its peculiarity. For to speak of everything is to employ the predicate is a thing or is a fact, depending on whether we think the world is the totality of things or of facts. Both
102

predicates, Wittgenstein held, express only formal concepts, corresponding to formal or internal properties, and thus the sentences in which they occur say nothing, though they show much. I shall have much more to say about these views in chapter 4, where Wittgensteins position on ethics will be discussed in detail; in chapter 7, where the focus will be on the logic his views presuppose; and in Part Three, where his metaphysical views will be relevant. But reference to them here would be instructive.

The realism/antirealism debate, in metaphysics and logic as well as in ethics, must be radically rethought wherever it involves putative statements employing formal concepts about the totality of things or of facts, that is, about the world. If ethics involves such statements, as Moore and Wittgenstein believed, then according to Wittgenstein, though not Moore, both moral realism and moral antirealism must be rejected. For such statements say nothing, even though they show much. The controversy between moral realism and moral antirealism thus becomes a special case of the controversy between what in the Introduction I called metaphysical realism and metaphysical antirealism. Insofar as the latter concerns the world, neither alternative can be stated properly, if Wittgenstein was right. We do not face a stark choice between them.80 Antirealism is usually a negative position, merely denying the reality of whatever items are in question, and today usually asserting that with respect to them language is all there is. This is why it is deeply unsatisfactory, whether in metaphysics or in ethics. Wittgensteins distinction between saying and showing offered an alternative to antirealism that did not constitute a return to realism, i.e., to the equally unsatisfactory acceptance of the items in question as being unqualifiedly out there. I have called such an intermediate position semirealism.
103

Moores theory of right action involves reference to the world as a whole. For this reason and in this sense, he might have agreed with Wittgenstein that insofar as it includes that theory, ethics is transcendental, that it concerns the limits of the world, not its contents.81 Such a view of ethics was not entirely novel. For Plato the philosophic life culminated in a glimpse of the Form of the Good, which he held to be indescribable. Aquinas placed Good in the company of Being, One, Truth, and Beauty, the so-called transcendentalia, which were said to range across the categories, i.e., the highest genera, and thus to lack even the status of categories of things in the world, much less the status of things. In philosophical theology God has described as a being of infinite goodness that is the source and measure of all other goodness, earthly and unearthly.

As if using words from Wittgensteins Tractatus, Kant said that, unlike what he called practical anthropology, moral thought is concerned, not with what happens, but with what ought to happen, even if it never happens.82 His distinction between what happens and what ought to happen was in tune with, though of course not the same as, Wittgensteins distinction between saying and showing. If ethics is not about what happens, i.e., is the case, yet truth is correspondence to what is the case, then ethics contains no truths. Kant did not draw this conclusion, but Wittgenstein did. The logical positivists also drew it, used it to attack Moore, but misunderstood it.

Whether we ourselves should draw the conclusion depends on how circumspect we are in wielding the notion of truth. That truth is correspondence (bereinstimming, agreement, in Kants terminology) to fact is a truism of common sense. But as a philosophical theory it is too crude for ethics, as well as for logic and mathematics, for reasons to be explained in chapter 7, and indeed even for ordinary general statements, for
104

reasons to be explained in chapter 10. Wittgenstein saw this, perhaps Kant also did. Moore might have seen it with respect of statements predicating goodness overall, especially rightness, had he considered the matter. To be sure, all three accepted the truism. Moore discussed the nature of truth extensively but inconclusively in lectures delivered in 1910-11 but published only posthumously. He wrote that to say that a belief is true is to say that it corresponds to a fact,83 and that this means that the fact to which it refers is, or has being.84 But he seemed to vacillate, writing also that the notion of a fact can be understood as standing for what corresponds to true belief,85 in which case what he had said about truth would be circular, and acknowledging that the truth of conditional statements did not seem to fit it at all.86 Like Kant, Moore was confident about the definition of truth as correspondence only when understood as nominal, and, like Wittgenstein, he was willing to restrict the range of true statements to which it applied.

3. The Good and the Right. It is often asked whether a dehumanized ethics such as Moores could be relevant to action. The question is as ancient as Aristotles complaints about Platos Form of the Good. It is especially familiar from discussions of egoism. Are there circumstances in which one ought to sacrifice ones life? The usual opinions fail to distinguish between the goodness of remaining alive and the goodness of dying in a certain way, though it is familiar to most parents who would sacrifice their lives in order to save their childs. Its not just the goodness of the childs life that matters. There is also the goodness of ones dying in order to save it. Similarly, a soldier may
105

sacrifice himself in order to save his comrades, not just because of the goodness of their lives, but also because of the goodness of the sacrifice, which may well seem to outweigh that of remaining alive. But our question here is not about egoism, though still ambiguous. It may be asking (1) whether Moores property good can bear a relation to the rightness of actions, (2) whether one can be motivated to action by it, (3) whether one can be motivated by cognitive states of which it is an object, by itself or as a constituent of states of affairs, or (4) whether there can be such cognitive states in the first place.

The connection between good and right cannot be just happenstantial, but neither it trivially definitional. In Principia Moore did define duty as that action, which will cause more good to exist in the Universe than any possible alternative,87 but also described what is good in itself or has intrinsic value as what ought to exist for its own sake.88 In Ethics he repeated that it is always our duty to do what will have the best possible consequences, but denied that this is a mere tautology.89 In the preface to the second edition of Principia he wrote that he had used good in a sense that bears an extremely important relation to the conceptions of right and wrong.90

The idea of being motivated to action by Moores property good indeed has doubtful coherence, but only because the idea of being motivated by any property seems incoherent, for purely logical, not ethical or psychological, reasons. Dyadic relations do not hold between individuals and properties. xRF is ill-formed. But it is not incoherent to ask whether one can be motivated by cognitive states of which the property good, or a state of affairs that includes it as constituent, is the object. Could such cognitive states be
106

reason-giving? Could they be action-guiding? The question is difficult to answer because the specialists in such matters, namely, psychologists working on motivation, still know too little, probably because of the moral and social constraints on serious experimentation on human subjects, while philosophers engage only in speculation or intuition of what would be proper to say. Hume announced that reason is, and ought to be, the slave of the passions, but we may wonder how he knew this. Perhaps Hume could be excused for thinking that passions are discoverable through introspection. But his own views should have kept him from thinking that their being motives, i.e., their motivating, could be so discovered. Nevertheless, there have been advances in psychology relevant to motivation. The preoccupation with primary drives seems to have ended. There is growing recognition of the spontaneity of the young childs artistic and play behavior, the importance of surprise for its cognitive development, the childs preference for what is novel, the power of curiosity, the tendency to explore, a desire to know for the sake of knowing, which are evident even in nonhuman animals. Progress has been made away from the egoism and hedonism presupposed by most so-called rationality theories.91 The genetic basis of motives like empathy is readily acknowledged. Chomsky accepts Platos thesis in the Meno that much of our knowledge is innate.

Can there be cognitive states the object of which is the property good, or a state of affairs that includes that property as constituent? Philosophers offering a negative answer usually rely on causal or quasi-causal theories of knowledge, which are metaphysical and epistemological, accepted either because they seem scientific or because of thoughtexperiments about what we would or would not say in imaginary situations. They ask,
107

Would you say that S knows (perceives, sees, is aware of) x if x bears no relation to S? But they ignore the obvious answer, No, I would not, but x does bear a relation to S, namely, that of being known by S, in favor of unobvious answers employing the notion of causality, as if the right answer could not be also the obvious one. Such philosophers must be assuming that the nature of causality is clearer than the nature of knowledge, or at least that they understand the former better than the latter. Some even deny the cognitive status of arithmetic, a paradigm of unquestionable knowledge, if it is taken to be about numbers, a certain kind of abstract entities. Presumably, they would say that, if arithmetic were about numbers, a mathematicians intellectual life would be as inexplicable as they would find the mathematicians moral life to be if it involved awareness of Moores nonnatural property good.

Elsewhere I have suggested that the property good, described by Plato as blinding, by Aquinas as one of the transcendentalia, together with Being and One, by Moore as nonnatural, and by Wittgenstein as unsayable, is best considered a generic property, though one on the highest level of generality.92 It is the genus of which Moores personal affection and aesthetic appreciation, Aristotles eudaimonia, Mills pleasure, and other goods reasonably proposed by reasonable people are species. This is why we cannot see it in the way we see a shade of yellow. But then neither can we see even Color in that way, though Color is a generic property on an incomparably lower level of generality. We might say (though Wittgenstein did not) that Goodness only shows itself in its species, just as Color only shows itself in yellow, red, blue, etc., and is not seen as a property separate from its species. This would be so even if Goodness were one of what the medievals called the transcendentalia, which ranged across the highest genera, the
108

categories, and therefore were not genera, nor definable per genus et differentiam. The medieval distinction is important but not relevant to the present point.

Whether or not this suggestion is right, the theorists denying that we can have cognitive access to Moores property good need to pay more attention to the epistemological and metaphysical details of the issue. Their own epistemology is open, of course, to the familiar objection to externalism, namely, that it cannot answer the skeptic, though finding such an answer has been the raison dtre of epistemology at least since Descartes. But our theorists also ignore, or require us to count as unintelligible, Kantian and contemporary antirealist accounts of cognition, which we ought to be able to understand and respect even if we do not accept them. According to Kant, although insofar as it is a fact in the world cognition is a zoological matter and thus subject to the demands of scientific causal explanation, the world itself is comprehensible only as an object of cognition insofar as it is not a zoological matter.

The theorists also owe us answers to numerous metaphysical questions, such as those explored in detail by David Armstrong93 and Evan Fales.94 The first, of course, is What is causation? They cannot just revert to Hume and in effect beg the question against Moore by saying that causation is constant conjunction in time, whether accidental or nomological. Is it then, as Armstrong and Fales have argued, a relation between universals, properties, or at least grounded in such a relation? And are properties universals in the first place, or are they rather particulars, tropes, or perhaps both, as Moore in fact held?95 How would the arguments against Moore read in the case of each possibility? Fales argues that there must be properties we can identify independently of
109

their causal powers if a vicious infinite regress is to be avoidable. His examples are the properties characterizing the content of sense perception, though his ultimate concern is with the identification of properties in physics.96 But if some properties can be identified without reference to their causal powers, so might Moores property good, whether or not it has causal powers. If our theorists deny this on the ground that it is a nonnatural property, then they must tell us in detail how they understand this. Is a property nonnatural because it is not in time, as Moore held? What is it for a property to be in time? Would they say, circularly, that to be in time is to have causal powers? Moreover, since a property, natural or nonnatural, enters in causal relations only indirectly, as a constituent of states of affairs, we would need to be told a great deal about the nature of states of affairs and how and what properties might be constituents of them. For example, can a state of affairs that is in time and part of the causal order have a constituent that is not? To be sure, such a constituent would not itself be in time and enter in causal relations, but then is this not true of all properties?

Yet another question the theorists need to consider is whether there are uninstantiated properties. If there are, must they, too, have causal powers? Are moral properties, even though real, uninstantiated? If they are, their lack of causal powers may be due just to their lack of instances. But even this is not obvious. Being an angel perhaps has no causal powers, but is this true of the property of satisfying the definition of the circle? Michelin and Goodyear must have excellent knowledge of that property even though surely it is never instantiated. Is Moores property good like that of being an angel or like that of being a circle? Or is it rather a trope, like the particular shape of my pen, and if so is it also a generic property, like shape as such? Perhaps generic properties have no causal
110

powers but their species do. A tires having shape helps no car roll but its being round does. A traffic lights being colored stops no driver but its being red sometimes does. Are Moores ideal goods personal affection and aesthetic appreciation species of goodness, as I have suggested? Do they have causal powers? If they do but their genus, goodness, does not, is this so because the latter is a nonnatural property or just because it is a generic one, like shape or color as such? And, finally, are Moores ideal goods instantiated properties? Perhaps they are not, perhaps they are mere ideals. Can mere ideals moral, political, religious have causal powers? If they cannot, how are we to understand our striving and sometimes dying for them?

Needless to say, I shall not attempt to answer these questions here, but answers are needed, detailed and carefully worked out, if we are to take seriously the complaint that Moores property good is irrelevant to action. Appealing to naturalism or the scientific point of view, much less to our intuitions, is not enough. Without such answers the complaint might be like the 17th century natural philosophers complaint that Newton appealed to occult and immaterial gravitational forces, rather than to intelligible and robust bumping, or like H. A. Prichards complaint that Einsteins theory of relativity was unintelligible because it relied on a non-Euclidean geometry. How a body could motivate another body at a distance was incomprehensible to those 17th century natural philosophers, and Prichard could form no mental image of a non-Euclidean space. Perhaps those of Moores heirs who have been unable to comprehend his nonnatural property good were also just unable to picture it as pushing or pulling.

111

Therefore, I shall ignore the metaphysical and epistemological concerns behind the questions about the relevance of Moores ethics. But a different question does arise. It is both legitimate and deep. By requiring reference to the whole world in judgments of duty, Moore could tell us nothing specific about how we ought to act in any particular situation. This is why he virtually admitted that his theory of right was profoundly skeptical. There might be an action we ought to do, but as a matter of empirical fact, not philosophical theory, we could not know what it is. In view of the mind-boggling range of its consequences and organic relationships, throughout an infinite future,97 even probability statements about them could not be seriously made. Radical moral skepticism seems inevitable.

But Moores moral skepticism does not lead to amoralism. Ideal utilitarianism is not mere consequentialism. An action may be intrinsically good even if it ought not to be done, even if it did not make the world better. In Rosss terminology, if not meaning, it may be a prima facie duty even if not an actual duty. This is why respect for the good, Aquinass first principle of natural law, may continue to inform the ideal utilitarians actions. Such respect would be akin to love, whether practical or pathological, not to calculation. It can have as its object only the intrinsic goodness of the action, its being a prima facie duty, not its being an actual duty. Only a part of a world, not a whole world, can be loved.

I am not suggesting an inference, surely specious, from the intrinsic goodness of an action to its rightness. No claim is being made that the former makes the latter probable, even to a tiny degree, or that it justifies or is a reason for the action. In Hutchesons useful terminology, if not meaning, it is at most an exciting reason, not a justifying reason. No appeal is being made to it a what Kant called Grund and Moore
112

called evidence. Rather, my suggestion is that if in acting one is motivated and guided only by respect for the good, but only the intrinsic goodness of an action is intellectually visible, then one is motivated and guided only by respect for the intrinsic goodness of the action. One has no knowledge of the totality of its consequences and organic relationships, indeed one has even no genuine conception of it. Thus, qua agent, the ideal utilitarian can only be a deontologist, not a consequentialist. We have to settle for rightminding, even if it does not coincide with right-acting. This is why Moores ideal utilitarianism was not inimical to moral common sense, which views with distaste the spirit of calculation, cost-benefit analysis, that ordinary consequentialism cultivates. The ideal utilitarian has no justifying reasons but plenty of exciting reasons for doing good particular actions without guile: their plain goodness. Thus Moores dehumanized ethics may be seen as the marriage of love, not convenience of the two great ways of moral thinking: the utilitarian and the deontological.

Chapter Four: 1SAYING AND SHOWING THE GOOD

1. Logic and the World. In Tractatus Logico-Philosophicus Wittgenstein made the case for dehumanized ethics much more briefly but also more powerfully than Moore had done fifteen years earlier. He did so with his distinction between what can be said and what can only be shown and combining it with his picture theory of meaning and thought. Contrary to received opinion, neither the distinction nor the theory is idiosyncratic or obscure. As we shall
113

see, the distinction has a noncontroversial application to ordinary pictures and representations generally. And, of course, the traditional view of meaning and thought was representational: both involve ideas, even mental pictures (images). Wittgensteins picture theory of meaning and thought was merely a subtler version of that theory. The distinction between saying and showing was essential not only to his views in ethics. It was essential also to his logical views on generality, to be discussed in chapter 7, as well as to his metaphysical views regarding the self and the world, to be discussed in chapters 11 and 12.

The distinction provided a welcome alternative to the stark choice between realism and antirealism, in ethics, metaphysics, logic, even the philosophy of religion. Wittgenstein thought it concerned the cardinal problem of philosophy. Many of his readers find it obscurantist. Tough-minded philosophers ask, What are those things that can only be shown? How can there be such things? But their question misses the point of the distinction, namely, to provide an alternative to both realism and antirealism. What only shows itself is not part of reality, like Secretariat or oxygen. But neither is it unreal, like like Pegasus or phlogiston. This is why in the Introduction I called it semirealism, distinguishing it thus from both standard realism and standard antirealism.

Wittgenstein applied his semirealism chiefly to what he called logical objects.98 The implications for ontology were obvious and direct. But his semirealism also had important implications for ethics for what might be called ethical objects. Shortly after completing the Tractatus, he wrote that the point, the meaning, of the book was an ethical one.99 In this chapter I shall be mainly concerned with that ethical point. But I
114

shall devote considerable attention to the logical/ontological views in the Tractatus, which serve as the grounds of the ethical views but are developed in much greater detail, partly because they will be indispensable in chapter 7 for our account of Wittgensteins antirealism in logic.

Sophisticated but sensible moral realism holds that an action ought not to be done if its consequences would be very bad. It also holds that the action might nonetheless be intrinsically good (or prima facie right, in W. D. Rosss terminology), or even good merely by enhancing the goodness of what G. E. Moore called an organic unity or whole. Doing justice would be bad (wrong) if the heavens should fall, though presumably it is intrinsically good and also enhances the goodness, or lessens the badness, of any organic wholes of which it is a part. To say that justice is to be done even if the heavens should fall is moral posturing, not moral thinking. Frederick the Greats committing suicide if captured by the enemy in order to protect his country from extortion might be good (right), a possibility even Kant envisaged though he believed that suicide is bad (wrong) in itself. The presence in a philosophy department of its only aesthetician may enrich it, even if the person is only a mediocre aesthetician.

Nevertheless, as we saw in the previous chapter, such moral realism remains beholden to the future. It does enjoin us to take into account all the consequences of an action and all the organic wholes of which it is a part, since genuinely moral thought sets no date and no place beyond which what happens doesnt matter. Todays small children will retire more than half-a-century from now, but if we ought now to help make their retirement happy. The disastrous flood in Bangladesh occurs far from Iowa, but Iowans who can
115

ought to provide aid. Thus our moral realism inherits the major defects of ordinary consequentialism. The first is epistemological: we cannot know what we ought to do because we cannot know all the consequences and noncausal ramifications of our actions; we cannot even make serious probability judgments about such an indefinite, possibly infinite, totality. The problem is familiar, having been discussed at length by Sidgwick, and arises out of commonsense considerations, not philosophical theories. It renders literal cognitivism questionable. I have discussed it in detail elsewhere.100 The second major defect of ordinary consequentialism is metaphysical and not at all familiar: with respect to such a totality, realism, not just literal cognitivism, is questionable. Wittgenstein was the first to see this. His view rested on fundamental logical considerations, which here I shall only sketch. They will receive detailed discussion later, especially in chapters 6 and 7.

In the Tractatus, Wittgenstein wrote, [I]t is impossible for there to be propositions of ethics (6.42). But what he meant was not at all what his positivist successors thought indeed, it was just the opposite. He went on to explain: Propositions can express nothing that is higher.Ethics cannot be put into words. Ethics is transcendental (6.426.421). Yet, so was also logic: Logic is not a body of doctrine, but a mirror-image of the world. Logic is transcendental (6.13). Wittgenstein took this to follow from the very nature of logic: The propositions of logic describe the scaffolding of the world, or rather they represent it. They have no subject-matter (6.124). The reason is that Propositions cannot represent [kann nicht darstellen] logical form: it is mirrored [spiegelt sich] in them. Propositions show [der Satz zeigt] the logical form of reality. They display it [Er weist sie auf] (4.121). Indeed, What can be shown [gezeigt werden
116

kann], cannot be said [kann nicht gesagt werden] (4.1212). Nevertheless, it can be shown. Wittgenstein was not a realist, whether in logic or in ethics, but neither was he an antirealist. He was a semirealist.

The propositions of logic can only show the logical form of reality because the characteristically logical expressions are formal: [T]he variable name x is the proper sign for the pseudo-concept object. Wherever the word object (thing, etc.) is correctly used, it is expressed in conceptual notation by a variable name (4.1272). This is why it is nonsensical to speak of the total number of objects. The same applies to the words complex, fact, function, number, etc. They all signify formal concepts (4.1272). The statement This is white does say something. What it says can even be pictured literally, not just logically. But the putative statement This is an (individual) object does not, for it presupposes what it purports to say, its having sense depends on its being true.101 Yet the statement is not gibberish. A logical category might not seem to be something higher in the way the concerns of ethics and religion do, but logically it is the highest, most general, classification. In fact, the reason Wittgenstein applied the phrase the higher (das Hhere) to the concerns of ethics and religion was that they are about the limits of the world and thus exceed the limits of what is sayable for logical, nor ethical or religious, reasons.

Wittgenstein did not complete the list of formal concepts in 4.1272. I shall return repeatedly to their important role in his philosophy. For our present purposes, however, what matters is that the distinction between saying and showing applies to both object and fact. This entails its application also to world. It is nonsensical to speak of all
117

facts or of all objects. For we cannot speak of totalities determined by properties that can only be shown, i.e., formal properties. According to Wittgenstein, the world is the totality of facts, which on final analysis are configurations of objects in states of affairs, it follows that we cannot speak of the world. It is important to note that this would also follow if the world were the totality of objects. It does not presuppose the proposition that the world is the totality of facts.

2. The World and the Good. The proposition that we cannot speak about the world is essential to Wittgenstein s ethics. It is a consequence of the thesis that we cannot attribute formal properties. Whether the world is the totality of facts, as Wittgenstein held, or of objects, as one is more likely to think, we cannot genuinely speak about it. The totality of facts is the totality determined by the one-place predicate is a fact, and the totality of objects is the totality determined by the one-place predicate is an object. But being a fact and being an object are formal properties, which can only be shown. Thus the world cannot be said but at most shown. If the ethical involves reference to the world, it too cannot be said but at most shown.

And, indeed, we find Wittgenstein writing: The sense [Sinn] of the world must lie outside the world. In the world everything is as it is, and everything happens as it does happen: in it no value exists and if it did exist, it would have no value. If there is any value that does have value, it must lie outside the whole sphere of what happens and is the case. For all that happens and is the case is accidental (6.41). In the Notebooks, he
118

had written: To believe in a God means to understand the question about the meaning of lifeto see that the facts of the world are not the end of the matterto see that life has a meaning.102 The sense of the world is not something in the world, because it is the sense of the whole world. This is why it may be said to lie outside the world, at its limits. But why is value also not in the world? Because, we are told, The world and life are one (5.621). The sense of the world is also the sense, meaning, of life. So, the latter, too, is not in the world. Yet the sense or meaning of life is the ultimate value. It constitutes [t]he solution of the riddle of life, but that solution lies outside space and time (6.4312). It is not an item in the world. Later, in 1929, Wittgenstein explained: What is good is also divine. Queer as it sounds, that sums up my ethics. Only something supernatural can express the Supernatural.103 The supernatural was as central to Wittgensteins ethics as the the supersensible was to Kants. Elsewhere, also in 1929, he wrote, [Attributions of] absolute value are nonsensical but their nonsensicality [is] their very essence [A]ll I wanted to do with [those attributions] was to go beyond the world and that is to say beyond significant language.104 Wittgenstein meant, however, not that the attributions are gibberish, but that they are not logical pictures, in the sense required by his theory of meaning. I shall return to this point in section 3

What Wittgenstein called the riddle of life presumably concerns the sense or meaning of life, which has been the central topic in serious ethics. To ask about ultimate value is to ask about the meaning of life, about what makes life worth living. (Platos Republic began with Cephaluss claiming that it is not pleasure, as commonly thought, but justice.) And the meaning of life does involve the meaning of the world. One who asks about the
119

meaning of ones life does sometimes phrase the question as asking about the sense of it all. Life can hardly be fully meaningful in a meaningless world. Indeed, that the world exists at all, that there is something rather than nothing, may be the ultimate object of joy or sorrow and certainly of wonder.

Wittgensteins identification of life with the world is striking, and so is his almost Spinozistic conclusion that to ask about value and thus seek the sense of the world require view[ing] the world sub specie aeterni as well as feeling the world as a limited whole, which, he adds, is something mystical (6.45). Ethics does ask what makes life good, worth living, meaningful, but the good life is the world seen sub specie aeternitatis105 This is why Wittgenstein also says: If the good or bad exercise of the will does alter the world, it can alter only the limits of the world, not the facts not what can be expressed by means of language. In short the effect must be that it becomes an altogether different world. It must, so to speak, wax and wane as a whole (6.43).

Wittgensteins point is not that an action, an exercise of the will, does not alter the world, which of course it does. Nor is it that actions are ethically irrelevant. His point is that the value, the goodness or rightness, of an action does not consist in its producing some particular event or events in the world, as standard consequentialism holds; rather, it consists in the world itself becoming different as a whole, at its limits. So our question about the consequences of an action must be unimportant. At least those consequences should not be events (6.422). It is in this sense that How things are in the world is a matter of complete indifference for what is higher (6.432). Realization of value, whether goodness or rightness, does not consist in the occurrence in the world of
120

some particular event or events, but in the world itself becoming different, at its limits, in its waxing and waning as a whole (6.43). However, all this can only be shown. For there cannot be ethical propositions. The reason is not that, as Wittgensteins early followers thought, there is nothing for such propositions to be about, but that what they purport to say cannot be said because it is the higher, which can only be shown (6.42).

Wittgensteins assertion that value is not anything that happens because it is not accidental reminds us of Kants third fundamental proposition of morality in the Foundations of the Metaphysics of Morals, that duty is the necessity of acting from respect for the law. Duty and necessity express a sense of must, not of accidental. Hiss assertion that value is not something that happens brings to mind Kants assertion that morality is about what ought to happen, not what does happen, the latter being the business of practical anthropology. What happens is already in the world, what ought to happen is not yet or perhaps ever.

Wittgensteins remarks about ethics left most of his readers bewildered, and his claim that there is something mystical, which cannot be expressed, was unpalatable to his tough-minded heirs in analytic philosophy. But that there is what he called the mystical is a consequence of his fundamental views about logic, and these cannot be easily rejected by tough-minded philosophers. According to Wittgenstein, It is not how things are in the world that is mystical, but that it exists (6.44). The existence of the (actual) world, indeed of something, is a basic presupposition of logic, but how to express existence (with a predicate or just the existential quantifier?) has remained deeply controversial since Anselm proposed his ontological argument for the existence of God,
121

for example in Kants denial that existence if a real predicate, Meinongs theory of objects and Russells vehement attack on it, and present-day quandaries about what it is for one world to exist or be actual, and the rest just possible. Children find the difference between what exists and what does not, say, between their real and imaginary playmates, simple and easy to comprehend, but logicians do not.

Wittgensteins ethical views were not the ravings of a troubled young man. They were a consequence of his tying the notions of ethics to the notion of the world. As we saw in the previous chapter, this had been done also by G. E. Moore, an essential part of whose ethics were the notions of reference to worlds considered in isolation and of the totality of the consequences of an action. Wittgenstein did not owe his views about ethics to Moore. Straightforward cases of one major philosopher owing his views to another are, not surprisingly, rare. Wittgenstein had read and admired Schopenhauer before he met Moore. But Principia Ethica had been published only nine years before Wittgenstein met Moore. It would be surprising if a book by one of his few friends, which was the rage in British philosophy at the time and he had read, did not influence his thought at all.

Contemporary antirealism in ethics usually begins by rejecting Moores view that goodness is a nonnatural property. But even if goodness were a natural property, rightness might still be nonnatural. As understood by Moore and most consequentialists, it involves reference to all the consequences of an action and all the organic wholes to which it belongs. If Wittgenstein is right, such reference would be impossible, for the reasons explained earlier. There could be no genuine statements about such a totality,
122

just as there could not be about the totalities of objects and of facts. Thus, in a different though no less important sense of this much abused word, those totalities could be said to be nonnatural. Yet rightness, natural or nonnatural, is the raison dtre of ethics as a discipline. It is what moral appraisal of action is about.

3. Understanding the distinction between saying and showing. Wittgensteins distinction between saying and showing might be less puzzling if stated in English as the distinction, familiar to kindergarteners as well as fiction writers, between telling and showing show and tell. Kindergarteners and fiction writers might or might not agree that some things can only be shown, but they would be unanimous that showing is often more effective than telling and sometimes alone practicable. Of course, they dont mean what Wittgenstein meant. But his picture theory of meaning suggests a reasonably clear account of what it is to be something that can be said, namely, to be capable of being pictured, whether literally, as in a painting or photograph, or in a logical picture, and of what it is to be something that cannot be said yet can be shown, namely, to be incapable of being pictured yet can be shown in the picture.

The picture theory of meaning was offered in the Tractatus first as an account of thought in general: A logical picture of facts is a thought [Gedanke] (3), A state of affairs is thinkable: what this means is that we can picture it to ourselves (3.001). Only then did he also offer it as an account of sentential meaning. The theory was not idiosyncratic. It accorded with the received, traditional doctrine that thinking consists in operating with ideas, mental representations, perhaps even mental images. It also implies, as that
123

doctrine did, the traditional doctrine that the truth of judgments is correspondence to reality, to be established, if possible, by comparison. But Wittgenstein applied these traditional doctrines mainly to language. He did so by proposing an unusual but not implausible conception of declarative sentences as logical pictures, which made the distinction between thinking and speaking seem insignificant: A proposition [Satz] is a picture of reality (4.01), Reality is compared with propositions (4.05), A proposition can be true or false only in virtue of being a picture of reality (4.06). This was part of the linguistic turn he took.

To think of something, traditional philosophy of mind held, is to represent it in the mind. Notoriously, this was most plausible and least unclear in the case of the representations called mental images, which could also be called mental pictures. To speak of something, Wittgenstein now held, is also to represent it in a picture, but the picture need not be a painting or a photograph, nor even a mental image. It could be just a logical picture a propositional sign, a sentence. Wittgenstein in effect strikingly broadened the traditional conception of a picture. He did so with a few succinct sentences: There must be something identical in a picture and what it depicts, to enable the one to be a picture of the other at all (2.161), What a picture must have in common with reality, in order to be able to depict it correctly or incorrectly in the way that it does, is its pictorial form (2.17), A picture can depict any reality whose form it has. A spatial picture can depict anything spatial, a coloured one anything coloured, etc. (2.171), A picture whose pictorial form is logical form is called a logical picture (2.181), Every picture is at the same time a logical one (2.182).

124

A sentence (propositional sign) is a logical picture because it depicts what it says by sharing with it only logical form, rather than, say, also shape or color, as picturing it in a paintings, photograph, or perhaps mental image would do. But Wittgensteins view of a sentence was not the grammarians, it was also strikingly broad: The essence of a propositional sign is very clearly seen if we imagine one composed of spatial objects (such as tables, chairs, and books) instead of written signs (3.1431). It would be very unwieldy but possible to use tables, chairs, and books instead of words to refer to objects, and to use configurations of tables, chairs, and books instead of sentences to make statements. A sentence could also be composed of mental objects. In a letter to Russell, Wittgenstein wrote that a thought consists of psychic constituents that have the same sort of relation to reality as words, though he added: What those constituents are I dont know106 This is why Wittgensteins often used thought [Gedanke] and proposition [Satz] interchangeably. The linguistic turn implied by his assertion The limits of my language mean the limits of my world must not be confused with the crude linguistic turn that became fashionable later, even though it was initiated by the Tractatus. If our cognitive access to reality consists in representing it, but the representations need not be more than logical, then whether the access is psychological or linguistic is irrelevant. This is the defensible linguistic turn the one Wittgenstein took. Descartes, Locke, and Kant accepted the first part of the antecedent of this conditional, but the second part did not even occur to them. Had it done so, they might have accepted both and the history of modern philosophy would have been dramatically different.

Now, to be something that can be said is to be capable of being pictured, whether literally, as in a painting or photograph, or in a logical picture. To be something that
125

cannot be said is to be incapable of being pictured. But what cannot be said can be shown. There is no mystery here. In a painting, for example, much is shown that is not and cannot be pictured in the painting or by any part of it. The painting may represent a tree next to a barn, each represented by a part of the painting, and the spatial relation between the parts of the painting that represent the tree and the barn may represent their relation of being next to each other. But nothing in the painting represents that relations being a relation, nothing says that their being next to each other is a relation (rather than, say, a shape or color). Yet the painting shows this, indeed must show it if it is to represent what it does represent. What it shows cannot be denied as one might deny, for example, that the painting is a portrait of Churchill. The absence from the painting of what it only shows is not like Churchills absence from it. Of course, paintings do not consist of words, and sentences are only logical pictures. But like all pictures, physical or mental, paintings are logical pictures, though not all logical pictures are paintings.

Wittgensteins distinction between saying and showing may initially seem mysterious but it is familiar even to beginning students of logic in one of its applications. That a statement is logically true, a tautology, that its negation is a contradiction, often shows itself and is immediately seen in its logical form, without reference to what the statement is about. But that a statement is true in some other way, say, empirically, does not show itself, its truth cannot be just seen in its form we must also know what it is about, what it says, by attending to what the descriptive expressions in it stand for. In neither case do we take the logical form itself to be about or stand for anything.

126

Although in accord with the traditional views of thought as representation and of truth as correspondence, Wittgensteins picture theory of meaning and his distinction between saying and showing were thus a major, far-reaching revision of those views. He accepted them only on the level of atomic sentences, where indeed alone they are plausible. But this is an extraordinarily primitive level. Sentences on the highest levels, those of logic and ethics, allow only showing, not saying. Even ordinary molecular and general sentences, where saying is possible, fail to be pictures in Wittgensteins, or indeed any view, even if their atomic components do not. A sentence of the form If p then q is not itself a picture, even if p and q are. But this is not true only on Wittgensteins theory. It is also true on the traditional theory: no physical or mental representation can be made of what the sentence says. This is why such sentences have been found problematic in the history of philosophy. Not surprisingly, in his later work Wittgenstein abandoned the picture theory of meaning and the correspondence theory of truth. Even if they are not totally defective, their applicability was unacceptably limited.

We can now understand the Vienna positivists passionate opposition to what can only be shown, in particular to the ethical. No picture of any kind physical, mental, or merely logical can be made of what can only be shown: this is why it cannot be said. For example, no painting can literally depict the goodness of a person or the rightness of an action. It follows that what can only be shown is not observable, since presumably anything observable can, at least in principle, be pictured, physically or mentally, and therefore logically. This, of course, is exactly why the traditional empiricists denied that there are unobservable entities we could not have ideas of them, since ideas are copies of impressions. It is also why the logical positivists (the logical empiricists)
127

denied that there are things that only show themselves, and in particular that there are ethical objects. Yet their most familiar claim about ethics did appear to coincide with one of Wittgensteins. If propositions are pictures, then there can be no propositions of ethics. Wittgenstein avoided unbridled realism sufficiently to inspire logical positivism, but he also avoided unbridled antirealism sufficiently to protect the ethical.

Perhaps the motivation behind the two most common accounts of Wittgensteins distinction is also plain, though often tacit, empiricism. The first holds that what only shows itself is not anything at all. Prima facie, there is little to be said in favor of this account. Surely, what only shows itself, in logic as well as in ethics, is not like Pegasus or phlogiston. Its claims to some status in reality would remain even if they could not be fully met.

The second, more common, account holds that what only shows itself cannot be said for reasons of surface grammar. It resembles the common interpretation of Freges claim that the concept horse is not a concept: is a horse is a grammatical predicate, not subject. Warren Goldfarb writes, All we are doing [in speaking of logical form] is noting that names have to be put together in one way or another in order to make sentences.107 Of course, we are noting this, but its not all we are doing. Why do we put the names together, why do we need sentences at all, rather than just names? Even if, for superficial grammatical reasons, some things cannot be said with a sentence, why could they not be said with a list of names, as of course Wittgenstein would also have held? It might be an awkward way of saying something, but why is it not a genuine way? Lists of
128

names sometimes do seem to say, tell us, something, e.g., in inventories. Wittgensteins answer was that we need sentences, not just names, because the world is the totality of facts, not of things. It was a metaphysical, not grammatical, answer. A world in which Jack is the father of Jim and a world in which Jim is the father of Jack may contain the same things and thus the same inventory, but they are different worlds. Moreover, why do only some ways of putting words together say anything, i.e., count as well-formed sentences? Wittgensteins answer was that the formal properties of what names stand for, i. e, objects, allow only some configurations of objects into states of affairs and thus only some logical pictures. Perhaps there are better answers to these questions, but answers are needed and few have been given.

The etymology of the word reality deserves attention by both realists and antirealists. Res is the Latin for thing, and no argument is needed that logical objects such as negation and ethical objects such as goodness are not things like Secretariat or oxygen. But also no argument should be needed that they are not things like Pegasus or phlogiston.

Wittgenstein used show (zeigen), of course, as a metaphor. He resorted occasionally to other terms, as in 4.121: cannot represent [kann nicht darstellen], is mirrored [spiegelt sich] finds its reflection [sich spiegelt], displays it [er weist sie auf]. And say (sagen), even if not used metaphorically, is sufficiently vague in everyday language (German or English) to allow for a variety of uses, as does its synonym tell. A sentence says or tells us something, but so does a list of names, a thermometer, and a
129

human face. Rudolf Carnap chided Wittgenstein for writing a whole book and then concluding that what he had said in it could not be said. But a simple way of avoiding the apparent inconsistency would be to introduce, in place of say, the two terms say1 and say2, explaining that what is said1 is what according to Wittgenstein is said, and what is said2 is what according to him is only shown. For example, the sentence This page is white says1 that this page is white but says2 that it is an object. Then Wittgensteins otherwise puzzling conclusion in the Tractatus would be just that what was said2 in the book could not have been said1. The difference would be explained sufficiently as the difference between what can be pictured and what cannot. There can be a picture of what This page is white says1, perhaps a picture (painting, photo) of a white piece of paper, but there can be no picture of what it says2, say, that the page is an object. Similarly, while the sentence This page is blue would be a case of saying1, the phrase it is not the case that, which might have preceded it, would be a case of saying2. A picture of a white piece of paper does not represent by itself what It is not the case that this page is blue expresses. If it did, it would represent what all sentences of the form It is not the case that this page is , where is any color other than white, would express, .

Wittgensteins distinction between what is said and what is only shown can thus be thought of as serving to emphasize, in a striking way, certain fundamental differences between two kinds of items. What is said can be an independent item in the world, what is only shown cannot. That there is such an item as negation going about in the actual objective world is plainly absurd: there can be no painting or mental image of it. That there are such items as goodness and rightness going about is no less absurd, and so is the idea of picturing them. Yet, nothing would be a cognizable world without negation.
130

And, we may add, nothing would be a lived-in world without good and evil, right and wrong .108 For nothing would be life without action and nothing would be action if not subject to appraisal, whether directly or indirectly, as a good or evil, right or wrong, action.

Although the point of the Tractatus was ethical, according to Wittgenstein, he was mainly concerned, as the title makes explicit, with logic. No explanation is needed, therefore, of the brevity of his remarks about ethics. Even in metaphysics, his focus was on its most abstract level, that of logical form. Not only would nothing be a perceived, cognized, or lived-in world without logical form, nothing would be a world at all without it. For nothing that violated or could not be captured by logic could be a world. Even a world consisting solely of immaterial or nonspatiotemporal objects, e.g., angels, would still need to conform to logic. Had Wittgenstein supposed that his simple objects are angels, his logical and metaphysical views would have remained unchanged. Unlike the logical antirealist, he did not deny that there is logical form in the world. Rather, he drew attention to the radical difference between logical form and the things that exhibit it by pointing out that, unlike them, it cannot be pictured, though it can be shown: There are, indeed, things that cannot be put into words. They make themselves manifest. They are what is mystical (6.522).

A basic presupposition of logical analysis is that sentences unlike in surface grammar may share the same logical form. In analysis we begin with the surface grammar of the sentence but search for what it must have in common with other sentences, especially in other languages, if they are to have the same truth-value and the same implications. That
131

is the logical form of the sentence. But unlike its surface grammatical form, its logical form is invisible. Moreover, if we take for granted that truth is correspondence to fact, then we must take for granted also that there is something in the world to which the logical form of a true sentence corresponds, viz. the logical form of the fact it asserts. Otherwise, there would not be a fit sufficiently specific and definite for truth. Yet the logical form of the fact is quite unlike the things the sentence is about, viz. the constituents of the fact. It is invisible, like the logical form of the sentence, even if they are visible. But it is also quite unlike invisible things. The subject-predicate form would be present also in the putative facts asserted by theological sentences such as God is wise and mathematical sentences such as 3 is an even number, and while God and numbers are invisible, there is an obvious difference between why they are invisible and why logical form is invisible. The page and color that the sentence This page is white is about are visible, but its subject-predicate logical form is not. In the case of God is wise, both the logical form of the sentence and what it is about, God and wisdom, are invisible.

I have called Wittgensteins view logical semirealism in order to distinguish it from both logical realism, which cheerfully allows for such statements as This piece of paper is a particular, and logical antirealism, which no less cheerfully dismisses them as gibberish. It is a sophisticated view. It should not be confused with mere denials of reality. Its aim is not to fight superstitions or fairy tales, as antirealists sometimes think is their aim. Denying that some things can be said is not like denying that there is such a horse as Pegasus or such a substance as phlogiston. What is denied is that certain items, which are not at all matters of superstition, are like Secretariat or oxygen. This is not to assert that
132

they are like Pegasus or phlogiston. It is to assert that they are different from all four.

Wittgensteins position in ethics may be called moral semirealism for similar reasons: to distinguish it from both moral realism and moral antirealism. The difference between goodness or rightness and oxygen or yellow color that moral antirealism insists on, even if grossly misunderstands, has been amply demonstrated in its long history. It is felt by anyone who asks the familiar and seductive but logically puzzling question, Why should I do what I ought to do? Few agree with moral realism that goodness or rightness is at all like oxygen or yellow. Few agree with Moore. Yet, that the good and the right are like Pegasus and phlogiston is also seldom taken seriously, except by enfants terrible and philosophes subtile. Few are unable to understand the difference between what ought to be the case and what is the case, or think of it as resembling the difference between phlogiston and oxygen.

Moral antirealists are seldom sensitive to the distinctions that led Wittgenstein to logical and then to moral semirealism. In this respect they differ radically from the philosophical roots they share with Wittgenstein in Kants transcendental idealism. Kant proposed a unified account of the empirically real and the transcendentally ideal, and he vigorously defended the need for acknowledging both. Nothing comparable in scope or depth can be found in standard moral antirealism. What might be the reason? Perhaps it is the tacit prejudice that the distinction between reality and nonreality does not admit of refinement. Meinong called it prejudice in favor of the actual, by which he meant objects in time.
133

Early in the twentieth century, philosophers on both sides of the English channel, including Moore, Russell, Meinong and Husserl, routinely distinguished between being and existence, meaning by the latter what Meinong meant by actuality, and unhesitatingly asserting, for example, that relations and numbers have being, but denying that they exist, that they are in time. What exists, they held, is in time. What has being but is not in time does not exist. Indeed, usually we speak of the existence of spatiotemporal items, not of items such as relations or numbers. But, unless we are philosophers, we do not say that relations or numbers do not exist, e.g., that there is no such relation as fatherhood or such a number as 5. We usually just ignore the issue and say nothing. Wittgenstein wrote the Tractatus when the distinction between being and existence was familiar, and undoubtedly he was sensitive to the reasons for making it. It is not the same as his distinction between what can be said and what can only be shown, but resembles it in motivation. Contemporary antirealists ignore both distinctions, and assume that being, existence, reality, and actuality are all the same, and expressed by the particular (existential) quantifier. Wittgenstein in effect showed that this assumption is not so much wrong as primitive.

He was particularly sensitive to its failure to fit the special status of the ethical. As Kant pointed out, ethics is concerned not with what does happen, but with what ought to happen, even if the latter has never happened nor will happen. Wittgenstein put essentially the same point by saying bluntly that the ethical is not in the world at all. Therefore, it also is not the sort of thing that can be observed and therefore pictured in
134

ordinary physical or mental pictures. But, going beyond anything that Kant or anyone else had held, Wittgenstein concluded that it cannot be pictured even in logical pictures, that it cannot be said: Ethics cannot be put into words. If this conclusion was mystical, its mysticism was grounded in logic

It would be a mistake to ask at this point whether Wittgensteins semirealism, in ethics or metaphysics, is true or right. In the Introduction I remarked that semirealism is not a theory about certain peculiar objects or facts, those that are semireal, but rather that it exemplifies a conception of philosophy more sophisticated than such philosophical theory about object or facts might suggest. I shall say much more about this conception in chapter 13.

Part Two: METAPHYSICS HUMANIZED It is solely from the human standpoint that we can speak of space, of extended things (Kant)

Chapter Five: THE ROLE OF LANGUAGE IN COGNITION

1.

Empirical and a priori questions.

135

I suggested in the Introduction that anthropocentrism in epistemology and ethics is natural and understandable but indefensible, while in metaphysics it is unnatural and almost incomprehensible but at least as defensible as Kants transcendental idealism and contemporary antirealism. It is due in metaphysics not to humans natural interest in humans, as it is in epistemology and ethics, but to purely philosophical considerations. The present Part is devoted to making these considerations clear. In effect, it explains and defends the humanization of metaphysics. For, as we saw in the Introduction and will further see here, there can be no justification for just rejecting antirealism and returning to pre-Kantian metaphysics. Yet, we also saw, anthropocentrism in metaphysics leads, however tacitly, to the absurdity of human creationism. The explanation of how antirealism may avoid anthropocentrism, and thus how antirealist metaphysics may be dehumanized, will be our task in Part Three.

The subject matter of metaphysics is traditionally described as the world or reality as a whole. Although world usually signifies, if not just the earth, the spatiotemporal world, the cosmos, it may include also sense data as well as the immortal souls believed to be associated with human bodies. Reality usually has a broader sense. In addition to the spatiotemporal world, reality may include also abstract, nonspatiotemporal entities such as Platonic universals and numbers, as well as God, who by being the creator of the world is not in the world. But I shall not enforce this distinction between world and reality and will use the terms interchangeably.

I have used the term realism for the metaphysical view that the world is independent of our faculties of cognition, in the contemporary sense of cognition in which cognition
136

includes perceiving and understanding (conceptualizing) as well as describing (by writing or speaking). And I have used the term antirealism for the opposite view, including Berkeleys subjective idealism, Kants transcendental idealism, Hegels absolute idealism, as well as recent views such as Hilary Putnams internal realism, Michael Dummetts antirealism, and Nelson Goodmans irrealism. Metaphysical realism and antirealism are usually understood as concerned with all reality, and therefore should be distinguished from realism and antirealism regarding just an alleged part of it, e.g., moral values (moral realism/antirealism), abstract entities (Platonic realism/antirealism), the theoretical entities of physics (scientific realism/antirealism), and so on.

My concern in this book is with metaphysical realism and antirealism. But, as we saw in the Introduction, even within them, important distinctions are needed. One can be a realist regarding things but an antirealist regarding facts and thus regarding the world, if the world is the totality of facts, not of things. One can be a realist regarding the spatiotemporal/physical and causal structure of the world but an antirealist regarding its logical structure: One may believe that there are physical objects but deny that there are logical objects. Moreover, asserting and denying the reality of an item might not be the only options. Bertrand Russell held that some things do not exist but do subsist, and Wittgenstein held that some things cannot be said but do show themselves.

The project of metaphysical antirealism amounts to the humanization of metaphysics. Its contemporary version is linguistic antirealism. Unlike Kants transcendental idealism, which focused on the role of our faculties of sensibility and understanding in shaping cognition and thus the world as cognized, linguistic antirealism focuses on the role of our
137

language faculty, or organ as Steven Pinker has called it. Contemporary antirealists thus avoid Kants questionable, appeal to inhabitants of consciousness such as sensations, ideas, and concepts. As David Armstrong has remarked, Concepts [understood as mental items] are a more mysterious sort of entity than linguistic expressions.109 The language faculty may in some sense be mental, but its role in cognition can be understood and defended, as indeed it was by Wittgenstein, Gilbert Ryle, J. L. Austin, and P. F. Strawson, without appeal to any manifestations of its work in consciousness. Linguistic antirealism thus avoids the seemingly absurd implication of antirealism that the world is an inhabitant of one or several humans consciousness. Moreover, even if the human mental faculties are in some sense properties of the human brain, the role of the language faculty in cognition can be understood and defended without appealing to neuroscience. Linguistic antirealism thus avoids also the absurd implication that the world is in one or several humans brains.

Of course, it would be no less absurd to hold that the world is in language, that it is linguistic. But in Part Three we shall find that this seeming implication of linguistic antirealism can be avoided. Wittgenstein did write: The limits of my world are the limits of my language, but he also wrote, I am my world. And Heidegger insisted that it is language, not we, that speaks. These are obscure pronouncements, to which I shall return. They suggest that linguistic antirealism can be understood in such a way that it involves no reference to language as a human, a zoological, phenomenon. But there would be little motivation for accepting linguistic antirealism if human cognition were not believed to be essentially or at least importantly linguistic. I shall therefore begin by considering the role of language in cognition.
138

The central event in both Anglo-American and continental 20th century philosophy was its linguistic turn. According to Michael Dummett, after Frege philosophy of language replaced epistemology as the central branch of philosophy, just as after Descartes epistemology had replaced metaphysics. The phrase linguistic turn was coined by Gustav Bergmann and popularized by Richard Rorty in order to characterize recent Anglo-American philosophy, but it applies also to much of contemporary continental philosophy. The turn began with the publication of Wittgensteins Tractatus LogicoPhilosophicus after the First World War and the birth shortly thereafter of logical positivism in Austria, but it hastened with the migration in the following two decades of most of the positivists to Great Britain and especially the United States. In addition to claiming, as the 19th century positivists had done, that all knowledge is empirical and in principle scientific, thus raising the question whether there was a subject matter left for philosophy, the logical positivists held that philosophy did have a subject matter but that it was language. To distinguish philosophy from the empirical disciplines of linguistics and lexicography, however, they identified that subject matter more specifically as what they called the logic of language. For logic, of course, has always been a branch of philosophy and a paradigm of an a priori discipline.

But the linguistic turn probably would not have been taken were it not for the assumption, often tacit, that cognition, or at least thought, without language is impossible. The philosophers who took the turn considered this assumption true as a matter of empirical fact, as indeed it seems to be, except, as we shall find in section 4 of this chapter, in the crucial case of what in the Introduction I called logical cognition. Yet few attended to the
139

relevant empirical facts. For example, a highly relevant fact, to which I shall return, is the apparent occurrence of cognition of numbers or at least of numerosity in both prelinguistic children and some nonhuman animals. No armchair philosophy can discuss it responsibly, yet obviously it is an empirical datum crucial for the general question whether cognition without language is possible. And the ultimate value of the linguistic turn, thus of much of the subsequent course of Anglo-American philosophy, depends largely on how that question should be answered.

I said in the Introduction that philosophy cannot competently investigate empirical facts, and this is true also of those relevant to that question. But this does not mean that philosophy should be unaware of the latter. After all, they were the larger context of the linguistic turn and thus of the varieties of antirealism most common today. And some are so obvious as to require no special competence or investigation, e.g., that mathematics requires symbols. But other relevant facts are not obvious, including the alleged fact that speakers of different human languages perceive and think of the world differently. In this chapter I shall offer a look at what has been said about these questions, first by philosophers and then by linguists and psychologists, though on the present topic disciplinary distinctions are not always sharp. No attempt will made to provide a comprehensive account. For that a separate book, indeed several books, would be required. I shall only illustrate what can be found in this important part of intellectual history. And we must keep in mind throughout that the linguistic turn and linguistic antirealism are logically independent of each other, even though the adoption of the former encourages the adoption of the latter. Linguistic antirealism is a metaphysical position. The linguistic turn was essentially methodological.
140

Unfortunately, the discussions by philosophers of the question whether cognition without language is possible have seldom been detailed, and the discussions of it by scientists have seldom been useful. Philosophers have relied usually on speculations or, at best, exercises in introspection. Scientists have been hampered by the virtual impossibility, moral and social, of genuine experimentation with children, before or after they acquire language, just as the impossibility of genuine experimentation in economics has made the latter the dismal science. That the most advanced sciences, the so-called hard sciences physics, chemistry, biology are experimental, not just empirical, while the soft sciences economics, sociology, political science, and much of psychology are not, is not an accident. Experiments vastly augment not only the number but, more importantly, the variety of relevant empirical data and thus make detection of significant, rather than merely coincidental, correlations much easier.

That there is a connection between cognition and language is implied by the historically influential use of logos in Greek for both reason and word. The definition that man is a rational animal could be understood also as saying that man is a speaking animal. This is why the question whether cognition without language is possible has often been taken to ask whether thought without language is possible. The first definition of thought in the Oxford English Dictionary reads as follows: The action or process of thinking; mental action or activity in general, esp. that of the intellect; exercise of the mental faculty; formation and arrangement of ideas in the mind. We are also told that think is The most general verb to express internal mental activity, excluding mere perception of

141

external things or passive reception of ideas. The definitions of the corresponding words in major dictionaries of other languages are similar.

As the OED implies, thought and think have a wide sense, that of mental action or activity in general, as well as a narrow sense, that of mental action or activity of the intellect. They have these two senses also in philosophy, though philosophers, especially Brentano and Husserl, usually add that the action or activity is intentional, directed upon an object, real or not, and describe being conscious of something as an act. Descartes used them in their wide sense. After he offered the argument I think [cogito, pense], therefore I am, he explained, What is a thinking thing [res cogitans, une chose qui pense]? It is a thing that doubts, understands, conceives, affirms, denies, wills, refuses; that imagines also, and perceives. Clearly, by thought Descartes meant what later philosophers, in particular, Brentano, Husserl, and Sartre, called consciousness. Philosophical theories of thought in this sense are essentially philosophies of mind. For example, soon after Descartes offered his argument, the British empiricists proposed theories of mental activity as a succession or association of discrete ideas, whether abstract ideas (roughly the concepts of later philosophy), ideas of sensation (later called sense-data), or ideas of the imagination (later called mental images).

Descartes drew a sharp distinction between a thinking thing and the human body. A thinking thing is simple, it has no parts. The body, in contrast, is a paradigm of an entity that has parts, organs an anatomy, in the modern literal as well as the etymologically pregnant sense. Indeed, one of the standard arguments for the immortality

142

of the soul was that it is simple and therefore incapable of corruption, disintegration, falling apart, which is the common conception of how a thing ceases to exist.

Later philosophers more commonly used thought in its narrow sense, that of mental action or activity of the intellect. In that sense, thought is the level of cognition expressed in Greek by nosis and in English by understanding, conceptualization, or judgment. Unlike a neonates sense perception, thought so- understood involves the application of concepts and seems distinctively human. Indeed, in the 17th century, John Locke declared that brutes think not, on the grounds that they are incapable of entertaining abstract ideas and using language.

In the 18th century, Kant rejected the atomism of the empiricists view of mental activity as a succession of ideas by pointing out the synthesizing function of the understanding (Verstand). In the 19th century, Franz Brentano distinguished among presentation, judgment, and phenomena of interest. Judgment necessarily involves presentation, because one must be aware of what the judgment is about (whether by sensing it or through a concept). And judgment makes possible the characteristically mental actions or states of acceptance or rejection, belief or disbelief.110 In the 20th century, H. H. Price followed Kant by regarding thinking as the employment of concepts, even in recognition, and argued that a concept is essentially a principle of classification.

Unlike primitive sense perception and imagination, a judgment has as its object a proposition, what is judged to be true or false. Propositions usually are understood in philosophy as the primary vehicles of truth-value. Judgments, sentences, beliefs,
143

opinions, etc., are said to be true or false in virtue of the propositions they express. A proposition is expressed by a declarative sentence, while a judgment is expressed by the assertion of that sentence. Frege used thought (Gedanke) in a sense close to that of proposition. Thoughts, he held, belong in a third realm, neither mental nor physical, and thus they are not judgments, which presumably are mental.111

Since a judgment is expressed by the assertion of a declarative sentence, it can be expected to have parts and a structure analogous to those of a sentence. Indeed, Jerry Fodor has argued that there is a language of thought, thus attributing to thought the sort of structure characteristic of language, though he regards thought to be a state or activity of the brain.112 And Gustav Bergmann analyzed all mental acts as complexes, facts or states of affairs, each consisting of a momentary particular and two attributes that the particular exemplifies, namely, a species that determines the kind of mental act it is (e.g., perceiving, remembering, imagining) and a proposition (sometimes called text by Bergmann) that determines which state of affairs is its object (intention). He held that that even objects of ordinary sense perception and imagination are states of affairs, and thus require sentences for their expression in the ideal language.113

In this book the question whether there is thought without language would have a clearer and more promising focus if we ask, instead, whether there is cognition without language. For it is the term cognition, not mind or even judgment, that captures literally what is relevant to the epistemological and metaphysical issues that give rise to antirealism. Yet, even then, the question remains too broad, partly because an affirmative answer seems obvious. Neonates do enjoy sense perception, surely a level of cognition, however
144

rudimentary, months or years before they learn to speak; they also communicate, though not by speaking but by crying. Later, though still before learning to speak, they enjoy also recognition, e.g., of Mother, a primitive level of conceptual cognition, and perhaps they even engage in rudimentary reasoning, e.g., that crying leads to being fed. We shall find in section 4 that only logical cognition can be considered clearly impossible without language, because there the impossibility is logical.

Since language is an empirical subject matter, the study of its role in cognition should also be empirical. In keeping with the stance, explained in the Introduction, regarding philosophical inquiries into empirical matters, the present chapter will offer no theories about language. It will be limited to a brief review of what others have said, often providing lengthy quotations. But, to deserve attention, an empirical study of language must be serious. Casual empirical observations and incautious generalizations from personal experience can be grossly misleading.

A useful discussion, scientific or philosophical, of the question whether cognition without language is possible must be specific regarding the respects in which it might or might not be possible without language. It must also be specific regarding the kinds of cognition, language, and possibility that are being considered. Elementary and obviously relevant distinctions must be made. The familiar views tend to be intolerably general and vague. Surely, there are fundamental differences between sense perception and calculation, between the language of babes and the language of bards, and between causal impossibility and logical impossibility. Judgments about the causal possibility or impossibility of cognition without language are especially hazardous. One may be
145

unable to balance ones checkbook without talking to oneself, but other people can. When driving in a strange city, one may be compelled to issue to oneself instructions like Turn left on 3rd street, then right at second stop-light, but other people might not. The professional activities of some people, especially philosophers, consist almost entirely of talking, reading, and writing. (Sartre titled his autobiography Words.) But farming, fishing, and acrobatics do not.

Language might not be causally necessary even for some advanced levels of cognition such as designing a house and planning a trip. But surely it is necessary for doing physics or mathematics. A flys perceptual cognition surely does not require language, but any conceptual cognition that whales engage in might require it. And in the case of extraterrestrial life forms, angels, or God, language might not be necessary for any cognition. To be sure, a human being who lacks a language cannot do physics or mathematics, but surely God can, and for all we know so can intelligent beings outside our solar system. We may not understand what such cognition might be like, but then neither can we form auditory mental images of the high-frequency sounds that dogs but not humans can hear. We certainly need language, symbols, for cognition in physics and mathematics, but this may be due to our limited cognitive powers. A god might cognize directly, for example, the ultimate constitution of matter, without relying on language or any symbols, including those of mathematics.

The truth is that, so far, there is neither sufficient knowledge nor conceptual maturity in psychology or biology to judge responsibly whether thought or cognition is causally

146

possible without language. But judgments about whether it is logically possible can be made in philosophy. This question is not empirical, it is properly philosophical.

2. Philosophical opinions. Let us briefly review what philosophers have said about the role of language in cognition. In the Cratylus Plato distinguished between names and the things they name, and considered whether names are arbitrary and conventional. At the beginning of modern philosophy, Bacon wrote, Men believe that their reason governs words; but it is also true that words react on the understanding; and this it is that has rendered philosophy and the sciences sophistical and inactive.114 Hobbes dwelt on language in detail, arguing that true and false are attributes of speech, not of things.115 Locke held that brutes abstract not, because they talk not: the power of abstracting... [is] an excellency which the faculties of brutes do by no means attain to.116 Leibniz wrote, nothing exists in the intellect that was not before in the tongue except intelligence itself. Hegel remarked that [t]he forms of thought are, in the first instance, displayed and stored in human language.117 These contentions were accepted by most 20th century philosophers. As we have seen, Wittgenstein declared that the limits of the language ... mean the limits of my world, and Heidegger announced that Language is the house of being.

But in Thinking and Experience, published in 1953, H. H. Price wrote, it is sometimes supposed that no intellectual activity of any kind can occur without the use of words. This is not true of recognition.... Recognition is a prelinguistic process in the sense that it is not dependent on the use of words... [W]ords themselves have to be recognized.118
147

Prices view was plain common sense. Recognition is a level, indeed a fundamental level, of all cognition, primitive or advanced. It is present in children long before they acquire language. In adults it occurs sometimes when no name or even description of what is recognized is available, for example in the recognition, reading, of facial expressions.119 And creative work in music or painting, as well as in mathematics and advanced theoretical physics, only minimally involves the use of a natural language, though in mathematics and physics it does require mathematical symbolism. Price felt obliged to say what I quoted because of the dominance at the time of the philosophy of ordinary language, according to extreme versions of which, he wrote, an intelligent being...must always be talking to himself or to others.

Indeed, in his Philosophical Investigations, which by coincidence was published posthumously also in 1953, Wittgenstein wrote, Is thinking a kind of speaking? One would like to say it is what distinguishes speech with thought from talking without thinking.-- And so it seems to be an accompaniment of speech. A process, which may accompany something else, or can go on by itself. Say: Yes, this pen is blunt. Oh well, itll do. First, thinking it; then without thought; then just think the thought without the words....

When I think in language, there arent meanings going through my mind in addition to the verbal expressions: the language is itself the vehicle of thought..

148

While we sometimes call it thinking to accompany a sentence by a mental process, that accompaniment is not what we mean by a thought. Say a sentence and think it; say it with understanding: And now do not say it, and just do what you accompanied it with when you said it with understanding!120

Expounding what he took to be Wittgensteins view, Renford Bambrough argued in 1960 that objects called by the same name are distinguishable from other objects just by the fact that they are uniquely called by that name: they neither have a common property nor bear to each other resemblances except for what Wittgenstein had called family resemblances.121 In 1973 Michael Dummett wrote that to possess a concept is to be a master of a certain fragment of language122 and in 1975 that Only with Frege was the proper object of philosophy finally established: namely, first, that the goal of philosophy is the analysis of the structure of thought; secondly, that the study of thought is to be sharply distinguished from the psychological process of thinking; and, finally, that the only proper method for analyzing thought consists in the analysis of language.123 In 1982 we find Donald Davidson declaring that a creature cannot have a thought unless it has language.124 In 1995 Quine wrote that thought, as John B. Watson claimed, is primarily incipient speech, though Quine allowed that in the case of artists, acrobats, and engineers, sometimes they are thinking with nonverbal muscles.125 However, after thus paying homage to muscular thought, Quine declared that beliefs, meanings, ideas, properties, and propositions are entia non grata, adding, But believing, doubting, hoping, expecting, regretting, all continue alive and well, and their objects, by my lights, are sentences.126 Even Nelson Goodman, who for years had held that what he called

149

symbol systems need not be linguistic, slipped in 1988 and denied that there is a readymade world beyond discourse.127

Nevertheless, analytic (though perhaps not continental) philosophy soon went beyond these extremist views. A year after Goodman wrote what I just quoted, Keith Donnellan, discussing Kripkes puzzle about belief, wrote, the puzzle we feel really has nothing to do with language or with languages at all.128 And, three years later, we find even Dummett asserting that linguistic practice is no more sacrosanct, no more certain to achieve the ends at which it is aimed, no more immune to criticism or proposals for revision, than our social, economic or political practice.129 Though intended as an attack on the philosophy of ordinary language (especially J. L. Austins), Dummetts assertion had independent philosophical significance by tacitly implying that criticism of language presupposes a cognition that is not dependent on language. It is noteworthy that, except perhaps for Quines casual reference to Watson, these opinions about language were offered without mention of any empirical data or of the relevant empirical sciences.

Some have come to believe that, though all cognition, from simple perception to mathematical reasoning, is necessarily symbolic, representational, the representations need not belong to a natural language they can belong to a language of thought, in Jerry Fodors phrase.130 For Fodor this is just a philosophical hypothesis, but he thinks that it has the support of neuroscience. Tim van Gelder explains: Contemporary orthodoxy maintains that it [cognition] is computation: the mind is a special kind of computer, and cognitive processes are the rule-governed manipulation of internal symbolic representations.131 Van Gelder then draws the natural conclusion that because the cognitive system traffics only in symbolic representations, the human body and the
150

physical environment can be dropped out of consideration; it is possible to study the cognitive system as an autonomous, bodiless, and wordless system whose function is to transform input representations into output representations.132 If cognition is what such a cognitive system does, cognition does not involve language, even if some neural states are described as symbolic, though neither is it also directed upon objects and thus is not cognition of reality or of anything. At most these neural states would be like the characters employed by a computer, but as John Searle had shown in 1980, the idea that they mean anything other than what the programmer means by them is grossly mistaken.

3. Scientific opinions. Let us briefly review what linguists and psychologists have said about the role of language in cognition. More than a century after Locke announced that brutes abstract not, the German linguist Wilhelm von Humboldt wrote, Language is the formative organ of thought. Intellectual activity, entirely mental, entirely internal, and to some extent passing without trace, becomes, through sound, externalized in speech and perceptible to the senses. Thought and language are therefore one and inseparable from each other. But the former is also intrinsically bound to the necessity of entering into a union with the verbal sound; thought cannot otherwise achieve clarity, nor the idea become a concept. The inseparable bonding of thought, vocal apparatus and hearing to language is unalterably rooted in the original constitutions of human nature, which cannot be further explained.133 Humboldt offered no evidence in support of these opinions. He did speculate, however, that existing languages differ from each other markedly in regard to perfection, which he attributed to national, ethnic, and racial differences among their

151

speakers. In the United States, Emerson famously held that language is the archives of history, but laid no claim to arriving at this opinion scientifically.

Early in the 20th century, Ferdinand de Saussure wrote (or was reputed to have said, since the text was reconstructed from lecture notes): Psychologically, setting aside its expression in words, our thought is simply a vague, shapeless mass. Philosophers and linguists have always agreed that were it not for signs, we should be incapable of differentiating any two ideas in a clear and constant way. In itself, thought is like a swirling cloud, where no shape is intrinsically determinate. No ideas are established in advance, and nothing is distinct, before the introduction of linguistic structure.134 Saussure did not say how he (or the philosophers and linguists to whose authority he appealed) learned all this. For example, did he study languageless animals and human infants before they learned to speak?

William James famously announced that the world perceived by an infant is a blooming buzzing confusion, but he also failed to say how he knew this.135 James did think, however, that thought without language is perfectly possible. In support, he quoted at great length a deaf-mute instructor, Mr. Ballard, who was reported to have written: It was during those delightful rides, two or three years before my initiation into the rudiments of written language, that I began to ask myself the question: How came the world into being? When this question occurred to my mind, I set myself to thinking it over a long time. My curiosity was awakened as to what was the origin of human life in its first appearance upon the earth, and of vegetable life as well, and also the cause of the existence of the earth, sun, moon, and stars. Concerning this passage, Wittgenstein later
152

wrote, Are you sure one would like to ask that this is the correct translation of your wordless thought into words? Do I want to say that the writers memory deceives him? I dont even know if I should say that. These recollections are a queer memory phenomenon, and I do not know what conclusions one can draw from them about the past of the man who recounts them.136

Despite Wittgensteins misgivings, if sincere such reports of personal experience are useful, even if unscientific. They must be read with caution, of course, but if sincere the evidence they provide is unavailable elsewhere. For example, if Helen Kellers acquisition of language was indeed as she (and her teacher, Anne Mansfield Sullivan) claimed, then it was an extraordinary example of how, despite the absence of the usual sort of exposure to language, one can acquire a rich, indeed stylistically first-rate, linguistic proficiency. Yet Kellers editor did write: Philosophers have tried to find out what was her conception of abstract ideas before she learned a language. If she had any conception, there is no way of discovering it now; for she cannot remember, and obviously there was no record at the time. She had no conception of God before she had the word God.137

Among such reports of personal experience, the mathematician Kalvis M. Jansons is exceptionally instructive. He writes,

From an early age I found that many things were easier to think about without language. This usually, but not always, meant thinking in terms of pictures and was particularly true when trying to make or understand
153

intricate mechanisms. To me, abstract pictures and diagrams feel more important than words. For example, in maths I never feel that I fully understand something unless I have found a way of visualizing the system, although this is not usually a substitute for mathematical symbolism. Many of my original mathematical ideas begin with some form of visualization and, once that seems to be a sound model of the system I am analysing, the analysis usually falls into place. Occasionally this fails to happen and then I return to and modify the visual model....

It would be a mistake to believe, however, that nonverbal thought has to involve pictures. For example, three-dimensional space can be equally well represented in what I often think of as a tactile world. If I am trying to remember a complicated knot when no rope is available, I usually imagine the finger movements involved and the feel of the knot being tied without picturing it in my mind or moving my hands at all. Knots are examples of things that are extremely hard to describe and remember in words, and people who attempt to do so usually forget them very quickly and are poor at spotting similarities between complicated knots.....138

Not all reports of personal experience on our topic are as useful as Jansons. This is especially true of some familiar claims by leading psychologists, which must be understood as reports of personal experience since no other evidence is cited. For example, J. B. Watson wrote that thinking is merely talking, but talking with concealed
154

musculature.139 Leonard Bloomfeld held that thinking is talking to ourselves, suppressing the sound-producing movements and replacing them by very slight inaudible ones.140 (I should mention that today there is evidence that thinking is not prevented by drugs that suppress muscular activity.) B. F. Skinner wrote, thought is simply behavior verbal or nonverbal, covert or overt.141 Edward Sapir appealed to the frequent experience of fatigue in the speech organs after intensive thinking.142 This, surely, was a report of personal experience. But Sapir went further: Human beings do not live in the objective world alone..but are very much at the mercy of the particular language which has become the medium of expression for their society..The fact of the matter is that the real world is to a large extent unconsciously built up on the language habits of the group.We see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices of interpretation. 143 This is a crude version of what I have called linguistic antirealism, but offered as a scientific hypothesis and then left unsupported by empirical evidence. Sapir also held that meaning is either a visual image, a class of images, or a feeling of relation.144 He did not say how he knew all this. Sapir also charged philosophers with projecting grammatical categories into their conceptions of the world so that innocent linguistic categories may take on the formidable appearance of cosmic absolutes.145 In 2008, field linguists discovered a new language, Koro, spoken in some villages in northeastern India.146 One of the linguists was reported to say that languages like Koro construe reality in very different waysThey uniquely code knowledge of the natural world in ways that cannot be translated into a major language. But we were not told how the linguist knew this.

155

Sapirs student, Benjamin Lee Whorf, did engage in extensive empirical work on Indian languages. But he, too, cited no particular evidence when he wrote, Thinking...follows a network of trails laid down in the given language...The individual is utterly unaware of this organization and is constrained completely within its unbreakable bonds.147 Whorf thought that he and a Hopi could not discuss the same world. According to him, [S]egmentation of nature is an aspect of grammar...We cut up and organize the spread and flow of events as we do, largely because, through our mother tongue, we are parties to an agreement to do so, not because nature itself is segmented in exactly that way for all to see.148 Perhaps it was natural that what came to be known as the Sapir-Whorf hypothesis was offered in American linguistics. It was in America that non-IndoEuropean languages were plentiful and readily available for study: there were more than 100 American Indian languages. In this respect American linguistics had an important advantage over European linguistics.

Watson, Bloomfield, Sapir, Whorf, and Skinner were the most influential scientists on our topic in the first half of the 20th century. This may explain in part the unbridled linguisticism characteristic of American philosophy in that period, especially since respect for science was part of the logical positivist doctrine that dominated it at the time. Their supposedly scientific view seemed identical with the philosophical view that Hilary Putnam described as holding that objects arise out of discourse, rather than being prior to discourse.149 But were Watson, Bloomfield, Sapir, Whorf, and Skinner right? It is generally agreed that to answer this question on empirical grounds and with confidence we would need to study the prelinguistic infants cognition. And such studies have been notoriously unsuccessful, probably because serious experimentation on children, e.g.,
156

delaying a childs learning a language by a few years, would be grossly unethical. Less alarming experiments on children tend to be inconclusive, perhaps because small children, not just infants, seldom cooperate with or even take seriously the investigators, who often happen to be their parents. The theorizing about early cognition and language acquisition by Brbel Inhelder and Jean Piaget, for example, tends to be little more than speculation.150

The most important development in linguistics, of course, was that initiated by Noam Chomsky. He is not only a major linguist, but also a major philosopher whose views (especially in his later writings) on scientific methodology and the mind-body problem are of unsurpassed sophistication and subtlety. Chomskys linguistics is of special relevance to the topic of this chapter. Its central thesis was anticipated by Humboldt, who wrote, Since the natural disposition to language is universal in man, and everyone must possess the key to the understanding of all languages, it follows automatically that the form of all languages must be essentially the same, and always achieve the universal purpose.151 This thesis is essentially Kantian and it is defended throughout Chomsky's works.

Discussions of Chomskys linguistics need not bog down in what Chomsky rightly regards as misguided puzzling over whether what he calls universal grammar has psychological reality, rather than just utility as a theoretical hypothesis. Surely he is right that there is no principled difference between the propriety of such hypotheses in linguistics and the propriety of hypotheses in physics.152 Chomsky has always held that the status of innate linguistic competence is ultimately biological.153 The innate
157

structures to which he appeals are not accessible to consciousness, presumably because they are in the brain.154 This is a quasi-Kantian theme.

Chomsky's nativism is not what is argued in this book, if for no other reason than that, unlike what a philosophical book can properly attempt, it is essentially scientific and subject to empirical confirmation. But it does imply that at least what I have called the logical structure of language is not learned from experience. Indeed, Chomsky has argued that the familiar [quantifier-variable] notation is read off of the logical form that is the mental representation for natural language.155 But this is offered as a substantive scientific hypothesis concerning a causally necessary condition of language and thus of logical cognition. And the mental representation in question is ultimately identified with a feature of the brain. The hypothesis is hardly philosophical, even though, as Chomsky repeatedly makes clear, it is proposed on a very high level of abstraction and his approach conforms to sophisticated scientific method.

But a nativism such as Chomskys is not as novel as many take it to be. It should not seem innovative to anyone familiar with Aristotles distinction between first actuality and second actuality: one of Aristotles applications of it was to knowledge of grammar! Nelson Goodman wrote about Chomskys view: until the term innate idea is applied, what is advocated is the rather trivial truth that the mind has certain faculties, tendencies, limitations.156 But even if this truth is trivial, its importance was neglected in modern philosophy and psychology until Kant drew attention to it.

158

Chomskys quasi-Kantian appeal to innate psychological structures is not a genuine alternative to antirealism in metaphysics. At most, it is a sort of biological Kantianism, a biological transcendental idealism.157 Indeed, Chomskys application of it is not restricted to universal grammar. He applies it to perception we know that the visual system of a mammal will interpret visual stimulations in terms of straight lines, angles, motions, and three-dimensional objects 158 He applies it also to conceptualization, in ways Kant would have found surprising:

[H]uman nature gives us the concept climb for free. That is, the concept climb is just part of the way in which we are able to interpret experience available to us before we even have the experience. That is probably true for most concepts that have words for them in language. This is the way we learn language. We simply learn the label that goes with the preexisting concept. So in other words, it is as if the child, prior to any experience, has a long list of concepts like climb, and then the child is looking at the world to figure out which sound goes with the concept. We know that the child figures it out with only a very small number of presentations of the sound.159

The role of language in cognition was discussed by Chomskys student Steven Pinker in his widely read book The Language Instinct. He wrote,

The idea that thought is the same thing as language is an example of what can be called a conventional absurdity: a statement that goes against all
159

common sense but that everyone believes because they dimly recall having heard it somewhere and because it is so pregnant with implications.....

Sometimes it is not easy to find any words that properly convey a thought. When we hear or read, we usually remember the gist, not the exact words, so there has to be such a thing as a gist that is not the same as a bunch of words. And if thoughts depended on words, how could a new word ever be coined? How could a child learn a word to begin with? How could translation from one language to another be possible?160

This is plain good sense. We may wince, however, when Pinker claims that psychologists have shown that babies can do arithmetic: The developmental psychologist Karen Wynn has recently shown that five-month-old babies can do a simple form of mental arithmetic. She used a technique common in infant perception research. Show a baby a bunch of objects long enough, and the baby gets bored and looks away; change the scene, and if the baby notices the difference, he or she will regain interest. The methodology has shown that babies as young as five days [sic] old are sensitive to number. In one experiment, an experimenter bores a baby with an object, then occludes the object with an opaque screen. When the screen is removed, if the same object is present, the babies look for a little while, then get bored again. But if, through invisible subterfuge, two or three objects have ended up there, the surprised babies stare longer.161 Wynns experiments had obvious merits, but Pinkers interpretation of their results is nave.

160

The naivet is not philosophical but scientific, though it fits Mark Twains description of science as an endeavor in which one gets such wholesome returns of conjecture out of such trifling investment of fact. Pinkers interpretation of the results of Wynns experiments is an exaggerated abductive inference from meager and ambiguous data, not much better than Benjamin Whorfs inference, from what he learned about Indian languages, that thought depends on language, an infrence Pinker justly criticizes.162 To claim, on the basis of experimental results such as Wynns, that babies have knowledge of arithmetic is like claiming that since birds know how to fly they have knowledge of aerodynamics. Indeed, another scientist, Alan M. Leslie, has written of [Baillargons] important discovery about the young infants understanding of mechanics, on the basis of an experiment similar to Wynns.163 In the same collection of papers as Leslies, we also find Gabriel Horn arguing that thought does not require language on the grounds that mice distinguish self from other objects and thus are aware of their bodies because they (generally) dont get stuck in holes too small for them to get through.164 Such opinions in psychology bring to mind recent reports of discoveries in genetics of a gambling-gene and an alcoholism-gene.

But we should not complain inordinately about them. Despite the exaggerated conclusions drawn, the underlying facts are important in their own right. While crediting babies with cognition of arithmetic at best ignores the nature of arithmetic, there is no doubt that any prelinguistic cognition of what in the scientific literature is called numerosity would be relevant to the question of the dependence of cognition on language. More recent experiments, similar to Wynn's but involving monkeys, seem to confirm that cognition without language of numerosities does occur in monkeys, but they
161

do not imply that the distinctively human cognition exhibited in arithmetic is possible without language. On the basis of his extensive, lifelong research, David Premak has argued that [t]he thoroughgoing superiority of the pre-language child on every measure used indicates that the cognitive advantage to the human is not one introduced by language.165

Steven Pinker has written: Grammar offers a clear refutation of the empiricist doctrine that there is nothing in the mind that was not first in the senses. Traces, cases, X-bars, and the other paraphernalia of syntax [these are Chomskyan technical lingo] are colorless, odorless, and tasteless, but they, or something like them, must be a part of our unconscious mental life.166 The passage tells us that syntactical, and therefore presumably also logical, form corresponds to nothing perceived or otherwise experienced in the world. But this appeal to our unconscious mental life is simplistic, for the Sartrean reason, to which we shall return in chapter 11, that consciousness has no inhabitants, and therefore no unconscious inhabitants. Pinkers frequent talk of the seat of the language instinct in the brain, like all talk of the causes or sources, mental or extra-mental, of various sorts of cognition, casts little light on the status or nature of cognition. To be told, for example, that logical cognition is grounded, not in appropriate referents in the world, but in an unconscious structure of the mind or the brain is no more illuminating than to be told that it is grounded in evolution or in Gods will.

4. Language in logical cognition.

162

Our brief survey of what philosophers and scientists have said about the general question of the role of language in cognition suggests that it has received at best inconclusive treatment. But there is a much more specific question that can be answered with reasonable confidence. It concerns the role of language in logical cognition. To answer it requires no empirical research and remains properly a task of philosophy, which is the home of logic.

Wittgenstein held that the logical constants do not represent, that there are no logical objects. I have suggested that we speak more broadly of the logical expressions : variables, quantifiers, sentential operators, the verb to be in its senses of identity, existence, and predication, and the symbols or syntactic structures expressing those senses, as well as sentences expressing propositions and that we count as logical objects what all logical expressions so understood might be taken to represent. I called the kind of cognition that requires logical expressions for its expression logical cognition. Logical cognition includes but must not be confused with the sort of cognition pursued by logic. The scope of the latter is much narrower than the scope of the former. Statements expressing logical cognition usually contain also nonlogical expressions and few are a priori, while those expressing the sort of cognition pursued by logic contain only logical expressions and usually are taken to be a priori. I suggested that we call the former logical statements and the latter statements of logic. For example, All men are mortal is a logical statement because it includes the logical expression all, but it is not a statement of logic. Statements of logic are purely formal and usually employ only technical symbols. An example would be (( x) x ~ (x) ~ x)), i.e., Something is if and only if it is not the case that nothing is .
163

I shall discuss logical expressions, logical statements, and logical cognition in detail later. What is relevant to the question whether thought is possible without language is that if logical expressions do not represent, i.e., if they stand for nothing, then logical cognition is impossible without language, that all there is to it is language, though in chapter 9 I shall provide important qualifications. We have no access to what logical cognition is about except through language because it is not about anything. In this respect logical cognition is dramatically different from, say, perception, where in addition to talking about its objects we can perceive them. Not even God could know that all men are mortal by perceiving all men. God might perceive all individual things, perhaps an infinity of them, but he could not know by perception that they are all the individual things that there are: this is not something anyone can perceive because there is nothing that the word all stands for.

If the logical expressions do not represent, then a limited but far reaching linguistic species of conceptual antirealism, the species I called logical antirealism, ought to be accepted. Any advanced cognition requires statements that make essential use of logical expressions. If logical expressions stand for nothing, then those statements have no extralinguistic significance even if they contain nonlogical components that do.

Many expressions, e.g., in fiction, also stand for nothing, but they serve no cognitive function. The function of logical expressions, however, is unquestionably cognitive. Linguists distinguish between the structure of a sentence and its lexical elements, and acknowledge that both have a semantical function. But linguists tell us little about the
164

semantical function of the structure of a sentence. Even philosophers have little to say about it. It obviously depends, at least on part, on the logical expressions the sentence contains. The structure of a sentence is relevant to cognition because it contributes to the truth-value of the sentence. But it does not do this by standing for something in the world.

We may call the role of logical expressions in cognition transcendental, in a sense related to Kants yet different in that it applies to language, not to mental items or faculties. Indeed, the linguistic turn has often been described as a transcendental turn. If limited to logical cognition, it is more measured, discerning, and cautious and therefore more plausible than the more common, indiscriminate linguistic turn exemplified in the statements quoted earlier from Wittgenstein and Heidegger. The limits of my language may mean the limits of my world, but surely there is more to the world than language. Language may be the house of being, but surely it is not also the furniture in that house.

Logical antirealism rests on philosophical, a priori, not on empirical, a posteriori considerations. That logic is an a priori discipline has seldom been questioned. To be sure, in recent years doubts have been expressed, most notably by Albert Casullo,167 that any knowledge is a priori. But the fact remains that questions in philosophy, and particularly those in logic, are discussed and settled without making empirical appeals.

Our knowledge of the propositions of logic is a priori in accord with the standard definition of a priori knowledge as knowledge independent of, prior to, experience in respect to its evidential basis. By experience is usually meant sense perception and
165

introspection. Evidential basis stands for what one would cite in support, justification, of the speech action of claiming knowledge when asked, How do you know? or of the corresponding mental action of believing what one knows, if there are such actions. When what is known is inferred, the evidential basis is what the premises of the inference assert. Otherwise, what is known is self-evident, the evidential basis itself. But this definition merely says that a priori knowledge is not a posteriori knowledge. It is no more informative in philosophy than a definition of water as a liquid that has no color, odor, and taste rather than as H2O, as in chemistry. Can we offer a positive, more informative definition?

A priori knowledge has also been defined as prior to experience in the sense of being innate. But would this mean that one who knows arithmetic, the paradigm of an a priori discipline, knew it actually before birth? If it means only that one knew arithmetic potentially before birth, then all knowledge is innate. Another definition is that a priori knowledge is knowledge based solely on reason. But what is reason? Presumably, not reasoning in the sense of inferring: a proposition known a priori may be self-evident, known without inference. By reason many mean intellectual (rational) intuition, a certain kind of awareness.168 Such awareness need not be supernatural or even unfamiliar. If you hear the door bell ring three times, you may be aware of how many times it rang, but you do not hear a number. When you see a books color, you may be aware of the generic property color, but not by seeing it in addition to seeing the books color. If you feel joy, you may be aware of its goodness, but you do not feel that goodness as you feel the joy. Such awareness is intellectual in the sense that it involves intelligence, not sense perception or introspection, though these also are kinds of
166

awareness, and in that we can think about, but not picture or image, its objects. A posteriori and a priori knowledge are thus both based on awareness, though on radically different kinds of awareness. This should not be surprising. What would be surprising is if one could have knowledge without ever being aware of what one knows.

It might help to consider arithmetic, the paradigm of an a priori discipline. Mathematicians do not observe or feel the numbers 2 and 4, nor do they perform experiments. Of course, they learn the names of numbers empirically, which is true of all language-acquisition, but mathematicians do not think that their knowledge of propositions (equations) such as 2 + 2 = 4 is based on sense perception or introspection. Is it then based on intellectual intuition? If not, are mathematicians aware of anything that is relevant when they consider an equation? Some say that mathematics is concerned only with words and symbols. But is mathematical knowledge then just a matter of words or symbols?

Whatever the case with mathematical propositions may be, we have no awareness of logical objects, sensory or intellectual, because there are no such objects at all. In the case of the statements of logic we seem left with just our capacity to make them.169 We seem left with the conclusion that our knowledge of logic is grounded solely in language. This is the linguistic version of logical antirealism.

Despite the extensive debates over the general issue of realism/antirealism, little has been said about the specific issue of logical realism/antirealism, even though all advanced cognition involves cognition of logical structure. Perhaps this neglect has been due to the
167

comforting thought that logic, which is a branch of philosophy so advanced that it is often considered a separate discipline, has already dealt with and settled the issue. This, of course, is an illusion. Indeed, in the Tractatus, Wittgenstein held that the logical constants do not represent, and offered his truth-functional logic as an account of them. But, as we shall see in the following chapter, Russell felt compelled to allow for the existence of irreducibly negative and general facts, which surely would count as logical objects. He showed that attempts to reduce negative facts to incompatibility and general facts to conjunction or disjunctions of singular facts are unsuccessful.

It is fitting to end this chapter by quoting Nelson Goodmans reflections on its topic:

The conviction that thoughts are always in words has thrived on the recent philosophical focus on language, but has been primarily sustained by the incredibility of or incomprehensibility of the most familiar alternatives: looking upon the mind as a sort of motion-picture theater, or upon thoughts as ideas couched neither in words nor pictures nor any other symbols.

Yet the view of thinking as purely verbal leaves many an open question.

What of the difference we do perceive, and describe however inadequately as between having and not having a picture in the mind of a person or place? What happens when we have a thought we cannot put into words? Furthermore, since we do work and communicate overtly with symbols of all kinds, can thinking be with words alone? And do we have in the mind
168

images of words but no images of anything else? Or if a thought of a word is not an image, what is it?

To such questions, I can respond only with some tentative and conditional guesses or counterquestions: (1) We can obviously think of pictures, but can we think in pictures? - Yes, if we can think in words. (2) Are there thoughts without words? -Yes, if there are thoughts without pictures. (3) Are there pictures in the mind? Yes, if there are words in the mind. How can a picture be in the mind? Well, how can a word be in the mind?170

Chapter Six: METAPHYSICAL REALISM AND LOGICAL ANTIREALISM 1. The Logic of Realism In chapter 5 I considered the belief that cognition, at least thought, essentially involves language. This belief encouraged, and remains characteristic of, the version of metaphysical antirealism that came to dominate 20th century philosophy when it took the linguistic turn. I have called this version linguistic antirealism. I have also suggested that the role of language in cognition is especially evident in logical cognition, the sort of cognition that necessarily involves notions that are characteristically logical, such as negation and generality. I also suggested that antirealism with respect to logical cognition, which I called logical antirealism, is the most plausible version of linguistic antirealism and of metaphysical antirealism generally.
169

To gain a clear grasp of antirealism, we must see it in the context of the development of its opposite, realism. The guiding principle of realism may be put, roughly, as follows: acknowledge as being there what must be there if cognition and truth as correspondence are to be possible. This principle governed the construction of the ontological inventories, or categorial schemes, of Aristotle, Frege, Russell, and Bergmann. I select these philosophers for consideration here because they exemplify especially clearly the role of the principle in the development of realist ontology.

In the Categories Aristotle used the notions of said of and present in as primitive, and with unsurpassed elegance proposed the following inventory: (1) items said of but not present in something (the items would be, in the category of substance, what he called secondary substances and we may call substance universals, e.g., cat), (2) items present in but not said of something (particular accidents, in the nine categories of accident, e.g., this cats white color), (3) items both present in and said of something (universal accidents, in the nine categories of accident, the color white), and (4) items neither said of nor present in something (primary substances, or substance particulars, in the category of substance, e.g., this cat).

Freges inventory included (1) objects (roughly, what we call particulars or individuals), (2) first-level functions (called concepts but in a nonpsychological sense that is closer to that of property in current philosophy), which take objects as arguments and yield truth or falsity as values, (3) second-level functions, or our quantifiers, which also yield
170

truth or falsity as values but take first-level functions as arguments, (4) thoughts, the vehicles of truth-value, roughly what we call propositions, and (5) functions that take thoughts as arguments and yield compound thoughts, roughly what we call propositional connectives.171

Russells inventory, at least at the time of The Philosophy of Logical Atomism, included (1) particulars, (2) universals (properties), (3) atomic facts, (4) negative facts, thus at least one kind of molecular facts, and (5) general facts. Bergmann proposed a much richer inventory: (1) particulars, (2) universals, (3) the ultimate sorts of universality and particularity, in virtue of which an item is a particular or a universal, (4) facts, which he called complexes and which he allowed might be only potential, i.e., possible but not actual, (5) the propositional connectives, (6) the quantifiers, (7) the modes of actuality and potentiality, which pervade facts and render them either actual or potential, and (8) the three nexus of exemplification, set-membership, and meaning.172

The logic of the development of realism that these inventories exemplify now emerges. All four include the category of particulars. Even sense data, with which, a la Berkeley, Russell toyed, are particulars. The privileged status of the category of particulars may explain why some opponents of antirealism (perhaps motivated by the title of Nelson Goodmans Ways of Worldmaking) object to it by saying, Surely there were a sun and an earth before there were human beings and thus, for all we know, before there were minds, or language and concepts. But the antirealist claims, not that the sun and the earth were created by our thought or language (Goodman was not a sort of creationist), but that they
171

cannot be perceived, conceived, and described independently of our perception, thought, and language which is hardly more than, in Hilary Putnams phrase, a virtual tautology.173

Would realists be willing to say (1) that the sun and the earth are particulars, (2) that they have properties, (3) that the sun exemplifies the property of being a star and the earth the property of being a planet, (4) that it is a fact that there was a sun and an earth before there were human beings, (5) that the sun is not a planet and the earth not a star, (6) that they and other stars and planets were all that there was before there were human beings, (7) that they actually, not just possibly, existed then, (8) that the sun and the earth are members of their pair, (9) that some things, mental or verbal, mean, refer to, the sun and the earth. If realists refuse to make such assertions, they would display, however unwittingly, merely the continuing hold of 17th and 18th century British empiricism the confusion of knowing and thinking with perceiving and imagining. We are comfortable with particulars because they come to mind first when we asked what we perceive. But even if they are all that we perceive, they are hardly all that we think and know. Realists tend to ignore the question whether there could have been such items as the sun or the earth unless there were also items corresponding to the italicized words in the first sentence of this paragraph. Surely, they owe us a serious discussion of how sparse the ontology of their realism could be. That it cannot limit itself to such items as the earth and the moon i.e., to particulars became evident in the logical development of realism beyond the philosophical terra firma of particulars.

172

The first move in that development was to introduce the category of properties, a move with which Plato dazzled philosophers. It was almost inevitable because we think and speak not only of particulars but also of what they are. The move to Russellian facts (or Fregean thoughts or Bergmannian complexes) came more than two millennia later but seems also inevitable. We speak in sentences, not lists of names. If the description of the world requires sentences, what items in the world require this and might be said to correspond to sentences? As we saw in the Introduction with the example of Jacks admiring Jill but Jill not admiring Jack, surely not the items that correspond to bare names and predicates. Hence, the introduction at least of atomic facts. But a further move to molecular, even if only negative, and then to general facts seemed also required. Atomic sentences are woefully inadequate for any cognition that is at all advanced. Any language beyond that of babes requires molecular, at least negative, as well as general sentences, whether universal or particular. And then a move to negation (perhaps also the other connectives) and generality themselves, i.e., to what makes negative facts negative and what makes general facts general, appeared needed. This move was made explicitly by Frege and Bergmann, as well as by Russell in Theory of Knowledge,174 though not in The Philosophy of Logical Atomism. Bergmann even included in his inventory particularity, universality, actuality, and potentiality, as what allows us to distinguish between particulars and universals, and between actual and merely possible facts, as well exemplification, set-membership, and meaning, which respectively tie particulars to their properties, members of sets to their sets, and thoughts to what they are thoughts of, their intentional objects. Like Meinong, Bergmann has often been accused by other realists of going too far, but those realists do not explain where and how the line should be drawn. The usual half-hearted attempts to rescue realism from excesses
173

by adopting some sort of reductionism, whether of properties to tropes, of negative facts to incompatibility, or of general facts to conjunctions or disjunctions of atomic facts, were all shown by Russell in The Philosophy of Logical Atomism to fail because they presuppose what they claim to reduce.

Aristotles, Freges, Russells, and Bergmanns ontological inventories are in many ways similar, despite their major differences. All four include particulars and properties, though Frege called particulars objects and properties concepts. Frege's, Russell's, and Bergmann's inventories include a category of items that correspond to sentences, but Aristotles does not. Frege called those items thoughts, Russell called them facts, and Bergmann called them complexes. Fregean thoughts were said by Frege to belong to a third realm, which is neither mental nor physical, but this would be true also of Russellian facts and Bergmannian complexes. A fact may include constituents that are mental or physical, but in no clear sense can itself be called mental (admiration may be something mental, but is the fact that Jack admires Jill also mental?) or physical (what are the weight, size, and location of the fact that one weighs 160 pounds?) This is why facts are sometimes said to belong only in logical space, just as Fregean thoughts were said to belong to a third rtealm. Freges, Russells, and Bergmanns inventories included negative and general facts, but only Freges and Bergmanns included negation and generality themselves. Except in the abandoned Theory of Knowledge, Russell committed himself only to negative and general facts, not to what makes them negative or general. Only Bergmann followed the logic of realism further by including particularity, universality, exemplification, set-membership, meaning, actuality, and potentiality.
174

Realists were inexorably required by the realist principle to acknowledge such additional items, even though most did not. The reason is that we can hardly suppose that our thought or language corresponds to the actual, objective world of particulars if we do not suppose that most, if not all, of them are also in it. There would not be a sufficiently specific, definite fit between thought or language and the world without them. For example, the sentence Jack admires Jill could not, if true, correspond just to Jack and Jill, or even to Jack, Jim, and the relation of admiring, since the sentence Jill admires Jack also would correspond to those things. It must correspond to the fact that Jack admires Jill. Moreover, the sentence Jack admires Jill presupposes a certain order of the items it is about, and thus something in which the items have that order.

The motivation for antirealism is now not difficult to discern. As Russell said regarding disjunctive facts, it is not plausible that somewhere in the actual objective world there are all those things going about that the realist principle seems to require.175 This is implausible because, except for particulars and perhaps properties, the things supposed to be going about are all logical, in the broad sense of logical explained in the Introduction. It becomes tempting, therefore, to conclude that in the case of such things there are only the words or syntactic structures supposed to correspond to them.

But this conclusion is itself implausible. At least, it is too hasty. By distinguishing between saying and showing, Wittgenstein proposed a third alternative, which I called semirealism. In chapter 4 we saw that the distinction is neither idiosyncratic nor obscure.
175

It has an obvious application even to ordinary pictures. And Wittgensteins related picture theory of meaning and thought conformed to the traditional view that both involve representations, perhaps even mental pictures.

As we saw in chapter 4, Wittgensteins rejection of realism in logic was unequivocal and explicit: The possibility of propositions is based on the principle that objects have signs as their representatives [Prinzip der Vertretung von Gegenstnden durch Zeichen]. My fundamental idea is that the 'logical constants' are not representatives [nicht vertreten]; that there can be no representatives of the logic of facts (4.0312). But his rejection of antirealism in logic was also unequivocal and explicit: is it really possible that in logic I should have to deal with forms that I can invent? What I have to deal with must be that which makes it possible for me to invent them (5.555).176 Even though the logical constants represent nothing, there must be something that makes our use of them possible and is thus presupposed by it.

Wittgenstein went to some lengths telling us what that something is. It included the existence of objects, which are (1) the meaning (reference, Bedeutung) of names, (2) the values of the individual variables in general propositions, including the laws of logic, (3) the constituents of atomic facts (states of affairs), and thus (4) the substance of the world.177 It included the senses of elementary (atomic) propositions,178 the existence or nonexistence of the facts (states of affairs) such propositions assert,179 and thus their truth or falsity as well as truth or falsity of molecular propositions.180 It included the truthpossibilities of elementary propositions, the combinations of which are made explicit in the truth tables for the molecular propositions composed of elementary propositions. It
176

included the totality of the singular propositions that general propositions show but do not assert. It included formal concepts like those of being an object and being a fact. It included therefore also the totality of objects and the totality of facts, i.e., the world itself. All these can only be shown, they lack unqualified reality.

The central doctrine in the Tractatus was that what logic is about is inexpressible, cannot be said. This phrasing motivates the antirealist reading of the book, for example, Cora Diamonds181 and Warren Goldfarbs,182since one might reasonably conclude that what cannot be said lacks reality. But the Tractatus also tells us that what cannot be said nonetheless can be shown or, as Wittgenstein sometimes writes, e.g., in 5.5561 and 6.522, that it can show itself (zeigt sich). One might then no less reasonably conclude that what cannot be said but can be shown is part of reality. After all, it is not like Pegasus or phlogiston. It does show itself. This motivates the realist reading, for example David Pears183 and P. M. S. Hackers.184 Wittgenstein, however, avoided both antirealism and realism. He did write There are, indeed, things that cannot be put into words. They make themselves manifest [Es gibt allerding Unaussprechliches], but his next sentence was They are what is mystical [Dies 'zeigt' sich, es ist das Mystische] (6.522). These things are not of concern just to soft disciplines like ethics and religion. They include logical form and formal properties, which are constitutive of the subject matter of logic. We cannot say of a given item that it is an object, a complex, fact, function, number, or the total number of objects or of facts. Therefore, we cannot even say of the world that it is the world. But what we cannot say shows itself in what we can say. It is not nothing. Surely, the world is something. Surely, so is what Wittgenstein called the higher.

177

Yet, neither is it straightforwardly a part of reality. If it were, it would not be inexpressible.

Cora Diamond describes a realist interpretation of the Tractatus as a chickening-out,185 and Warren Goldfarb calls it irresolute.186 But I doubt they would say that logical form is like Pegasus or phlogiston. Wittgenstein is explicit that in some sense it is in reality [Wirklichkeit]: Propositions can represent the whole of reality, but they cannot represent what they must have in common with reality in order to be able to represent it--logical form (4.12), Propositions show the logical form of reality. They display it (4.121). Logical form can only be shown, not because it is nothing but because its presence in both language and reality is what makes any saying possible. Saying presupposes showing. Something is said only if something else is shown.

But while logical form is not like Pegasus or phlogiston, neither is it like Secretariat or oxygen. Although Wittgenstein was not an antirealist, he also was not a realist. This should not surprise us. In both his early and his later works he avoided simplistic distinctions: And yet you again and again reach the conclusion that the sensation itself is nothing. Not at all. It is not a something, but not a nothing either!187

The Tractatus began with statements about the world (Welt), although in 2.06 Wittgenstein used also reality (or actuality, Wirklichkeit).188 Had he provided an explicit inventory of what according to him is in the world, which he did not because the inventory would have employed formal concepts, it would have been sparse. At most, it
178

seems, it would have contained two categories: objects and facts. Moreover, it is not clear whether the category of objects includes both particulars and properties. Indeed, it is not clear that Wittgenstein allowed the distinction between particulars and properties, since his objects seemed to be incomplete or unsaturated, like Freges concepts: If I can imagine [denken, think] objects combined in states of affairs, I cannot imagine them excluded from the possibility of such combinations (2.0121).189 Nor is it clear that he allowed for molecular and general facts, rather than just for atomic facts (Sachverhalte), though he did seem to allow for negative facts.190 The remaining Fregean, Russellian, and Bergmannian categories, if he considered them at all, were assigned by Wittgenstein to what can only be shown, usually under the rubrics formal properties and logical form. Wittgenstein's distinction between saying and showing amounted to a rejection of standard realism, but it did not amount to acceptance of standard antirealism. In this respect, he was not unlike his distant precursor Kant, who described his position as transcendental idealism but also empirical realism.191

As I noted above, Wittgenstein's objects were not unlike Freges concepts.192 We cannot speak about concepts: The concept horse is not a concept, Frege wrote. What expresses a Fregean concept is a grammatical predicate, and the concept horse is not a predicate. The relevant predicate is is a horse, but it cannot serve as the grammatical subject of a well-formed sentence, including a sentence of the form x is a concept. Yet, of course, there is such a concept. Some objects are horses. Frege did not deny that there are concepts. He was an unmitigated realist. But neither did he explain their peculiar status. This is why his readers have found his assertion that the concept horse is not a concept confusing.
179

2. Ontological, cosmological, and logical antirealism. According to metaphysical realism, the existence and nature of things (res), reality, the world, is independent of our cognition of them, in the broad contemporary sense of cognition that includes perceiving, remembering, imagining, conceptualizing, and describing. Metaphysical antirealism denies this. I call both metaphysical in order to mark their supremely general status and distinguish them from realism and antirealism in ethics, aesthetics, the philosophy of science, the philosophy of mathematics, and other branches of philosophy, where realism and antirealism are restricted to a specific subject matter.

It may be useful to review certain remarks made in the Introduction. I stated the thesis of metaphysical antirealism as follows: The world, at least insofar as it is known or knowable by us, depends on our capacities and ways of knowing it, on our cognitive powers or faculties sense perception, introspection, intellectual intuition, imagination, memory, recognition, conceptualization, inductive and deductive reasoning, use of language and of other symbolism. The thesis does not involve equivocation like that in the sentence As king of England, John was subject to no one, which is true but compatible with the also true As duke of Aquitaine, John was subject to the king of France. We know that in addition to being king of England John was duke of Aquitaine, and this is why we also know that he was not subject to no one only as king of England. The thesis of antirealism does not say that the world depends on our capacities and ways of knowing. It says that it depends on them insofar as it is known or knowable. But its metaphysical bite is felt when we acknowledge the tautology that we know no world that
180

we do not know, though this tautology must not be confused with the proposition that there is no world, or anything in or about the world, that we do not know.

I sketched the standard argument for this thesis, in both the Kantian and its more recent versions, as follows: (1) We cognize only what we have the capacity to cognize. This is a tautology. Therefore, (2) there is no reality, no world, that is independent of our cognitive capacities. But (2) does not follow from (1). What follows is another tautology, (3) that we cannot cognize reality independently of our cognitive capacities. Contemporary antirealists often argue on the basis of (1) for (2), not for (3), probably because the negation of (2), Kants view (4) that there is a reality, things-in-themselves, which is independent of our cognitive capacities, seems to them idle. But there is a very good reason for (4), namely, that (2) implies a sort of cosmic humanism, human creationism, namely, the proposition that the world depends for its existence and nature on certain members of one of its planets fauna.

As we saw in the Introduction, metaphysical antirealism comes in many varieties, as different as Berkeleys unqualified idealism and Kants transcendental idealism, Hegels absolute idealism, Wittgensteins logical antirealism, and Heideggers phenomenology of being-in-the-world, as well as Michael Dummetts antirealism, Hilary Putnams internal or, later, pragmatic realism, and Nelson Goodmans irrealism in contemporary philosophy. Berkeley held that the existence of the things we perceive is dependent on our perception, Kant that their nature is dependent on both our perception and our understanding, and Wittgenstein, Heidegger, Dummett, Putnam, and Goodman that it is dependent also on our language.
181

In one form or another, metaphysical antirealism has dominated modern philosophy at least since Berkeley, even though it is metaphysical realism that remains the bedrock of everyday and scientific thinking. But, as we saw in the Introduction, numerous though seldom made distinctions within it are required. And the reasons for accepting it have seldom been stated in detail. Usually they have consisted in abstract and obscure generalities such as Nothing can be conceived that cannot be perceived, For knowledge to be possible, objects must conform to knowledge, or Thought without language is impossible, which are no less controversial than what they are reasons for.

In the Introduction I distinguished antirealism from semirealism, as well as from skepticism and relativism. I also distinguished the several varieties of antirealism: ontological, cosmological, perceptual, conceptual, linguistic, and logical. These distinctions may have seemed fussbudgety, but they are needed. One can be an antirealist but not a skeptic, and one can be a skeptic but not an antirealist. One can be an antirealist but not a relativist. One can be a cosmological but not an ontological antirealist. One can question the reality of logical objects such as facts, and thus of the world if it is conceived as the totality of facts, but not the reality of things, such as animals, vegetables, and minerals. Moreover, one may be neither a realist nor an antirealist regarding some items, e.g., those that subsist (Russell) or those that can only be shown (Wittgenstein), hence the need to acknowledge semirealism. Within both cosmological and ontological antirealism, we need the distinction between perceptual and conceptual antirealism. Since a concept may be a purely mental, psychological, and therefore subjective, or a linguistic, public, and therefore objective, item, we need the distinction between Kants
182

psychological version and the 20th century linguistic versions of conceptual realism. Further distinctions are needed within logical antirealism: one can accept a realist interpretation of singular statements but only a semirealist interpretation of compound and universal statements and, as we shall see, an antirealist interpretation of generic statements. Such distinctions are needed because each variety of metaphysical antirealism calls for different considerations.

The distinction between ontological and cosmological realism/antirealism may seem purely notional, an indulgence in technicalities for their own sake. But it is not. If, as Wittgenstein thought, the world is the totality of facts, then one who rejects the category of fact also rejects cosmological realism, realism with respect to the world, but not necessarily ontological realism, realism with respect to things. Moreover, even if the world is not the totality of facts, it certainly is not just the totality of things, a mere collection of unrelated items, a whole without structure. The realism/antirealism debate for too long has appeared to allow only two alternatives: all and nothing. It has been needlessly vague because such crucial distinctions have been neglected. One aim of this book is to show that the issue is far more complex than participants in the debate usually suppose.

I argued in the Introduction that a clear and plausible defense of antirealism should shun purely abstract arguments and appeal to specific and readily understandable truths. One such truth is that the expressions distinctive of logic play an essential role in all developed talk and thought. It is the truth on which logical antirealism builds its specific and not implausible defense of antirealism. The logical antirealist denies that there are
183

logical objects, entities corresponding to the logical expressions, the logical constants. Standard examples of such expressions are the sentential operators: not (~), and (), or (v), ifthen (); the quantifiers: all (), some ( ), and the verb to be in its senses of identity (=), existence (there is, exists), as well as of predication (is, are) even when the latter is expressed only by syntactic order. The verb to be is a logical expression in both Socrates is human and God is, though in standard logical notation the former is expressed not by a separate sign but by the juxtaposition of the subject and predicate terms (e.g., Hs, if H is short for is human and s for Socrates), and the latter is either ignored as ill-formed or translated into a statement employing the particular quantifier. Now, no abstract argument or obscure philosophical principle is needed to convince nonphilosophers as well as most philosophers that, though essential to any cognition above that of infants, logical expressions do not stand for anything in the world, that there is no all, or. is, or and. In this respect logical antirealism differs sharply from the usual versions of perceptual, conceptual, or linguistic antirealism.

Frege used the phrase logical objects for the objects of arithmetic in the context of his project of reducing arithmetic to logic, a project adopted later by Russell and Whitehead.193 Wittgenstein used it for the entities a logical realist may think are required for understanding the sentential connectives. But in the Introduction I applid the phrase also to facts, in Russells sense, including those that are general, negative, and molecular (or compound). Clearly, it is applicable to thoughts, in Freges sense. Both Russellian facts and Fregean thoughts are categories of entities accepted because of broadly logical considerations. Both were explicitly introduced as the category of entities
184

that declarative sentences stand for or express. I have accordingly counted declarative sentences as logical expressions, too, facts being the logical objects to which, if true, they supposedly correspond. The analysis of sentences with respect to logical form has been a central concern of logic.

The logical realist holds that at least some logical expressions correspond to objects, that there are logical objects. As we saw in the previous section, Frege, Russell (at one central stage of his philosophy), and Gustav Bergmann were its clearest proponents. Wittgenstein, in Tractatus Logico-Philosophicus and later works, seemed to be a logical antirealist, but we also saw that his position is better described as semirealism. It is logical antirealism, however, that had been characteristic, even if only tacitly, of earlier philosophy. For example, Kant noted that logic abstracts from all objects of knowledge and has itself alone and its form to deal with, and that the concern of logic is to give an exhaustive exposition and a strict proof of the formal rules of all thought.194 Thus Kant in effect acknowledged the chief thesis of logical antirealism: there are no logical objects even though logic is present in all thought. All thought, even when it does have objects, must conform to logic, but logic has no objects. This could have been an argument for transcendental idealism additional to those Kant explicitly offered, though he did not present it as such.

Logical antirealism leads directly to cosmological antirealism if the world is the totality of facts and fact is a logical expression. But since thought and talk about individual things also usually employ logical concepts and expressions, arguably logical antirealism also leads to ontological antirealism. This is especially evident in phenomenalist theories,
185

which analyze statements about individual material things as statements about all outcomes of certain possible experiences and thus as essentially involving generality. However, I shall not consider here whether logical antirealism leads to ontological antirealism.

An antirealism that is only logical and only cosmological is so moderate as to be virtually a part of common sense, of the thoughtful nonphilosophical judgment that all theorizing must respect. Common sense would rebel against the postulation of objects corresponding to the sentential connectives. It is satisfied by counting them as syncategorematic, mere ancillaries in assertions, incapable of use as subjects or as predicates, and thus standing for no entities. Common sense also could be easily convinced that sentences, the expressions Russell took to stand for facts, are syncategorematic. They function as neither names nor predicates. Facts were not on Aristotles or any other list of categories before Russell and Wittgenstein included them in theirs. It would occur only to philosophers that the sentential connectives and sentences themselves might stand for distinct entities in the world.

Only a few philosophers, most notably Russell, have also thought that, not only there are facts, but also there are general facts, which correspond to universal sentences, as well as molecular facts, which correspond to molecular, especially negative, sentences. Frege, who together with Russell was the target of Wittgensteins attack on logical realism, abhorred Russells category of fact (on the grounds that to say that p is a fact is to say no more than that it is true that p) but he did appeal to the category of objective entities he called thoughts, which are expressed by sentences. Frege also accepted the existence of
186

general and negative thoughts, as well as the existence of compound thoughts such as conjunctions, disjunctions, and conditionals. His vigorous arguments, as well as Russells, for the irreducibility of general, compound, and negative statements were, in effect, arguments for logical realism with respect to them.

In this chapter, in order to avoid confusion, I shall follow Freges, Russells, and Wittgensteins use of general as a synonym of universal, even though strictly speaking general statements include both universal and particular (existential) statements, as well as generic statements (to be discussed in chapter 8). All three philosophers took for granted that particular statements are to be understood as negations of negative universal statements (( x) x ~ (x) ~ x)), and thus to require no separate discussion. None thought that true generic statements stand for generic facts; indeed, they did not even consider them.

In the preface to Ways of Worldmaking, Nelson Goodman wrote, I think of this book as belonging to that mainstream of modern philosophy that began when Kant exchanged the structure of the world for the structure of the mind.195 Indeed, contemporary antirealism is best understood as the heir of Kants transcendental idealism. Kant argued that although there is reality as it is in itself (things-in-themselves), we can know reality only as it is for us (things-for-us). Not only our knowledge but all judgments, whether true or false, are shaped by our cognitive faculties, our senses and our concepts. We can no more get at reality as it is in itself, apart from us, than we can get outside our skins. Perhaps it is false that nothing unperceived or even unperceivable by us can be known by us. But the proposition that nothing unconceptualized or unconceptualizable by
187

us can enter in epistemic relations with our judgments is, in Putnams words if not sense, a virtual tautology. Only judgments can enter in epistemic relations to other judgments, and judgments necessarily involve concepts. To state the proposition as an explicit tautology, however, would require extensive accounts of the notions it involves, which would rely on similar and no less controversial other propositions. But there can be little doubt that the proposition is true. It should not be confused with the thesis of ordinary idealism. Kant did not hold, as Berkeley did, that everything is mental. Nor should the proposition be confused with the antirealist thesis that all reality is dependent for its existence or at least nature on us, which is hardly a tautology, not even a virtual one. There is no contradiction in thinking or speaking of things in themselves. This is why Kant described his view as transcendental idealism but also empirical realism. Contemporary antirealists are seldom sensitive to Kants distinction.

My central concern in this book is not to engage in the general and rather amorphous dispute between realism and the many varieties of antirealism but to appraise the much better defined version of metaphysical antirealism I called logical antirealism. It resembles Kants transcendental idealism but places the dependence of the world on our language rather than on our mental faculties, and even then it does so only with respect to the logical expressions in language, including sentences.

3. The logical structure of the world. The subject matter of metaphysics is said to be reality, being, what exists, or just the world, as in Goodmans Ways of Worldmaking. Its first question is, what is real, what kind of entities exist? Various answers have been given, e.g., that only material entities
188

exist (Hobbes), that only mental entities exist (Berkeley), or that in addition to material and mental entities also abstract entities (Plato) or God (Aquinas) exist. But however we answer the question, we take for granted that the world has a structure, that it is not a mere collection or assemblage of isolated items.

What kind of structure is fundamental, absolutely necessary, to the world? Not a causal structure: Hume rejected causal connections, except in the bland sense of spatiotemporal correlations. Not a physical or even just spatial structure: the dualist holds that in addition to material entities there are irreducibly nonphysical and nonspatial mental entities such as thoughts and feelings, and the idealist even holds that everything is mental; both deny that mental entities are in space: they lack geometrical characteristics, they do not enter in relations such as two-miles-from. Nor is a temporal structure absolutely necessary: the Platonist holds that there are abstract entities, e.g., universals and numbers, which not only are in neither space nor time they do not enter in relations such as two-yearsearlier-than and the theist holds that the spatiotemporal world and thus presumably of space and time themselves were created by God. The fundamental structure of the world, which though seldom mentioned or even considered is denied by no one, is logical. Astronomy, physics, chemistry, and biology focus on its causal, spatial, temporal, and physical structure. Philosophy, at least as it was understood by the founders of contemporary logic and analytic philosophy Frege, Russell, and Wittgenstein focuses on its logical structure. If the first question of metaphysics is what is real or exists, its second question is, what kind of structure the world must have? And surely the answer is that it must have a logical structure. Indeed, Aristotle held that the science of being qua

189

being, i.e., metaphysics, begins with the study of the principles of the syllogism, i.e., logic.196

What is meant by logical structure? The answer lies in what is meant by logic. Logic is concerned with the relations between sentences (propositions, statements, judgments) that hold in virtue of their logical form. And the logical form of a sentence is best understood as what is left when we completely abstract from its subject matter, what it is about. According to Russells canonical account, in respect to logical form sentences (propositions) are either atomic (e.g., Socrates is human), compound (e.g., Socrates is human and Plato is human), or general (e.g., All humans are mortal).197 Essential to discerning the logical form of a sentence are the logical expressions it contains, e.g., the italicized words in the previous sentence. To say that the world must have a logical structure entails that any adequate description of it must employ such expressions. It is a tautology that we have no conception of a world that is not, at least in principle, describable.

Nothing would count as a world if it did not allow for atomic statements, e.g., This page is white. Nothing would count as a world if it did not allow for compound statements, e.g., This page is not red, or If this page is white then it is not red. And nothing would count as a world if it did not allow for general statements, e.g., All winters in Minnesota are severe or There is water on Mars. It is natural therefore to see the logical structure of the world as providing room for atomic facts about the properties and relations of individual things; for compound facts, most notably, facts about what is not the case (negative facts), what is the case only if something else is the case (conditional
190

facts), two facts being both the case (conjunctive facts), and either one or the other of two facts being the case (disjunctive facts); and for general, either universal or particular, facts.

Nothing would be a world if it did not allow for saying of a thing what it is, that is, for predication, were there not in it something like a relation or nexus of instantiation or exemplification. Nothing would be a world if it did not allow for distinguishing one thing from another, if negation and identity were not in the world. Nothing would be a world if it did not allow for generalization, if generality were not in the world; as Bergmann argued in his posthumously published New Foundations of Ontology, there could be no laws of nature in a world lacking generality because the laws of nature are expressed in general statements.198 The logical realist would say that the world does contain such peculiar entities and therefore has a logical structure independently of our cognition of it, in particular, of our language. The notion of logical structure is thus understood in terms of the notion of a certain kind of linguistic expressions, those I have called logical. Does the world have a logical structure independently of our use of those expressions? The logical realist holds that it does. The logical antirealist disagrees, insisting that the world has a logical structure only because we describe it with a language that contains logical expressions.

Sentences are the vehicles of the description of the world. Logical expressions are parts of sentences. I have counted sentences themselves as logical expressions, and the facts that supposedly make them true as logical objects. The logical antirealist will point out, however, that a fact is indistinguishable from the sentence it supposedly makes true. It is
191

a complex entity supposed to have a structure or form that seems to image the structure or form of a sentence. For example, Russells requirement that an atomic fact contain particulars and properties or relations corresponds to the grammatical requirement that a simple sentence contain a subject and a predicate, and his distinction between atomic and molecular facts corresponds to the grammatical distinction between simple and compound sentences. Indeed, the former correspondence was explicitly acknowledged in Wittgensteins doctrine that sentences are logical pictures of facts. It is thus fairly clear that when describing the ontological characteristics of facts Russell and Wittgenstein relied on the grammatical characteristics of sentences. Of course, they were concerned not with the surface grammatical form of a sentence but with its logical form. But one arrives at the latter only by beginning with the former. Logic is concerned with what sometimes is called logical grammar, not with ordinary grammar, but we have little if any conception of logical grammar apart from our conception of ordinary grammar.

The central place of sentences was explicitly acknowledged in Freges explanation of his technical notion of thought (Gedanke). Even though he held that thoughts belong neither in the physical nor in the mental world, but rather in a third realm,199 Frege also wrote, The world of thoughts has a model in the world of sentences, expressions, words, signs. To the structure of the thought there corresponds the compounding of words into a sentence.200 [T]he structure of the sentence serves as an image of the structure of the thought.201 Russells and Wittgensteins heavy indebtedness to Frege in logical theory was freely acknowledged by both. Like Fregean thoughts, Russells and Wittgensteins facts also were understood as analogous to sentences:

192

Although the German word Gedanke is a synonym of the English thought, as used by Frege it has no ordinary translation in English. Proposition might be least misleading, as long as we think of a proposition as an objective item distinct both from the sentences in the various languages that express it and from the ideas and judgments we entertain about it. If so, there would be no clear difference between a true Fregean thought and a Russellian fact. This is why Frege wrote, What is a fact? A fact is a thought that is true.202 His view that thoughts are neither mental nor physical may seem mysterious, but the description would also apply to propositions understood in the way just suggested as well as to Russells and Wittgensteins facts. Neither propositions nor facts are things that can enter in physical relations, nor are they mental images or judgments. Neither the proposition nor the fact that this table is two feet from me is itself two feet or any other distance from me, and neither is mental, just as the table is not.

In logic we usually speak of propositions, sometimes of sentences, but not of facts. This is a symptom of our ambiguous conception of its subject matter, our uneasy attempt to straddle the perceived chasm between the logic of sentences, which is about language part of the behavior of certain members of our planets fauna and thus essentially a zoological subject matter and the logic of facts, which supposedly is about the world and thus, in Platos view, a more appropriate subject matter for spectators of all time and all existence. According to logical realism, the chasm is real. According to logical antirealism, it is an illusion. The logic of the world, the antirealist holds, is not distinguishable from the logic of words, and Russells or Wittgensteins facts, as well as Freges thoughts, are merely hypostatized sentences, shadows that sentences cast upon the world. I shall have much more to say about the topic of facts in chapter 10.
193

The general thesis of metaphysical antirealism now appears more plausible. What makes a world a world, rather than a congeries of things, is what requires sentences, rather than mere lists of names, for our description of it, namely, a logical structure. This requirement is even more basic than that a world must be logically possible, if the latter means that it must not involve a contradiction, for even a contradiction has logical structure, often as simple as that of p and not-p. But the only conception we have of logical structure is that of the logical structure of sentences. This is why in speaking of a world we are appealing to the category of facts. This is why Wittgenstein unhesitatingly wrote that the world is the totality of facts, not of things. Sentences, of course, are parts of language, and their logical structure is a feature of language. But language is something human, ours. We have no genuine conception of a language that is both nonhuman and in principle untranslatable into a human language. Therefore, insofar as we can conceive of the logical structure of the world and thus of the world itself as a world, they themselves are ours, human, they depend on us. Of course, that the only conception of logical structure we have is that of sentences, of our language, does not entail that the world does not have that, or some other kind of, structure independently of language. But it seems to be a good reason for reaching that conclusion. For it does entail that our cognition of the world, insofar as it involves logical concepts and employs logical expressions, depends on language.

Many twentieth century philosophers have held that all cognition, not just that involving logical concepts, is dependent on language. Earlier I cited Wittgensteins assertion that The limits of my language mean the limits of my world, Heideggers that language is
194

the house of being and that language is the happening in which for man beings first disclose themselves to him each time as beings, and Quines that thought, as John B. Watson claimed, is primarily incipient speech. These opinions exemplify the linguistic turn philosophy took in the twentieth century. It was more fundamental than its earlier turns, such as the Platonic turn to abstract entities, the theological turn in the Middle Ages, and the idealist turn in the 18th century. For it applies to everything we think is real, including abstract entities, God, and minds. It was a turn to linguistic antirealism. But the linguistic antirealism I have sketched is not as extreme as that implied by the quotations from Wittgenstein, Heidegger, and Quine. It does not indiscriminately, as matter of general principle, hold that the world as we cognize it, as it is for us, depends on language. It restricts this dependence to logical expressions. It is only a logical antirealism, even though its implications are far-reaching. It does not hold that all cognition depends on language. It avoids the highhandedness, often characteristic of philosophy, that the quotations in the previous paragraph seem to exemplify.

It is thus a limited, moderate, and for this reason more plausible antirealism. It does not offer a general and a priori answer to the question whether cognition depends on language. This question, as we saw in chapter 5, is mainly empirical, to be answered by neuroscientists, psychologists, and linguists. And the proper answer may well vary according to what kind of objects of cognition we have in mind. Rocks are different from headaches, both are different from electrons, and all three are very different from numbers. It would be rash to suppose that what is true of our cognition of some is true of our cognition of all. There are also many, very different kinds of cognition. Some may depend on language while others do not. Surely, there is cognition in the form of sense
195

perception, enjoyed by infants and nonhuman animals, which does not involve language. Recognition is a fundamental higher form of cognition, and it too is found in infants and nonhuman animals. Driving a car and professional boxing involve specialized cognition that only superficially, if at all, finds or needs expression in language. Creative work in music or painting only minimally involves talking. Worldly people, especially in the law, politics, and diplomacy, rely heavily on reading facial expressions, which language notoriously fails if we try to describe. We achieve greater specificity and a sharper focus when we ask, Does logical cognition have a distinctive subject matter, is it, at least in part, about anything distinctively logical? If not, then to that extent logical cognition would appear to be nothing but use of language, since it would lack the nonlinguistic feature essential to other kinds of cognition: a subject matter, things they are about that make them, at least in part, the cognitions they are. Seeing a cat differs from seeing a dog because cats differ from dogs. Seeing a cat differs from hearing a cat partly because sounds differ from colors and shapes. It would not follow, however, that all there is to logical cognition is language. I used the phrase distinctive subject matter precisely because, outside logic, sentences containing logical expressions do have a subject matter, namely, what is denoted by the nonlogical expressions (the descriptive words) that they also contain. But our question is whether such sentences have a subject matter insofar as they are logical. The world may have a logical structure that is supplied by language, but it does not have only a logical structure: it also contains the entities that are structured. The tendency to forget this may explain why philosophers seldom see that logical antirealism is a version of metaphysical antirealism. They say, for example, Of course All humans are mortal is about the world it is about humans. They overlook the fact that while what the
196

sentence says depends on more than the logical expressions all and are, it does depend on them. Its truth-value depends on both the nonlogical and the logical expressions in it.

In contrast, the statements of the discipline called logic employ only logical expressions. If logical expressions stand for nothing, then the statements of logic have no extralinguistic significance a consequence commonly, however rashly, accepted today. But all human cognition beyond that of babes, not just the cognition expressible in statements made by logicians, must be expressible with statements that employ expressions such as the verb to be (especially in its sense of predication), the adverb not, sentential connectives like or, or the quantifiers all and some, even if, unlike the statements of logic, they also employ nonlogical, descriptive, expressions. All developed human cognition is logical, in this broad sense. All statements, not just the statements of logic, employ logical expressions. A language that lacked them would be either in principle untranslatable or as primitive as baby talk. Therefore, no developed human cognition is possible without language. This does not entail but is a good reason for holding that the world, insofar as it is the object at least of developed cognition, cannot have a character fundamentally different from what we humans take it to have.

What I mean by developed cognition need not require higher education it would include the cognition expressed by All sheep eat grass. If a shepherd did not know that all sheep eat grass, there would be precious little the shepherd knew. If we could not say that all sheep eat grass, there would be precious little we could say. In chapter 9 we shall see
197

that we are more likely to make the generic statement Sheep eat grass, and the antirealist implications of generality would be evident. For now, I shall remain within the familiar context of universal statements. In that context, while the words sheep, eat, and grass do stand for things in the world, all clearly does not. There is no such entity in the world as all, not even among things-in-themselves. All would not be an individual thing, nor a property or relation of an individual thing. This is why the phrase entity in the world as all, which I just used, is a grammatical monstrosity that only a philosopher would concoct. But it follows from this that there are also no distinctive, general or universal, facts, to which statements making essential use of all might correspond. The sentence All sheep eat grass stands for nothing distinctive in the world.

If we take advantage of the familiarity of Kants term transcendental, we may say that logical concepts and expressions are transcendental. They are essential at least to any developed cognition of the world, but stand for nothing in the world. This would be a way of stating the thesis of logical antirealism. It would hawould be news. Philosophers and grammarians have commonly called the logical expressions syncategorematic. But they offered little explanation or argument, and few saw the metaphysical implications.

4. Frege and Russell on negation and generality. In this chapter I shall take the words not and all as paradigms of logical expressions. They are fundamental not only to logic but all developed cognition. They express, respectively, the logical concepts of negation and generality (more accurately,
198

universality). I shall offer here a preliminary discussion of both. Chapter 7 will be devoted to a further, detailed discussion of generality, which is of special concern in cosmological antirealism. The thesis of logical antirealism would be sufficiently defended if established with respect to these two notions. As our realist foil, let us take Freges and Russells classic discussions. It is to those discussions that Wittgensteins fundamental thought that the logical constants do not represent was opposed.

Regarding generality, Frege wrote, It is surely clear that when anyone uses the sentence all men are mortal he does not want to assert something about some Chief Akpanya, of whom perhaps he has never heard.203 Frege classified what all expresses as an unsaturated entity that is a second-level function, saturated by first-level functions like those expressed by x is a man and x is mortal, which in turn are saturated by individual objects such as Chief Akpanya.204 A function is unsaturated, or incomplete, in that it requires attachment to something else, very much as the grammatically incomplete expression is human requires attachment to a grammatically complete expression such as Chief Akpanya, an analogy to which Frege often appealed. For example, in the thesis of materialism, (x)(x is material), i. e., Everything is material, the universal quantifier x represents a second-level function, and x is material represents the first-level function that completes it. The latter is also incomplete, but what completes it is an individual thing, say, this page. Freges terminology is awkward, but what matters for our purposes here is his explicit acknowledgement that there is an entity to which the word all corresponds. To be sure, it is an incomplete entity, but so is also the property of being human. Frege called a first-level function a concept (Begriff), but like many of his contemporaries he did not
199

mean by concept anything mental. He meant what property means in current philosophy.

Frege also held that there are negative thoughts, in his technical sense of thought, since for every thought there is a contradictory thought which appears as made up of that thought and negation, though not as mutually independent parts. The thought does not, by its make-up, stand in any need of completion; it is self-sufficient. Negation on the other hand needs to be completed by a thought. The two components arequite different in kind.One completes, the other is completed. And it is by this completion that the whole is kept together.205 Negation is incomplete, just as the second-level function expressed by all is incomplete. The difference is that negation is completed by a complete entity, viz. a thought, while a second-level function is completed by an incomplete entity, a first-level function. But negation is not something mental or subjective. It is not the act of denial.206 Nor is it a kind of judging: Negation does not belong to the act of judging, but is a constituent of a thought.207 It is an objective part of a no less objective entity, viz. a negative thought. The sentence It is not the case that Socrates is feline consists of two parts: it is not the case that and Socrates is feline. The later could stand by itself, the former could not. But what the whole sentence says is no more psychological or subjective than what Socrates is feline says. Russell too was a realist about generality and negation, though he argued for the existence of general and negative facts, not Fregean thoughts. Regarding generality, he wrote, When you have taken all the particular men that there are, and found each one of them severally to be mortal, it is definitely a new fact that all men are mortal. For, In order to arrive [by complete induction] at the general statement All men are mortal,
200

you must already have the general statement All men are among those I have enumerated. True general statements, such as All men are mortal, stand for general facts. Therefore, there are general facts. Moreover, because You cannot ever arrive at a general fact by inference from particular facts, however numerous, there must be primitive knowledge of some general facts.208 There is therefore the necessity of admitting general facts, i.e., facts about all or some of a collection.209

Russell also argued that there are negative facts and that negativeness is an ultimate.210 He wrote, There is implanted in the human breast an almost unquenchable desire to find some way of avoiding the admission that negative facts are as ultimate as those that are positiveUsually it is said that, when we deny something, we are really asserting something else which is incompatible with what we deny. If we say roses are not blue, we mean roses are white or red or yellow. But such a view will not bear a moments scrutiny.... The only reason we can deny the table is square by the table is round is that what is round is not square. And this has to be a fact, though just as negative as the fact that this table is not square.211

In discussions of our topic one often finds, in addition to logical realism and logical antirealism, the supposedly third option of logical reductionism. For example, in the case of universal statements, the realist holds that the universal quantifier x, or the word all, stands for a real entity, whether a Fregean second-level function or some other kind of logical object. The reductionist, finding the reality of such an entity implausible, as Wittgenstein almost immediately did upon meeting Russell in 1911, usually translates universal statements as conjunctions of the singular statements that instantiate them, and
201

claims that this is what they really say. But conjunction, the symbol or the word and, is also a logical expression and the reality of any logical object it stands for is no less dubious. The logcal antirealist denies that there are logical objects of any sort.

Sometimes the reductionist appeals implicitly to the very same concepts that are to be explained, and thus is no longer properly called reductionist. This is case with the claim that the sentential connectives are defined by the corresponding truth-tables and that this is all there is to what they mean. For example, not-p is said to be merely a truth function of p, on the grounds that not-p is true if and only if p is false. But we have no grasp of falsity except as the negation of truth, regardless of what theory of truth we hold, and thus the reductionist is implicitly appealing to negation when offering that explanation of negation. As Russell remarked, it is extremely difficult to say what exactly happens when you make a positive assertion that is false, unless you are going to admit negative facts.212

Another glaring example of such reductionism is Rudolf Carnaps claim that x (x) asserts the property of universality of the property , and x (x) asserts the property of nonemptiness to it.213 What are these properties of universality and nonemptiness if not what the quantifiers x and x respectively express? Presumably, Carnap followed the account of generality Russell had offered in Principia Mathematica: Let us call the sort of truth which is applicable to a first truthConsider now the statement (x). x. If this has truth of the sort appropriate to it, that will mean that every value of

x has first truth. Thus, if we call the sort of truth applicable to (x). x second truth,
we may define [(x). x] as second truth as meaning every value for x has first
202

truth, i.e., (x) x has first truth. Similarly, if we denote by ( x.) x the statement x sometimes, i.e., as we may less accurately express it, x with some value of x, we find that ( x). x has second truth if there is an x with which x has first truth.214 Russell needed this distinction between two truths for his theory of types, but the ontological idea behind it had already been proposed in his earlier article On Denoting, where he claimed that a universal statement such as (x) x means that x is always true, while ( x) x means that x is sometimes true. He pointed out repeatedly that always true and sometimes true express primitive, indefinable notions. But, if so, appeal to them can support neither the reductionist nor the antirealist thesis.

Elsewhere, reductionism is little more than a front for antirealism (for example, the reduction of mental states to brain states is little more than the denial of the reality of mental states). But, as we just saw, in the case of logic it seems to merely replace some logical statements with other logical statements, e.g., universal statements with conjunctions of singular statements even though a realist view of the sentential operators, including conjunction, is no more plausible than a realist view of the quantifiers.

This is often overlooked because of the common assumption that necessary equivalence is identity. If one is more comfortable with one of the sides of the equivalence, one is tempted to suppose that what it says is really nothing more than what the other side says. But, clearly, this is a mistake. It is necessarily the case that p if and only if p or p, but it would be silly, or at lest unmotivated, to speak of reducing either one to the other, of saying for example that what p says is really nothing more than what p or p says. In mathematics, to use Kants example, it is necessarily the case that 7+5=12, but there is no
203

clear and useful sense of reduction in which either side of the identity sign may be said to be reducible to the other. Hans Hahn was probably alone in thinking that 7+5 and 12 are synonyms by convention!

Freges and Russells views on generality and negation were explicit espousals of both logical realism and logical nonreductionism. Russells general and negative facts are considered to be objective constituents of the world, and they are not reducible to any other kinds of facts. Therefore, what makes them general or negative must also be in the world, though as we saw earlier Russell vacillated about this. Frege, however, was explicit: general thoughts involve a second-level function and negative thoughts include negation as one of their two components. Surely, Russells general and negative facts must also contain components represented, respectively, by the quantifier and the negation sign. If they did not, how would general facts differ from the corresponding singular facts and how would negative facts differ from the corresponding positive facts? If there is an item in the world represented by x (x) as well as an item represented by a, as Russell held, how would the two differ if not in virtue of something in one that is not in the other, presumably the presence in x (x) of what x represents and the absence from it of what a represents? Similarly, if there is an item in the world represented by not-p, as well as an item represented by p, how would they differ if not in virtue of something in one of them that is not in the other, presumably what not represents? For, according to Russell, the differences must be in the facts, in the world, not in our language or our minds. According to Frege, the presence of a secondlevel function in general thoughts is essential to the truth-values of general sentences and the presence of negation in negative thoughts is essential to the truth-values of negative
204

sentences. Surely, the presence of what is expressed by the quantifier in Russells general facts is essential to the truth-values of general sentences, and the presence of negation in Russells negative facts is essential to the truth-values of negative sentences. If, as Russell held, the truth of a statement depends on what is in the world, then what is essential to its truth must also be in the world.

The logical realist faces two tasks. The first, perhaps accomplished by Frege and Russell, is to refute logical reductionism, the view that upon analysis the question of the reality of logical objects does not arise because the logical constants that appear to stand for them have been analyzed away. By showing that at least in the case of negation and generality this is not so, Freges and Russells logical nonreductionism forces also antirealists to shun the comfort of slogans like A negative statement is just the corresponding false affirmative statement, Universal statements are just the conjunctions, and particular statements the disjunctions, of the singular statements that instantiate them, or Molecular statements are just truth-functions of their components. The second task the logical realist faces, however, is to convince us that, since logical reductionism is false, the logical constants do represent entities in the world. The plausibility of Freges and Russells views attached to their nonreductionism, not to their realism, which remained quite implausible.

Indeed, the idea that all and some correspond to entities has seldom been entertained. As we shall see in chapter 7, Gustav Bergmann, one of the few philosophers who thought deeply about this topic, did write: Each quantifier represents something which is sometimes presented. Had it never been presented, we would not know what the
205

quantifier meant.215 To respond to Bergmanns argument by saying that he relies on a naive conception of meaning would only display ignorance of his works. Nevertheless, it is an (abductive) argument for a statement of phenomenological observation, a statement about what is presented, not a report of that phenomenological observation, which is what it should have been, if Bergmann were right. And surely such a report would have been false. But Bergmann at least was aware of what would be necessary if logical realism is to be defended. We shall see in chapter 7 that so was Wittgenstein, and that in his later work Bergmann revised his uncompromisingly realist stand.

Freges and Russells arguments against reductionism in the case of negative sentences may be accepted, but antirealism about negation would remain inescapable.216 That negation corresponds to nothing in the world seems almost a tautology. Sartre claimed that it is consciousness that introduces negation into the world precisely because he believed that negation is not already in the world.217 The words no and not are learned early in childhood to signal the absence or nonexistence of a thing, to refuse an object offered by someone or to reject what one is told, and later to deny the truth of what one hears or reads. But they do not stand for the actions of refusing, rejecting, or denying, nor for any entity such as absence or nonexistence.

In denying that logical expressions stand for entities, the logical antirealist is not just denying a simplistic Fido Fido principle, according to which every word is a name. The antirealist is denying the natural, not at all simplistic though ultimately mistaken, assumption that if a word is to serve a cognitive role then it must relate to something in the world in a specfici, distinct, and comprehensible way, even if it does not name it, that
206

there must be something in what is cognized that grounds its cognitive role even if it is not named by the word, and that this something is accessible to us, if not directly in perception, as colors and books seem to be, then indirectly in sophisticated thought, as quarks, relations, and perhaps God seem to be. The logical antirealist claims that in the case of logical expressions none of this is true.

This claim should not be confused with the much weaker claim, to which chapter 5 was devoted, that language is causally necessary for logical cognition. Presumably, a human being who lacks a language cannot have detailed knowledge of astronomy or of medieval history. But surely God can, and for all we know so can intelligent forms of life outside our solar system. We may not understand what their knowledge would be like, but then neither can we visualize a non-Euclidean space or have auditory images of the highfrequency sounds that dogs but not humans can hear. The same may be said about humdrum cognitions like that expressed in an inventory of the thousands of chairs in a very large building. We cannot make the inventory without using language or other symbols. But, surely, a god could. Could the thought expressed in a universal statement like All my toys are upstairs be entertained by a child before learning the word all? Perhaps it could, if it were just a collection of several singular thoughts, e. g., that the doll is upstairs, that the ball is upstairs, and that the whistle is upstairs. But a mere collection of thoughts is not even what a conjunctive statement, a statement requiring the sentential connective and, expresses. And even if it were, it certainly is not what the universal statement, the statement employing all, expresses. A universal statement is not equivalent to the conjunction of the singular statements that are its instantiations, it is not made true only by them. As Russell would have pointed out, the conjunction would
207

have to include as additional conjunct the universal statement These are all the toys I have. A universal statement is made true not just by the atomic facts corresponding to its singular instantiations, but also by the further fact about the world that those are all the [relevant] atomic facts...[which] is just as much an objective fact about the world as any of them are.218

Would even God be capable of cognizing such a fact without using a language? God would know all the individual things there are, perhaps an infinity of them, without employing a language, but what would it be for God to know that they are all the individual things there are? Gods vision of all of them would not be enough. The thought that none has been omitted would also be needed. In our case, it seems, that thought could only be linguistic. To attribute to God literal use of language would be blasphemy, though an analogical attribution might be theologically defensible. In any case, what is at issue now is logical, not just causal, possibility.

The linguisticism of the position we have reached so far appears obvious. It does fit the usual formulations of linguistic antirealism, even though these formulations make no distinction between ontological and cosmological realism/antirealism, much less between metaphysical and logical realism/antirealism, or between antirealism and semirealism. For example, Michael Dummett writes, [U]ntil we have achieved an understanding of our language, in terms of which we apprehend the world, and without which, therefore, there is for us no world, so long will our understanding of everything else be imperfect.219 To attempt to strip thought of its linguistic clothing, he also says, is to

208

confuse it with its subjective inner accompaniments.220 Dummett may be right, but surely his judgement is rashly abstract.

Of course, cognition of the world involves also confrontation with nonlinguistic items in sense perception, introspection, perhaps also intellectual and even mystical intuition. Such confrontation need not involve language. A neonates exercise of normal sight or touch does not. We need not deny the independent reality of the items thus confronted. Thus we need not deny that we have cognition that is unmediated by language, or that its objects exist independently of that cognition. But it would be cognition at its most primitive level. Since it would not involve statements, it could not even appear to have facts as its objects and therefore, if the world is the totality of facts, not of things, to count as cognition of a world.

Chapter Seven: LOGICAL SEMIREALISM

1. Ineffability. In the previous chapter I offered a preliminary account of logical antirealism with respect to negation and generality. One need not be an antirealist to doubt the existence of such entities. Distaste for them has led to reductionism, the attempt to show that they are not really what they seem to be, which are often made in the context of the realism/antirealism debate. Reductionism is far less plausible in the case of not than in the case of sentential connectives like and, or, and only if. Analytic philosophers have provided little detailed discussion of negation, leaving that task to phenomenologists
209

like Heidegger and Sartre. But they have had much to say about generality. I shall focus on generality in this book, because of its obvious, direct, relevance to cosmological antirealism, one of our chief concerns in this book. Some is commonly understood in terms of not and all, but all seems to resists reduction. In the present chapter I shall consider in detail the especially noteworthy views of Wittgenstein and Gustav Bergmann on generality.

Wittgenstein usually called universal statements just general, as Russell had done. Sometimes, so did Bergmann. In discussing their views, so will I occasionally. All three took for granted that particular statements are equivalent to negations of negative universal statements (( x) x ~ (x) ~ x)), and thus to need no separate discussion, though in his early work Bergmann vacillated on this point. None considered the status of generic statements, statements like Patients with prior strokes benefit from taking Lipitor, the ubiquity of which in thought and talk provides the case for antirealism with decisive support. Generic statements will be the topic of chapter 8. The present chapter is devoted to universal statements, statements like All patients with prior strokes benefit from taking Lipitor.

The logical expressions for generality are of special relevance to cosmological antirealism, antirealism about the world. As Bergmann put it, there could not be any laws of nature if generality were not in the world.221 Logical antirealism regarding the other logical expressions may lead to ontological antirealism as well, antirealism about things. But I shall not consider this possibility here. In this book I attempt to present a version of antirealism that is measured, discerning, and as cautious and uncontroversial as possible.
210

Cosmological antirealism is such a version. It questions only the reality of the world understood as the totality of facts, not of things. It would not question the reality of the world if the world were understood as the totality of things. Cosmological antirealism does not question the reality of animals, vegetables, or minerals, and thus it remains safely in accord with common sense, which would find Wittgensteins notion of the word as the totality of facts, not of things, too philosophical to fathom.

According to Bergmann, the universal and the particular quantifiers stand for items with which we are sometimes presented, i.e., of which we are sometimes aware. Few have agreed with him. Few have agreed even with Russells weaker claim that at least there are general facts, of some of which we have primitive knowledge. Bergmann modified his view later, moving closer to the position Wittgenstein had held. Though adamant in opposing Russells logical realism, Wittgenstein did not merely deny that there are general facts. He also held a positive view on the topic, which Bergmanns later view resembled in important respects. Both represented, with unparalleled sophistication, the middle ground between simplistic realism and simplistic antirealism that I have called semirealism. A serious discussion of general statements must begin with what Wittgenstein and Bergmann said about them.

Of course, general statements had been the chief subject matter of logic since Aristotles syllogistic. But when Frege invented modern quantification theory, they became also a fundamental concern of metaphysics. Contemporary logicians seldom ask what, if anything, general statements correspond to in the world. But the founders of analytic philosophy, Frege and Russell, did, and the question was a major theme in Wittgensteins
211

early (pre-1929) and Gustav Bergmanns later (post-1959) works. All five Aristotle, Frege, Russell, Wittgenstein, and Bergmann were aware that there could be no genuine world if there were not generality in the world. This is why all five also saw the topics tie to what later became known as the realism/antirealism issue, even though for Aristotle generality involved universals, for Frege general thoughts, and for Russell, Wittgenstein, and Bergmann general facts or complexes.

Frege held that general statements express the saturation of second-level functions by first-level functions; Russell, that if true they stand for general facts; Wittgenstein, that they involve matters that can only be shown, not said; and Bergmann, initially, that they involve the entities generality and existence, for which the universal and the particular quantifiers respectively stand. As we saw in the previous chapter, Frege and Russell rejected the facile reductionist view of universal statements as the disguised conjunctions, and of particular statements as the disguised disjunctions, of their singular instances. So did Bergmann. In his early article Generality and Existence he wrote, What can be said with the quantifiers cannot be said without them.Consider (1) (x)G(x) and (2) G(a1). G(a2) G(aN). (1) implies (2). (2) does not imply (1).222 In New Foundations of Ontology he took this to be obvious, remarking that [(x) 1(x)] is not a conjunction, either finite or infinite, nor even analytically equivalent to one. Similarly, for [( x) 1(x)] and disjunction.223 In effect, the rule of universal instantiation is not a rule of equivalence

Bergmann argued that, like individuality, universality, and exemplification, generality and existence, which he took the quantifiers in universal and particular (existential)
212

statements to stand for, belong to the worlds form. One is presented with them, he wrote, but they do not exist rather, they subsist. Thus even in his early works Bergmann was inclined only toward semirealism regarding generality, not antirealism but also not realism. In that article, he used existence in two senses: the sense of some, which the particular quantifier expresses, and the sense of exists, in which the worlds form, as well as, say, Pegasus and the golden mountain, do not exist. Later, he expressed regret over the ambiguity. It is absent from New Foundations of Ontology, where Bergmanns views received, with remarkable subtlety, depth, and breadth, their most developed and detailed formulation.

Generality and Existence had been preceded by Ineffability, Ontology, and Method.224 Bergmann described the two articles as materially one. The first topic of Ineffability, Ontology, and Method was the ineffability of individuality, generality, and exemplification. He wrote, When I know that this is a green spot, I know also that (1) the spot is an individual, (2) the color is a character, and (3) the former exemplifies the latter (and not, perhaps, the latter the former). How could I know all this if it were not, in some sense, presented to me?225 But what was thus presented could not be also represented, at least not without futility. For, Looking at a nameI knoweven if I do not know which thing it has been attached to as a labelthe kind of thing, whether individual or character, to which it has been or could be attached.226

Bergmann went on to say that a certain name is on the lips of every likely reader, but that he would not mention it because he did not on this occasion wish to make assertions about the reading of a notoriously difficult text.227 The name of course was
213

Wittgensteins, and the text was Tractatus Logico-Philosophicus. Wittgenstein had written, In a certain sense we can talk about formal properties of objects and states of affairs It is impossible, however, to assert by means of propositions that such internal properties and relations obtain: rather, this makes itself manifest in the propositions that represent the relevant states of affairs and are concerned with the relevant objects (4.122). If I am to know an object, though I need not know its external properties, I must know all its internal properties (2.01231.) By external property Wittgenstein meant what Bergmann meant by character and other philosophers usually mean by property. By internal property he meant what he also called a formal property, e.g., that of being an object. We can now talk about formal concepts, in the same sense that we speak of formal properties..When something falls under a formal concept as one of its objects, this cannot be expressed by means of a statement. Instead, it is shown in the very sign for this object (4.126). Statements about an object say what external properties it has. Formal properties, however, Wittgenstein held, cannot be properly predicated, said, but they can be shown.

The similarity of Bergmanns views in Ineffability, Ontology, and Method and Generality and Existence to Wittgensteins views in the Tractatus was obvious, as Bergmann readily acknowledged. It involved, however tacitly, Wittgensteins distinction between saying and showing, which Wittgenstein later called the main contention in the Tractatus, since it concerned the cardinal problem of philosophy. 228 In chapter 4 I noted that some interpreters, for example, Cora Diamond and Warren Goldfarb, deny that according to the Tractatus there is anything that can be shown but cannot be said. However, Wittgenstein did write: There are, indeed, things that cannot be put into
214

words. They make themselves manifest. They are what is mystical [Es gibt allerding Unaussprechliches. Dies 'zeigt' sich, es ist das Mystische].229 Moreover, he held that what only shows itself is of utmost importance, that it is the higher (das Hhere). Surely, he did not think that the higher is nothing.

To understand Wittgensteins distinction between saying and showing and its role in the Tractatus we must take seriously its application not only to logic, but also to ethics and religion. To say that Socrates is an individual, rather than, say, a property or a relation, is not to make a remark about Socratess wealth of properties, but neither is it to say nothing. To speak of the sense of life is not like speaking of the length of human life, but it is hardly like speaking of nothing.230 To be told that God does not reveal himself in the world, because how things are in the world is a matter of complete indifference for what is higher,231 may depress us but it does not tell us nothing.

The reception of Wittgensteins and of Bergmanns views in the philosophical community were similar, perhaps because both dealt with metaphysical questions that few philosophers had even considered, and both offered answers of which no previous philosophers had been aware. Critics of Bergmann complained that his philosophy is a Meinongian jungle, or they just avowed that they found it too difficult. Critics of Wittgensteins Tractatus disparaged it as too metaphysical, or they just interpreted it in terms of the much later Philosophical Investigations as Bergmann sarcastically put it, they found misery in Wittgensteins glory, and glory in Wittgensteins misery.232

215

What only shows itself is not nothing. It does show itself. One may ask, If it is not nothing, then what is it, what kind of being or reality does it have, what is its ontological status? Natural though this question may be, it presupposes the sharp and unilluminating distinction between what is real and what is not real, which Wittgenstein attempted to bypass with his distinction between saying and showing. I have already discussed it in other contexts, and here I shall consider the details of how it affected his account of generality.

2.

Wittgenstein on Generality.

In Tractatus 5, Wittgenstein wrote that A proposition is a truth-function of elementary propositions. (An elementary proposition is a truth-function of itself). He had explained that The simplest kind of proposition, an elementary proposition, asserts the existence of a state of affairs (4.21), and that It is obvious that the analysis of propositions must bring us to elementary propositions (4.221). (In the Introduction to the Second Edition of Principia Mathematica, Russell wrote that Atomic and molecular propositions together are elementary propositions.233) It seems to follow that a general proposition is also a truth-function, presumably of its singular (elementary) instances. And so, in a letter to Wittgenstein written in 1919, Russell objected: [In an account of general (universal) propositions in terms of elementary propositions,] it is necessary also to be given the proposition that all elementary prop[ositions] are given.234 As we saw in the previous chapter, Russell had repeatedly argued that, for example, the universal proposition All men are mortal is not equivalent to Socrates is a man and Plato is a man and Aristotle is a man and..., unless another universal proposition such as All men have been enumerated or These are all the men there are is added to the
216

conjunction.235 If All men are mortal is translated as (x) (if x is a man then x is mortal), which seems to say that every object is such that if it is a man then it is mortal, Russells objection would be that it is not equivalent to the conjunction of all singular statements of the form if x is a man then x is mortal unless the proposition that all objects have been enumerated or taken into account is added to the conjunction. For example, the thesis of materialism, (x) (x is material), would not be equivalent to the conjunction of all statements of the form x is material, unless we added that we have considered all objects, all the values of the variable x.

Wittgensteins reply to Russells objection was vehement. He did not deny Russells logical point, which amounted to reminding that the rule of universal instantiation is not a rule of equivalence. Rather, Wittgenstein wrote, There is no such proposition [as that all elementary propositions are given]! That all elementary propositions are given is shown by there being none having an elementary sense which is not given And he proceeded to add: Im afraid you [Russell] havent really got hold of my main contention, to which the whole business of logical prop[osition]s is only a corollary. The main point is the theory of what can be expressed [gesagt, said] by prop[osition]s i.e., by language (and, which comes to the same, what can be thought) and what can not be expressed by prop[osition]s, but only shown (gezeight); which, I believe, is the cardinal problem of philosophy.236

Wittgenstein in effect pointed out the implications of his distinction between saying and showing for the notion of a totality of propositions. As we saw in chapter 4, the distinction is especially clear when applied to formal or internal properties, such as being
217

an object. These properties can only be shown. This is why we cannot speak of the totalities determined by them: it is nonsensical to speak of the total number of objects. The same applies to the words 'complex', 'fact', 'function', 'number', etc. They all signify formal concepts (4.1272). Wittgensteins list of formal concepts ends with etc. Had it been expanded, it would certainly have included also proposition, thus implying that we also cannot speak of the totality of propositions. If fact is a formal concept, surely so is proposition, and therefore also elementary proposition, since propositions are logical pictures of facts (4.01) and elementary propositions are the simplest kind of proposition, namely, those that assert the existence of atomic facts (4.21). Wittgenstein went on to say, It is obvious that the analysis of propositions must bring us to elementary propositions (4.221). He had written that a logical picture of facts is a thought (3), that in a proposition a thought finds an expression that can be perceived by the senses (3.1), and that what constitutes a propositional sign is that in its elements (the words) stand in a determinate relation to one another. A propositional sign is a fact (3.14).

There is no such proposition as Russells all elementary propositions are given, not because (as one might suppose) the term given is too obscure, but because the alleged proposition would say that everything that is a proposition and is not composed of propositions (i.e., is an elementary proposition) is given and proposition is a formal concept, like object, fact, or function. In his objection, Russell seemed to take for granted what later was called the substitutional interpretation of quantification, according to which, roughly, general statements refer to all the elementary propositions that are their substitution instances. According to the alternative, more intuitive,
218

objectual interpretation of quantification, general statements refer to all values of the individual variable they contain, that is, all objects.237 Whether the two interpretations in fact require such reference is a debatable question, but we need not consider it here. What matters is that if Russell had taken for granted the objectual interpretation, his objection would have been that the proposition all objects are given must be given, and Wittgenstein would have replied that there is no such proposition because object signifies a formal concept. Had the issue been the analysis of a universal fact as the conjunction of all the atomic facts that instantiate it, Russells objection would have been that it would be necessary to be given the proposition that all atomic facts are given. And Wittgenstein presumably would have replied: There is no such proposition! That all atomic facts are given is shown by there being no atomic fact which is not given. His reason would have been that fact, and therefore also atomic fact, is a formal concept.

This has to be understood with some care. It does not mean that there are no universal propositions. What is meant is that a universal proposition is not about propositions or about objects, and thus it does not say anything about all propositions or all objects, because no proposition could be about objects or propositions. For example, (x) (x is material) must not be confused with (x) (if x is an object then x is material), which employs the formal concept object and therefore says nothing. Indeed, this is readily appreciated in common discourse. It is how we ordinarily understand universal propositions. We find it natural to say that All men are mortal is about all men, but not, as logic suggests, that it is about all individual things, including the moon, my pen, your feet, etc., saying about each that if it is a man then it is mortal. Saying the latter belongs only in the classroom. All men are mortal does not include the predicate is an object,
219

the only predicates in it are is a man and is mortal. If stated as (x) (if x is a man then x is mortal), the only expression in it other than those predicates would be the variable x, which is neither a name nor a predicate.

Wittgensteins reply to Russell appealed, not to some analysis of general propositions that was different from Russells, but rather to the formal status of the property of being a proposition, and thus of being an elementary proposition. Like Russell, he refused to analyze a universal proposition as the conjunction, the truth-functional product, of its singular instances: I dissociate the concept all from truth-functions (Tractatus 5.521). On the other hand, he did not deny that, in a certain sense, a universal proposition is equivalent to that conjunction: Indeed the understanding of general propositions palpably depends on the understanding of elementary propositions (4.411). Suppose that I am given all elementary propositions: then I can simply ask what propositions I can construct out of them. And there I have all propositions, and that fixes their limits (4.51). Propositions comprise all that follows from the totality of all elementary propositions (and, of course, from its being the totality of them all). (Thus, in a certain sense, it could be said that all propositions were generalizations of elementary propositions) (4.52).

Are 4.51 and 4.52 consistent with 5.521, I dissociate the concept all from truthfunctions? The answer is provided by the sentences that immediately follow 5.521: What is peculiar to the generality-sign is first, that it indicates a logical prototype, and secondly, that it gives prominence to constants (5.522), The generality-sign occurs as an argument (5.523), and If objects are given, then at the same time we are given all
220

objects. If elementary propositions are given, then at the same time all elementary propositions are given (5.524). The occurrence of the variable x in (x) Fx gives objects, in the sense that the variable ranges over objects it is objects that it admits as values, rather than, say, properties or propositions. The generality-sign in (x) Fx is (x) (...x), the formal property of (x) Fx in virtue of which the latter is a general proposition. It may be said to occur in the proposition as an argument because it is part of what determines the truth-value of (x) Fx, just as a name put in place of the variable x in Fx is part of what determines the truth-value of the resulting singular proposition. If we replace x with Jack, then we obtain the proposition that Jack is F, which may be true, though if we replaced x with Jill then we obtain the proposition that Jill is F, which may be false. Similarly, if we replace (x) (...x), the generality-sign in (x) Fx, with ( x) (...x), then we would obtain ( x) (Fx), which may be true even if (x) Fx is false. If the variable x may be said to give objects in the sense that it takes objects as its values, then the generality-sign (x) (...x) may be said to give all objects. A universal proposition is, in a certain sense, about all objects (all values of x), but this is not something it says, it is something it shows. Wittgensteins claim that There are objects is a pseudo-proposition does not mean that there are no universal first-order propositions. For such propositions are not about objects, even though the quantified variable in them is understood as ranging unrestrictedly over all objects.

In 5.3 Wittgensteins wrote. All propositions are the result of truth-operations on elementary propositions. And he went on to say, Every truth-function is a result of successive applications to elementary propositions of the operation '(-----T)( ,....)'. This operation negates all the propositions in the right-hand pair of brackets, and I call it the
221

negation of those propositions ( 5.5). Wittgenstein had explained earlier that is a variable whose values are terms of the bracketed expressionHow the description of the terms of the bracketed expression is produced is not essential. We can distinguish three kinds of description: 1. direct enumeration, in which case we simply substitute for the variable the constants that are its values; 2. giving a function x whose values for all values of x are the propositions to be described; 3. giving a formal law that governs the construction of the propositions, in which case the bracketed expression has as its members all the terms of a series of forms (5.501). It follows that If has only one

value, then N( ) = ~p (not p) [i.e., the negation of all the values of the propositional variable = ~p]; if it has two values, then N( ) = ~p.~q (neither p nor q) [i.e., the = ~p.~q] (5.51), and that If

negation of all the values of the propositional variable

has as its values all the values of a function fx for all values of x, then N( ) = ~( x). fx (5.52), ~( x). fx being the logical equivalent to (x) fx.

It is now that Wittgenstein adds 5.521, I dissociate the concept all from truth-functions. What it says is compatible with All propositions are the result of truth-operations on elementary propositions (5.3), because of the difference between what in 5.501 he had called kinds of description 1 and 2. Unlike the case of ~p and ~p.~q , where has

propositions as its values (kind of description 1), in the case of (x) fx has as its values the values of the propositional function fx (kind of description 2).238 In the former case, the terms to which the truth-operation '(-----T)( ,....)' is applied, i.e., p and q, are propositions that are enumerated, i.e., explicitly mentioned, In the latter case, they are merely the propositions, whichever they might be, that are the values of the propositional function fx, and thus they remain implicit. General propositions are truth-functions only
222

in the sense that their truth depends on the truth of all of their substitution instances. Since these substitution instances are not mentioned, general propositions are truthfunctions only implicitly. By contrast, ~p and ~p.~q explicitly mention, enumerate, the propositions of which they are truth-functions, , namely, p and q,.

Let us note again that I dissociate the concept all from truth-functions (5.521) was immediately followed by What is peculiar to the generality-sign is first, that it indicates a logical prototype, and secondly, that it gives prominence to constants (5.522), and The generality-sign occurs as an argument (5.523). Pace G.E.M. Anscombe239 and Robert Fogelin,240 who think that the generality-sign in (x) Fx is the variable x, I have suggested that it is (x) (...x). The latter indeed may be said to indicate a logical prototype and to give prominence to the sign f in (x) fx, which is the only constant in it. The generality-sign only shows itself because fx is the form of all substitution instances of (x) fx. And the latter is a truth-function of its substitution instances in the straightforward, literal, sense that its truth depends on their truth. But this, too, only shows itself. It is not and cannot be said. For (x) fx is not replaceable by the conjunction fa . fb . fc .

5.523 is followed by 5.524, which reads: If objects are given, then at the same time we are given all objects. If elementary propositions are given, then at the same time all elementary propositions are given. In view of the two propositions that preceded 5.524, we may say that the variable x in (x) fx gives all objects in the sense that it is an object (individual) variable, and that the propositional function fx in (x) fx gives all elementary propositions in the sense that all elementary propositions are substitution
223

instances of fx, if f ranges over all predicates attributable to an object. This is why a general proposition might be said to refer to all objects, if we accept the objectual interpretation, or to all elementary propositions, if we accept the substitutional interpretation. But such reference would consist in showing, not saying. The variable x shows all objects because it is an individual variable, and the propositional function fx shows all elementary propositions because fx is the form of all elementary propositions. But, since object is a formal concept, (x) fx does not say that all objects are f. Nor does it say that all elementary propositions of the form fx are true, since elementary proposition is also a formal concept.

One of Russells complaints in his letter to Wittgenstein was that it is awkward to be unable to speak of [the negation of all the values of the propositional variable ]. Wittgenstein replied: This touches the cardinal question of what can be expressed by a prop[osition] and what cant be expressed, but only shown. I cant explain it at length here. Just think that, what you want to say by the apparent prop[ositin] there are 2 things is shown by there being two names which have different meanings.e.g., (a, b) doesnt say that there are two things, it says something quite different; but whether its true or false, it SHOWS what you want to express by saying: there are 2 things. Then Wittgenstein added: I suppose you [Russell] didnt understand the way, how I separate in the old notation of generality what is in it truth-function and what is pure generality. A general prop[osition] is a truth-function of all PROP[OSITION]S of a certain formI suppose you dont understand the notation [for the values of the propositional variable ]. It does not mean for all values of .241 I suggest that what is truth function in (x) fx is what is expressed by (x), and that what is pure generality is what is expressed by
224

fx. All propositions of the form fx may be said to be shown by that form, in the straightforward sense that it is their form this is the pure generality in (x) fx. And in a no less straightforward sense, the truth of (x) fx depends on there not being a proposition of the form fx that is false this is the truth function in (x) fx. But (x) fx does not say that there is no proposition of the form fx that is false. It is not a proposition about propositions. Since proposition is a formal concept, there can be no proposition about all propositions of the form fx or of any other form.

Although rooted in logical considerations, Wittgensteins distinction between saying and showing is more easily understood in terms of his picture-theory of thought and meaning. In chapter 4 I suggested that the theory was not idiosyncratic, but rather the natural successor of the traditional view of thinking as involving representations, ideas, mental images. Even in ordinary physical pictures, e.g., landscapes and portraits, much is shown, in a literal, everyday sense of shown, that is not itself pictured or even picturable, in any sense of pictured. The reason is that all physical pictures are logical pictures, though not vice versa. My example in chapter 4 was a painting of a tree next to a barn. Certain parts of the picture represent the tree and the barn, and the spatial relation between them represents their relation of being next to each other. But nothing in the picture represents, says, that the tree and the barn are objects, rather than, say, properties. And nothing in the picture represents their relations being a relation, nothing says that their being next to each other is a relation, rather than, say, a shape or color. Yet the picture shows all this, indeed must show it in order to represent a tree next to a barn. Indeed, there cannot be such a thing as picturing being an object or being a relation.
225

The sense in which something can only be shown is especially obvious in the case of an objects being an object. Consider the sentence This page is white. It does say something. What it says can even be pictured literally, not just logically. But the sentence This is an object, in Wittgensteins sense of object, says nothing, for it presupposes what it purports to say. Its even having sense depends on its being true. Yet the sentence is not gibberish. It does show something. It shows the logical category to which the item belongs. Bergmann called that category individuality but acknowledged its ineffability, the futility of saying about an individual object that it is an individual object. The sentence a is an object presupposes what it purports to say, since its subject term could only be a name, and in Wittgensteins technical uses of name and object names can name only objects: A name means (bedeutet) an object. The object is its meaning (Bedeutung) (3.203). This is why A name shows [zeigt] that it signifies an object (4.126).

While the distinction between saying and showing has a reasonably clear and important application to propositions of the forms x is an object and next-to is a relation, how it applies to other, more complicated cases is less clear though not less important. This is certainly true of its application to general propositions. To understand it better, we may take advantage of the notion of presupposition that P. F. Strawson proposed decades later, which we examined in chapter 2, and assume, at least provisionally, that presupposing something includes implicitly reference to it. Then, if we accept the objectual interpretation of quantification, we can hold that, though (x) x does not say that all objects are (since object is a formal concept) and thus does not refer to all objects, it
226

may be taken to presuppose such reference and thus to refer implicitly to all objects. It is (x) (if x is an object then x is ), not (x) x, that explicitly mentions and thus refers to objects, and for this reason it is inadmissible. The point may become clearer if we return to the statement All men are mortal. Translated as (x) (if x is a man then x is mortal), it does not say that all objects are such that if they are men then they are mortal, but it may be taken to presuppose this since the variable x ranges over all objects and thus to refer implicitly to all objects. What All men are mortal does say is just that all men are mortal. It is about men, not about objects. Were we to accept the substitutional interpretation of quantification, we could hold that (x) x does not say that all elementary propositions of the form x are true (since elementary proposition is a formal concept), but it can be taken to presuppose this and thus to implicitly refer to all elementary propositions. The actual statement All men are mortal does not say that all propositions of the form if x is a man then x is mortal are true, but it may be taken to presuppose this. What it does say is just that all men are mortal. It is about men, not about elementary propositions.

2. Formal and Material Properties. After his return to Cambridge in 1929, Wittgenstein continued working on the topic of generality. But what, according to the Tractatus, was shown by a general proposition, whether universal or particular, in Philosophical Remarks, written shortly after his return but published posthumously, was described as possibilities left open by the proposition, like an incomplete or partially indefinite picture: The general proposition I see a circle on a red background appears simply to be a proposition which leaves possibilities open. A sort of incomplete picture. A portrait in which, e.g., the eyes have not been painted in.
227

But what would this generality have to do with the totality of objects?242 Also: If I give a correct description of a visual field in which three red circles stand on a green ground, it surely wont take the form of saying ( x (x, y, z): x is circular and red and y is circular and red, etc. etc. You might of course write it like this: there are 3 circles with the property red.It is plain that the proposition about the three circles isnt general or indefinite in the way a proposition of the form ( x (x, y, z). x.y.z is. That is, in such a case, you may say: Certainly I know that three things have the property , but I dont know which; and you cant say this in the case of the three circles.243

A couple of years later, in Philosophical Grammar, also published posthumously, Wittgenstein again returned to the topic. Regarding the position he had defended in the Tractatus he now writes, My view about general propositions was that ( x). x is a logical sum and that though its terms are not enumerated here, they are capable of being enumerated.For if they cant be enumerated we dont have a logical sum.Of course, the explanation of ( x). x as a logical sum is indefensible. [But] it is correct that ( x). x behaves in some ways like a logical sum and (x). x like a product; indeed for one use of words all and some my old explanation is correct, for instance for all the primary colours occur in this picture But for cases like all men die before they are 200 years old my explanation is not correct.244 I take Wittgenstein to mean that the sentence about the primary colors would be an exception because primary color is an abbreviation, say, of red, green, or blue, and so the sentence would be an abbreviation of red, green, and blue occur in this picture, and not a general statement.

228

In Philosophical Grammar Wittgenstein also writes, If I say there is a black circle in the square, it always seems to me that here again I have something simple in mind, and dont have to think of different possible positions or sizes of the circle. And yet one may say: if there is a circle in the square, it must be somewhere and have some size. But in any case there cannot be any question of my thinking in advance of all the possible positions and sizes.I would like to say that in the proposition there is a black circle in the square the particular positions are not mentioned at all. In the picture I dont see the position, I disregard it.245 When asserting the general proposition There is a black circle in the square, one does not mention the other possible positions and sizes the circle might have had, one does not think of them, they are disregarded. Indeed, one does not even see the position it does have, it too is disregarded. Of course, the circle does have a position, one of an indefinite number of possible positions, but none is mentioned or seen. If the position of the circle were not disregarded, if it were seen, thought of, or mentioned, the case would be quite different. It would be like that of the singular proposition This black circle is in this square.

Wittgenstein did not explain these later remarks in detail. Nevertheless, they do fit what he had said in the Tractatus. (It is wrong-headed history of philosophy and poor psychology to think, as many do, that he had just abandoned the Tractatus!) In Wittgensteins earlier terminology, which he no longer employed, we might say that the possible positions of the circle are not said but show themselves. According to Russellian logic, the universal statement (x) x says that all individual objects are , that everything is , and that ( x). x says that some individual objects are , whichever they might happen to be. But in the Tractatus Wittgenstein had held that it is
229

nonsensical to speak of the total number of objects, indeed of objects at all, since object is a formal concept. Now, in Philosophical Grammar, he makes the further claim that general statements are often, perhaps usually, not even intended to be used in accordance with Russellian logic. The statement There is a circle in this square is not intended to say anything about all objects, not even about all objects that are in the square. In effect, Wittgenstein suggests that the general particular (existential) statement There is a circle in the square and the general universal statement There are only two things that are circles in this square are better understood as analogous to the singular statement This circle is in this square, rather than to quantified statements containing a variable ranging over all individual objects, or even over just all circles in the square. But how then do the general statements differ from the singular statement?

We find no answer in Philosophical Remarks or in Philosophical Grammar. The answer Wittgenstein had provided in his 1919 reply to Russell was that the generality of a general statement consists not in what it says but in what it does not say yet shows. In all three texts, however, he insisted that our use or understanding of general statements is far removed from what Russellian logic tells us. We do not use There is a circle in the square to say something about all things, or even about all circles, viz., that some are in the square. We certainly do not use it to say that it is not the case that no circles are in the square. It does entail the latter, but (like any statement) it also entails an indefinite number of other statements. Surely we are not making also all those other statements when we make that one statement.

230

Indeed, in Philosophical Grammar Wittgenstein expressed doubts about the very propriety of representing ordinary general propositions in the canonical forms of Principia Mathematica. For example, he wrote that the translation of There are two circles in this square as There are only two things that are circles in this square sounds crazy. Wittgensteins explanation of how one was led to this translation is worth quoting in detail: The original source of this notation [( n) and in general ( x)] is the expression of our word-language There is a with such and such properties. And here what replaces the dots is something like the book from my library or thing (body) in this room, word in this letter, etc. We think of objects that we can go through one after the other. As so often happens, a process of sublimation turned this form into there is an object such that and here too people imagined originally the objects of the world as like objects in the room (the tables, chairs, books, etc.), although it is clear that in many cases the grammar of this ( x), etc. is not at all the same as the grammar of the primitive case which serves as paradigm. The discrepancy between the original picture and the one to which the notation is now applied becomes particularly palpable when a proposition like there are two circles in this square is rendered as there is no object that has the property of being a circle in this square without being the circle a or the circle b.[T]he Russellian notation here gives an appearance of exactitude which makes people believe the problems are solved by putting the proposition into the Russellian form.246

I have repeatedly pointed out that the Russellian interpretation of universal statements ignores the fact that in actual talk and thought most generalizations are what linguists call generic statements, i.e., statements of the form Fs are Gs, rather than universal
231

statements, which are of the form All Fs are Gs, and in the next chapter will argue that even universal statements are usually intended only as generic by allowing for exceptions. But the reason for Wittgensteins misgivings about the Russellian interpretation of general statements were different. They fit Wittgensteins concern in later years with ordinary language. For example, he wrote, Mathematical logic has completely deformed the thinking of mathematicians and of philosophers, by setting up a superficial interpretation of the forms of our everyday language as an analysis of the structure of facts.247

3. Bergmann on Generality In Generality and Existence Bergmann used an example similar to Wittgensteins examples in Philosophical Remarks and Philosophical Grammar: being presented with a single square inside a circle. Bergmann asked, with what else must I be presented when I say This square is the only one inside this circle? He pointed out that the transcription of the statement would be F(a, b) . (x) [(x = a) v ~ F (x, b)], which contains the general operator (x) and a, b, and F standing respectively for this square, this circle, and the relation of being inside.248 Bergmann answered the question by saying that when presented with a single square inside a circle he was also presented with generality and existence. He was presented with generality in seeing that the square was the only square in the circle, and with existence, by which he meant in this case particularity, in seeing that there is a square in the circle. They are the entities that the universal quantifier, (x) or the phrase for everything, and the particular quantifier, ( x) or the phrase there is

232

at least one, respectively represent. Therefore, there are such entities as generality and existence, though they subsist, rather than exist.249

In New Foundations of Ontology Bergmann continued to hold that he was presented with generality and existence, though now he calls them the universal and the particular quantifiers. He went far beyond Generality and Existence by offering a much more complex account of quantification, which still resembled Wittgensteins but was a part of a rich, all-encompassing ontology, which Wittgenstein never attempted. Bergmann renounced his earlier distinction between existence and subsistence, holding now that whatever is thinkable exists.250 He pointed out that Existence is univocal.Yet the differences among some of the several existentsare very great indeedmomentous, or enormous,251 thus suggesting that the distinction between existence and subsistence is really a distinction between two radically different among the many kinds of existents.

Bergmanns assertion that everything thinkable exists should be no more surprising than Meinongs assertion that there are things of which it is true that there are no such things, but it is free from the latters paradoxical air, which continues to confuse Meinongs readers. There is a golden mountain, it has being, it exists, Bergmann would say, but of course it is fundamentally different from, say, the Rocky Mountains. Like the latter, it is a complex of facts, but unlike the Rocky Mountains it is pervaded by the mode of potentiality, rather than the mode of actuality. Bergmanns critics, like Meinongs, seem to attach magic significance to the words exist and being. But, important though they are, they still are words, conventional signs, and their ordinary use might not be suited for the purposes of ontology. The truth is that we can think and talk
233

about, even describe in some detail, say, a golden mountain east of Denver, just as we can think, talk about, and describe the Rocky Mountains west of Denver. What we must not do, of course, is to think that the former is an actual mountain.

Bergmann began his account of generality in New Foundations by denying that variables, whether free or bound, stand for anything.252 This was not surprising in itself, but it did require surprising changes in the analysis of general statements. Bergmann now argues that the universal quantifier in a general fact is a function, to be represented in the general statement by the sign \/, but without attaching a variable such as x. It takes as argument a 2-tuple that consists of (1) the individual thing in the singular fact asserted by some singular substitution instance of the general statement, and (2) that singular fact itself. The value of the function is the general fact.253 If, for example, the statement is all f1s are f2s, the 2-tuple might be <a, f1(a) f2(a)>. Bergmann used 2-tuple, instead of pair, because in New Foundations he also offered a highly original account of sets that prohibits casual uses of set-theoretical terms. Perhaps most surprisingly, however, he insists there that the conscious state or awareness of the general fact, which he calls the referent of the general statement, also includes an auxiliary act of consciousness, the intention (i.e., the intentional object) of which is the sentence itself, the words used in making the statement. Bergmann called this intention the text of the awareness.254 Since we cannot perceive or imagine a general fact, he held, the awareness of it is a believing or entertaining, not a perceiving or imagining.255 This explains why I cannot think any generality such as, say, all-men-are-mortal, without at the same time thinking the words all-men-are-mortal..One cannot believe, or doubt, or remember, and so on, any generality without also perceiving the appropriate words.256 Indeed, all awarenesses,
234

except primary Perceivings and Imaginings (and undoubtedly some Feelings), are inseparable from their texts. Thatnot only gives language its due without giving it too much; it also reassuringly recovers the sound core in a large body of recent and contemporary thoughtfrom Watson to Wittgenstein.257

The fact that all f1s are f2s is built by the function \/ not just from one argument but, indifferently, from an indefinite number of alternative argumentsfrom <a, f1(a) f2(a)>, from <b, f1(b) f2(b)>, and <c, f1(c) f2(c)>, and so on. In the text of [the awareness], howeverthere is no cue to this multiplicity.258 Bergmann gave all green (things) are square as an example. It is the text of an awareness that has as referent the general fact, presumably not actual, that all green (things) are square.

The 2-tuples that the function \/ takes as arguments, such as <a, f1(a) f2(a)>, e.g. <this, if this is green then this is square>, are not mentioned in the general statement, there is no cue in it to their multiplicity. But they all are essential to the general fact. The latter would not be actual if the singular facts in the 2-tuples were not all actual: all f1s are f2s would not be true if all its singular substitution instances were not true. From which of them the function \/ builds the general fact is ontologically indifferent. But it might not be psychologically indifferent, since the speaker or hearer of the general sentence must at least in principle be able to perceive or imagine one of them. I shall return to this latter point.

Thus the assay of all f1s are f2s, i.e. its ontological analysis, is not, conventionally (x) [ f1(x) f2(x)], but, rather, alternatively and indifferently \/ [<a, f1 (a) f2 (a)] > or
235

any of its variants; indifferently because all those variants are one and not many.259 The variants of \/ [(a, f1(a) f2(a)], of course, are \/ [<b, f1(b) f2(b)>], \/ [<c, f1(c) f2(c)>], and so on. Each is an alternative assay of the one and same general fact. We may note that in standard logic it is also indifferent, unless the context requires otherwise, whether we symbolize all f1s are f2s as (x) [ f1(x) f2(x)], (y) [ f1(y) f2(y)], or (z) [ f1(z) f2(z)]. But standard logic uses variables, which according to Bergmann represent nothing and therefore have no place in ontological analysis.

It may seem that Bergmanns insistence that the arguments the quantifier \/ takes, say, in the case all f1s are f2s, are 2-tuples is an unnecessary complication, but the reasons for it are compelling. What else could those arguments be? Not f1(x) f2(x), because it contains variables. Nor the properties f1 and f2 themselves. One might be presented with them, as well as with the quantifier, but this would not suffice for being presented with the fact that all f1s are f2s. According to Bergmanns principle of acquaintance, one cannot be presented with f1 and f2 except when they are exemplified.260 But even if one could, being presented with them as well as with the quantifier \/ would hardly count as being presented with the fact that all f1s are f2s, or indeed with any fact: the three presentations might be unrelated. They might be unrelated even if f1 and f2 are exemplified: one might be presented with them as well as with the quantifier by virtue of being presented with the fact, say, that all f2s are f1s, or with facts such as that all f1s are f3s and all f2s are f4s, rather than with the fact that all f1s are f2s. In general, Bergmann asked rhetorically, if f1 were the argument of \/ for the value (x) (f1x), what would be the argument of \/ for the value (x) [f1(x) v f2(x)? Surely not f1() v f2()!

236

The only [other] thing I can think of, and which therefore I propose [as the argument of \/] is a 2-tuple such as <a, f1(a)>.261 This is why the assay of (x) (f1x) is \/ <a, f1(a)> .

Without the singular fact that is one of the terms of the 2-tuple, there would be no relevant conscious state or awareness at all when one makes the general statement, for there would be nothing relevant for one to be aware of. Could the quantifier take as argument the singular fact f1 (a) f2 (a), rather than the 2-tuple <a, f1 (a) f2 (a)>? No, because even if \/ could take f1 (a) f2 (a) as argument, its value would not be a general fact. The notation must also be explicit with respect to which constituent of the singular fact the quantifier operates on, just as in standard logical notation it must be explicit which variable the quantifier binds. If variables are not used, this can be explicit only if the quantifier, so to speak, brings the constituent out of the singular fact, while also retaining the singular fact. The constituent and the singular fact must both be explicitly in the argument that the quantifier takes, and this amounts to saying that the argument must be the 2-tuple of which they are the terms. Bergmann expressed the point by saying that the individual is the target of the quantifier, while the singular fact is its scope. In the case of the statement all green (things) are square, the target might be any particular perceived or imagined object, even your hand, which, were the statement true, would be square if green.

Indeed, in the case of all f1s are f2s, or all green (things) are square, only one individual in each 2-tuple could be the target. But a singular fact often has more than one individual as constituent, and thus it could be the scope of the quantifier of different general facts. If the singular fact is that a is to the left of b, we must distinguish between
237

the general fact that all things are to the left of b and the general fact that a is to the left of all things. In standard notation, we would do this by distinguishing between (x) (x is to the left of b) and (x) ( a is to the left of x). In Bergmanns notation, we would distinguish between V <a, a is to the left of b> and V <b, a is to the left of b>. Bergmanns notation may seem obscure because it is unfamiliar, but he would say that from the standpoint of ontology the variables that the standard notation employs are far more obscure.

There are important similarities between Bergmanns later account of generality and each of Wittgensteins pronouncements on the topic: in the Tractatus, his letter to Russell in 1919, Philosophical Grammar, and Philosophical Grammar, though there are also obvious differences. The singular substitution instance of the general statement that stands for the singular fact Bergmann calls the scope of the quantifier is not, of course, asserted, it is not said, yet it must be, so to speak, in the background, if the statement is to express a relevant conscious state. We might even say that it must show itself. For it is the singular substitution instance that provides the general statement with its target and scope, both of which must, in some sense, be presented or given, though of course not as they would be if it was the singular, rather than the general, statement that was asserted. We could say that the 2-tuple from which, as its argument, the quantifier builds the general fact must also show itself. Indeed, the whole indefinite number of alternative arguments from which the quantifier builds indifferently the general fact may be said to show themselves. They must be there, in the background, like the indefinite number of possible positions of the circle on a red background in the example Wittgenstein gave in Philosophical Remarks. Like those positions, the alternative
238

arguments may be thought of as possibilities left open, neither enumerated nor capable of being enumerated, but with the speaker and hearer knowing that they are there, though not which they are. Thus, the general statement also may be said, as Wittgenstein did in Philosophical Grammar, to be indefinite, an incomplete picture, like a portrait in which, e.g., the eyes have not been painted in. There can be no question of thinking in advance of all the different alternative arguments the quantifier may indifferently take, they are not mentioned at all, they are unseen and disregarded. Yet they are there, like the different possible positions and sizes of the circle in the square that the statement there is a black circle in the square allows, even though the speaker has something simple in mind when making the statement.

In Philosophical Remarks Wittgenstein denied that the general proposition I see a circle on a red background has anything to do with the totality of objects, but in the Tractatus he could have said that it shows that totality. I have suggested that when Bergmann writes that what he calls the scope of the quantifier is not asserted, yet must be in the background if the statement is to express a relevant conscious state, we might say that it must show itself, and also that what he calls the alternative arguments from which the quantifier builds the general fact may be said to show themselves. If we did say this, would show have the sense that it had in the Tractatus, or in Wittgensteins assertion in his 1919 letter to Russell that while a universal statement does not say that all elementary (singular) propositions are given, this is shown by there being none having an elementary sense which is not given? The truth is that Wittgenstein did not explain that sense, just as Bergmann did not explain, for example, the sense of his term presented.

239

The reason, in both cases, was not dereliction of duty but the fact, obvious to them if not to their readers, that what they meant was too basic to allow for further explanation.

According to Bergmann, one must be presented with what he called the quantifier, i.e., generality, what (x) or \/ stands for, and Wittgenstein presumably would have disagreed. There are no 'logical objects, he wrote, thus announcing his break with the logical realism of Frege and Russell, though, as we have seen, he did not adopt logical antirealism. But it is not certain that Wittgenstein would have disagreed that thoughtful use of general sentences, at least those about perceivable objects or facts, involves being able to perceive or imagine, however peripherally and unfocusedly, something nonverbal of which the sentences would be true. Nor is it certain that he would have denied that such use would involve actual awareness, even if also peripheral and unfocused, of the sentence itself, what Bergmann called the text, whether by seeing, hearing, or imagining it. Bergmann held that these are phenomenological, or as he also put it, anthropocentric, even anthropological, facts that this is how we humans think and speak.

But Bergmann also offered a detailed account of these facts, which Wittgenstein did not. Bergmann explained that the text is needed in order to close the phenomenological distance between what is presented to us when we thoughtfully make a general statement and the assay of what the statement says, to close the gap between what the text of an awareness may lead one to expect, on the one hand, and the assay in fact proposed for its referent, on the other.262 The text is fused, absorbed, into the nontext, he wrote,263 it has fusing power.264 The general fact that all f1s are f2s is built
240

by \/ from an indefinite number of alternative arguments, from <a, f1(a) f2(a)>, <b, f1(b) f2(b)>, <c, f1(c) f2(c)>, and so on, but there is no cue to this multiplicity in the sentence all f1s are f2s, nor of course in its transcription, whether the conventional f1(x) f2(x) or Bergmanns \/ <a, f1(a) f2(a)>.

This phenomenological distance is unnoticed only because of the fusing power of the sentence. On no account of generality does a general statement contain explicit reference to the multiplicity of what makes it true. Whatever account we accept, we must rely on the statement to serve, so to speak, as proxy for that multiplicity. A merit of Bergmanns account is that it makes clear what all accounts of generality must admit, namely, that when saying, e.g., all green things are square, we could, as he pithily puts it, in principle also say regarding some individual thing: generalized for this: if this is green then this is square.265 Saying the latter would differ from saying the former only by making explicit that the assertion is a thoughtful one, not a mere utterance, that one actually has something relevant in mind when making it. In the old empiricist terminology, it would make explicit the presence before the mind of an idea, whether of sensation or imagination. In Bergmanns terminology, it would make explicit the presence of an individual actually perceived or imagined.

Bergmann suggests that if saying generalized for this: if this is green then this is square, rather than all green things are square, were our natural way of expressing the generality, then in general when we say that all f1s are f2s we would even be presented with the actuality of such complexes as \/ <a, f1(a) = \/ <b, f1(b), i.e., we would find the truth of the statement \/ <a, f1(a) = \/ <b, f1(b) obvious, indeed necessary.266 In the
241

Philosophical Investigations Wittgenstein would not have agreed, but in the Tractatus he might have been sympathetic. Surely, Bergmanns view is plausible. Can one thoughtfully assert that all green things are square without at least in principle being able to refer to some particular thing, perceived or imagined, even if it happened to be ones hand or ones computer, and say of it that if it is green then it is square? Bergmann of course held that one must actually, not just in principle be able to, make such reference, but the difference might be a matter of how we use the adverb thoughtfully, not a matter of ontological import. In any case, a detailed account of generality is needed, and Wittgenstein offered none of his own, in the Tractatus or in his later works.

To appreciate Bergmanns account, we should consider the alternatives to it. There is, first, the reductionist account of universal statements as conjunctions, and of particular statements as disjunctions, of their singular instances. As we saw, Bergmann found no merit in it, just as Frege and Russell did not. There is, second, Freges account of generality as a second-level function saturated by first-level functions. Bergmanns account resembles it, but Freges presupposed Freges ontology, which Bergmann rejected for reasons independent of the topic of generality.267 Third, there is Russells appeal to irreducibly general facts. Bergmanns view in Generality and Existence was similar to Russells, and his view in New Foundations of Ontology may be described as a refinement of it. The referent of all f1s are f2s, which Bergmann analyzed as \/ [<a, f1 (a) f2 (a)] > (or alternatively and indifferently any of the latters variants (\/ [<b, f1(b) f2(b)>], \/ [<c, f1(c) f2(c)>], and so on), is a fact, of course a general fact. But Bergmann provided an analysis of such a fact, which Russell did not. Indeed, Russell totally ignored the obvious and crucial question, In virtue of what are general facts
242

general? To have taken this question seriously was one of Bergmanns great merits. And, fourth, there is the view, often but wrongly attributed to Wittgenstein, that all there is to generality is the general sentences. Bergmann probably thought it was a case of a linguisticism too crude to deserve discussion, so he offered none, but he did agree that awareness of the referent of a general statement includes perceptual or imaginative awareness of the sentence itself.

The merits of Bergmanns position become evident when we contrast his transcription of the sentence all f1s are f2s as \/ <a, f1 (a) f2 (a)> with the standard transcription of it as (x) (f1x f2x). The latter includes the unrestricted individual variable x and therefore the sentence can be read as saying something about all individuals, as being about this computer, the page you are now reading, the moon, and so on. Bergmann thought that if we had no particular individual in mind when we assert the sentence we would have nothing relevant in mind, and so we would not be making a genuine statement at all. Locke, Berkeley, and Hume would have agreed. The traditional empiricist tenet was that to understand what we say or hear we must have an idea of what it is about. Bergmann seemed to accept this tenet in the case of general statements, if by idea is meant an object perceived or imagined, rather than a representation of it (he opposed representationalism vehemently). He believed that whenever we make a genuine general statement we must perceive or imagine a particular individual of which the statement is true. But he made the empiricist tenet more plausible by allowing that we can select, alternatively and indifferently, that individual.

243

The sentence (x) (f1x f2x) does not mention this computer, the page you are reading, the moon, or any other individual thing. In Wittgensteins Tractarian terminology, it does not say that, e.g., if this computer is f1 then it is f2. Nonetheless, presumably Wittgenstein thought that somehow it must show this. It must do so at least in the sense that, if a thoughtful, circumspect, utterer of (x) (f1x f2x) were asked whether it means that if this computer were f1 then it would be f2, the utterer would agree, or express consent in some other way. Bergmann did not use Wittgensteins terminology, but he might have done so in order to explain the relevance of, say, this computers being a term in one of the indefinite number of 2-tuples from which the quantifier indifferently builds the general fact that all f1s are f2s. Wittgenstein, of course, denied the empiricist tenet in his later works, but even there he probably would have agreed because it seems obviously true that for a statement about the sort of things that can be perceived or imagined to make sense the speaker or hearer must principle be able to, even if in fact does not, perceive or at least imagine something of which the statement would be true.

Bergmanns and Wittgensteins Tractarian positions on generality shared a negative but important feature in Bergmanns terminology, that a general statement does not mention the singular statement that provides it with its target and scope, and in Wittgensteins, that the general statement does not mention the elementary statements of which it is a truth function. They also shared an important positive feature. Bergmann argued that if one is aware of what is said by a general statement, then one is aware also of the sentence used in making it, that in this sense thought depends on language in the case of generality, indeed that it does so in all cases except some perceivings, imaginings, and feelings. This dependence, he held, is not causal or external; it is internal,
244

constitutive.268 Thought is inseparably intertwined with language, Bergmann wrote,269 thus endorsing much of the linguisticism he had vehemently opposed in the past. And Wittgenstein was the philosopher who in the Tractatus endorsed and in the Philosophical Investigations celebrated that linguisticism.

It is simplistic to think that metaphysics provides descriptions of the world that are additional to those provided by science and everyday thought, and to view metaphysical disagreements as disagreements about the truth of such descriptions. Metaphysicians do not discover entities that are hidden from the rest of us, including chemists, physicists, and astronomers, nor do they have the sort of training and instruments needed for making such discoveries. What they can do, however, is to see, draw attention to, and emphasize similarities and differences between fundamental kinds of items in the world that in everyday life, and even in science, go unnoticed, not because they are hidden but precisely because they are fundamental and there for everyone to see. Bergmann wrote, Is there a felt difference between the external property, as some call it, of being green and the internal one, as they also say, of being a property? Directly one cannot argue on either side. That is one reason, though to be sure not the only one, why at some place or places one must appeal to the phenomenological basis. All I can say, therefore, is that this particular difference pierces my eyes.270 By phenomenological basis he meant what in the same work he also called the phenomenological rock bottom and the jumping-off place.271

Bergmanns view that a general statement does not mention yet does involve the singular statements that provide the quantifier with its target and scope, and Wittgensteins
245

view that a general statement does not say yet does show that all elementary propositions are given, acknowledged, drew attention to, and emphasized, in their own but perhaps not incompatible terminologies, the fundamental differences between general and singular statements. These differences are there for all to see, but they pierce few eyes. Even Aristotle, the founder of logic, did not see them clearly when he counted both as subjectpredicate statements. Bergmann wrote of what is presented, and Wittgenstein of what is only shown. These are metaphors and need not signify fundamental disagreement. There is no established terminology for what Bergmann and Wittgenstein wanted to say, because of its novelty. Instead of caviling at the obscurity of their writings, we might do better if we open our eyes and perhaps even jump!

In chapter 4 I suggested that it would be a mistake to ask whether Wittgensteins semirealism is true or right, and much the same can be said about Bergmanns later ontology. Neither is a theory about any peculiar objects or facts. Both exemplify a conception of philosophy more sophisticated than such a philosophical theory might suggest. I shall say much more about this conception in chapter 13.

Wittgenstein and Bergmann were not realists regarding general statements. But neither were they antirealists. They did not hold that general statements assert facts in the way singular statements assert facts. But they also did not hold that all there is to general statements is the general sentences and our use of them. Their views regarding generality are paradigms of what I have called semirealism. Such a temperate position deserves respect, even if we remain noncommittal regarding its details.

246

There is more to be said in favor of logical semirealism, as contrasted with straightforward logical antirealism, than we find in Wittgensteins or in Bergmanns works. There are also certain distinctive experiences associated with the use of logical expressions, though they are not experiences of logical objects. The occurrence of such experiences does not support logical realism. But neither does it comport with ordinary logical antirealism. It will be the topic of chapter 9. First, however, we must devote attention to a severe limitation of both Wittgensteins and Bergmanns accounts of generality. Neither acknowledged the existence of the kind of general statements that are most common, namely, generic statements. It is the ubiquity and central role of such statements that lends support to unqualified antirealism in metaphysics.

Chapter Eight: GENERIC STATEMENTS 1. The Ubiquity of Generic Statements. Universal and particular (existential) statements are not the only general statements. There are also those that linguists call generic. In section 1 of this chapter I argue that they are far more important to cognition than universal statements, that universal statements themselves are commonly intended not as literally universal but rather as generic, and that generic statements are ubiquitous in both everyday talk and science. In section 2 I argue that, though realism about the world presupposes the category of fact, in the robust Russellian sense of fact, there are no generic facts even if there are universal

247

facts. In section 3 I argue that generic statements are not reducible to any other kind of statement.

The ubiquity of generic statements, whether explicitly generic or disguised as universal, provides the clearest and most eloquent argument for the antirealist thesis. Such statements call for an antirealist interpretation, unlike universal statements, which as we have seen admit of a semirealist interpretation. I noted earlier that the traditional arguments for antirealism have been excessively abstract and based on obscure philosophical assumptions. A clear and plausible defense of antirealism, I suggested, must bypass them and start afresh from specific and readily understandable truths. One such truth is the implausibility of a realist interpretation of universal, particular, and molecular statements, which led us in the previous chapter to endorse only semirealism regarding such statements. The ubiquity of generic statements is another such specific truth. It provides the least controversial yet virtually conclusive argument for antirealism about the world, cosmological antirealism, though not for antirealism about things, ontological antirealism. Neither common sense nor philosophy would accept the existence of facts that are just generic.

The 17th century French philosopher Antoine Arnauld found the statement Dutchmen are good sailors puzzling..272 It does not say that all Dutchmen are good sailors. Some are not. But neither does it just say that some are. Some Germans also are good sailors, but perhaps Germans are not good sailors. What, then, does the statement say? We may be uncertain whether Dutchmen are good sailors, but let us suppose it was common knowledge among those whose judgment mattered when Arnauld wrote, presumably 17th century shipmasters. We therefore also suppose that the statement was true. But if a fact
248

is the sort of brute extralinguistic entity that according to Russell makes a statement true and which Wittgenstein had in mind when he declared in Tractatus LogicoPhilosophicus that the world is the totality of facts, not of things, Dutchmen are good sailors corresponded to no such entity, for there was not such an entity in the world.

Perhaps there was a fact to which the particular statement Some Dutchmen are good sailors corresponded. Perhaps there would have been a fact to which the universal statement All Dutchmen are good sailors corresponded, had this statement been true. But there was no distinctive, third fact to which Dutchmen are good sailors corresponded. Its truth did depend on the truth of some statements of the form x is Dutch and x is a good sailor, and perhaps these statements did correspond to brute Russellian facts, but Dutchmen are good sailors was not the conjunction of these statements and thus did not correspond to the fact, if there was one, that made the conjunction true. Nevertheless, the 17th century shipmasters had knowledge of its truth, and that truth mattered greatly in their world. Yet what they knew was not in that world. This is what puzzled Arnauld. There was no similar puzzle in the case of the other statements mentioned here.

Dutchmen are good sailors is an example of a vast number of statements of the form Fs are G, some of great practical and scientific importance. Linguists call them generic. They are general, not singular, but also not universal statements. Nor are they particular (existential) statements, which are much weaker. They are usually made without intention to endorse the corresponding universal statement and are understood so by the listener. Arnauld gave as examples also Frenchmen are brave, Italians are
249

suspicious, Germans are large, Orientals are sensuous, and many others.273 In the recent literature of linguistics we find Birds fly (penguins are birds but do not fly), Frenchmen eat horse meat (most French people do not), and John smokes a pipe (sometimes he smokes cigarettes).274 As the last example shows, a generic statement need not have the grammatical form Fs are G, just as a universal statement need not have the grammatical form All Fs are G, much less For every value of x, if x is F then x is G. What matters is that the statement is intended and understood as replaceable, upon analysis, by a statement of that form.

Here are some other examples. After the German election in September 2005, an observer wrote, It is clear that Germans do not want to be governed by Angela Merkel. There is no other way to explain the CDUs collapse to a 35.2% in the election after reaching 49% only a couple of months ago in opinion polls. The author obviously did not mean that all Germans were unwilling to be governed by Angela Merkel. Yet, the statement is an example of coherent, perhaps astute political thought, and it might have been true. The Encyclopedia Britannica informs us that The solubility of a gas in a liquid rises as the pressure of that gas increases, but it also says that exceptions may occur at very high pressures. Economists say that reducing taxes leads to increased economic growth and therefore government revenue, but they do not deny that sometimes it does not. No pharmaceutical company promotes its drugs as 100% effective, and no responsible physician tells a patient that the recommended surgery is is 100% safe. Parents, physicians, and politicians insist that smoking causes lung cancer, but even politicians avoid saying that it always does. Physicians do not even say that it is always bad for your health the Surgeon General only says that it may be. Exercise prolongs
250

life is considered true but, notoriously, exercise often fails to prolong life. Abstention from universal statements is characteristic of serious thought and discourse.

Indeed, universal statements themselves are commonly intended and understood as though they are only generic. Strawson noted that there are many cases of subjectpredicate statements beginning with all which it would be pedantry to call false on the strength of one exception or a set of exceptions.275 It might not be pedantry in the case of universal statements in mathematics or highly theoretical areas in science. However, Strawson pointed out, they are also statements philosophers often consider analytic or disguised definitions, meaning-postulates, reduction-sentences, inference-tickets, conventions not statements of fact.

In everyday discourse, we do make universal statements that allow for no exceptions, e.g., All of Jacks children attended the wedding, but they are readily replaceable with conjunctions of singular statements (the statement can be supported by a list of the children,) which the typical universal statement is not. We make universal statements commonly for rhetorical purposes, e.g., saying All politicians are crookedwhen both speaker and listener know that some are not, or Everybody loves her though obviously complete strangers do not. In the interpretation and application of the law, universal statements are studiously avoided because the possibility of exceptions must be allowed this is why there are courts and lawyers. In the areas of science where most scientific endeavor takes place geology, zoology, botany, medicine, anthropology, psychology, linguistics, sociology, economics, even much of biology and chemistry universal statements are scarce. It is generic statements that are common, such as Morotopithecus
251

bishopi was a fruit-eater, which does not mean that all members of the species were fruit-eaters or that all they ate was fruit, or Patients with prior strokes benefit from taking Lipitor, which does not mean that all do. Scientists shun universal statements because they believe that such statements could be justified only by information about real causal connections, which they seldom if ever have. They tend to rely, instead, on statistical reports like 265 or 11.2% of the patients who took Lipitor in a double-blind, randomized, and placebo-controlled clinical trial suffered a stroke over five years, while 795 or 37% of those who took a placebo did.

Aristotle noted that it is the mark of an educated man to look for precision in each class of things just so far as the nature of the subject admits.276 Steven Pinker said, regarding his theory of language acquisition, I fully expect that [it] will be met with some counterexamples. My defense is that an acquisition theory that faces occasional counterexamples is better than no acquisition theory at all.277 The legal scholar Frederick Schauer remarks that Universal generalizations, whether the source of the universality be definitional or empirical, tend to interest philosophers, but most of the generalizations that the rest of us employ and encounter on a daily basis are not.278

Indeed, generalization invites exceptions, the exception proves the rule, and rules are made to be broken are sayings we hear often, the first implying that all general statements are really generic, and the other two that even a rule grounds only a generic, not a universal, statement. We resort to generic statements not because of indifference to accuracy but because there is no acceptable alternative. Usually, neither a universal
252

statement understood strictly nor a particular statement or conjunction of singular statements would express what we can say legitimately when making a generalization.

Arnauld contrasted universal truths about the nature of things and their immutable essences, which admit of no exception, with universal truths about existing things, especially human and contingent events, which admit of some exception and, if we supposed that they did not, would be judged falsely, except by chance.279 The former are metaphysically universal. The latter are only morally universal, like the usual sayings All women love to talk, All young people are inconstant, All old people praise the past. But Arnauld cautioned that with respect to propositions having only moral universality we ought not to reject them as false, even though we can find counterexamples to them280 I have suggested that such propositions are much more common than those about immutable essences. But, pace Arnauld, even the latter, including those Strawson would have called analytic, may admit of exceptions and thus are in fact only morally universal.

Consider the venerable definition Man is a rational animal, meaning by man human being and by rational, let us suppose, possessing intelligence deserving to be called intellect. It states the essence of man, what a man is, and logicians properly infer from it that all men are rational, indeed that this is necessarily so, by definition. But the logicians do not mean that neonates display intelligence deserving to be called intellect. So, metaphysicians revise the definition by inserting the adverb potentially. Some
253

neonates, however, are not even potentially rational they are born with severe and irremediable mental defects. The metaphysicians may revise the definition further, perhaps by appealing (in the past) to Aristotles distinction between first and second potentiality or (today) to the genetic roots of intellectual capacities. But, if they do, they are no longer interpreting the definition, they are trying to rescue it. The original intention was just to say that men are rational animals, and both the definition and the statement inferred from it should have been so understood and then left alone.

2. Facts, Generic Facts, and Realism. When Kant declared that things as they are known by us are not as they are in themselves, he assumed that we can know things only if they are objects of sense experience, phenomena. Our focus here, however, is not on things but on the world, and not on sense experience but on description. There is a difference between things and the world: the world may contain all individual things but it is the structure, not the mere collection, totality, of them. There is also a difference between sense experience and description. Sense experience is essentially subjective and by itself barely qualifies, if at all, as knowledge. Neonates enjoy sense experience but know little if anything. Description, by contrast, almost certainly employs a public language and thus is in principle subject to the constraints of intersubjective agreement. It is taught, and when sophisticated it may count as science. Now, as I argued earlier, the world is describable necessarily, though not solely, in general statements, whether universal or generic. But even if universal statements correspond to items in the world, generic statements do not. Even if there are universal facts, there are no generic facts. Yet almost all general statements are generic, explicitly or tacitly. Cognition of the world, though perhaps not
254

cognition of individual things, requires the use of generic statements. Yet the world contains nothing to which they correspond.

In chapter 6 I sketched the traditional argument for antirealism, in both the Kantian and its more recent versions. I pointed out that, because of its forbidding level of abstraction, the argument leaves unclear what it claims and what motivates it. This is especially true of its extreme version, according to which what we think we know corresponds to nothing in the world. I suggested that we should seek arguments for a moderate version that claims only that certain essential, or especially important, parts or aspects of what we think we know correspond to nothing in the world. I suggested therefore that our focus should be on arguments from specific and readily understandable truths. The form of such an argument would be as follows: (1) We cognize (perceive, understand, describe) the world as having a certain uncontroversial and familiar particular feature F. But it is obvious (2) that the world does not, perhaps cannot, have that feature. Therefore, (3) the world as we cognize it, as it is for us, is not as it is in itself. The major defenders of antirealism, from
Kant to Goodman,

did offer also arguments of this second sort. For example, Kant pointed out

that, though our knowledge of things rests on sense perception and we necessarily conceive of them as causally related, causal relations are not themselves perceived. In Ways of Worldmaking Goodman provided a rich display of concrete examples of features of the world from apparent motion to the fairness of samples that are best understood as made by us. The argument for antirealism from the ubiquity of generic statements is also of this second sort: We think and talk about the world as containing facts that are the object of the cognitive activity of generalization, and generic statements are our chief vehicles of generalization, but there are no generic facts.
255

This is an argument for antirealism with respect to the logical structure of the world, the structure that any world must have in order to count as a world. As I argued in chapter 6, the world is a structured whole, not a mere assemblage. Astronomy, physics, chemistry, and biology focus on its spatial, temporal, physical, and causal structure. Philosophy focuses on its logical structure. The logical structure of the world corresponds to the standard classification in logic of statements as atomic, compound, and general. It thus provides for atomic facts about the properties and relations of individual things. It provides for compound facts. And it provides for general, i.e., universal and particular, facts. Nothing would count as a world if it did not allow for description by true atomic, compound, and general statements.

Nothing would count as a world if it did not allow for true atomic statements, e.g., This page is white. But also nothing would count as a world if it did not allow for true compound statements, e.g., This page is not red and If this page is white then so is the next page. And nothing would count as a world if it did not allow for true general statements, e.g., All men are mortal and There is water on Mars. But in addition to the general statements logic recognizes, i.e., universal and particular statements, there are also generic statements. A serious and knowledgeable person would say that winters in Iowa are severe, but not that all are. I have argued that generic statements are the chief vehicles of generalization, yet also that obviously they correspond to nothing in the world. This is an argument for antirealism with respect to a specific but essential part of our cognition of the world. It resembles but is not the same as antirealism with respect to the general statements that logic recognizes, which is familiar and was defended by
256

Wittgenstein in the Tractatus as part of his rejection of Freges and Russells logical realism and defense of his thesis that there are no logical objects. Wittgensteins logical antirealism or, as we have seen, more precisely semirealism, may be plausible.281 But it is not nearly as plausible as plain, straightforward, antirealism with respect to generic statements, which he did not even consider.

I have repeatedly noted that whatever the merits of the traditional arguments for antirealism from the role of cognition in sculpting reality may be, they depend on highly abstract presuppositions about reality, cognition, perception, and conceptualization that remain utterly implausible. Kant was fully aware of this. He compared his transcendental idealism to the Copernican revolution in astronomy. Despite our unquestioning acceptance of the heliocentric system, we continue to see and think of the sun as rising in the east at dawn, then slowly moving overhead in all its glory, and setting in the west at dusk. This is no ordinary, transient illusion. It is a phenomenological fact, in the sense of phenomenology expressed by Husserls slogan We must go back to the things themselves [zu den Sachen selbst!].282 The case with our unswerving acceptance of realism is similar. Putnam described the antirealist thesis as a virtual tautology. The heliocentric system is not a tautology, but it is so amply confirmed as to function in educated peoples thought as if it were. Nevertheless, the power of our belief that the sun moves overhead has remained unaffected. The fate of transcendental idealism is proof of the undying power of realism. Kant and philosophers like Goodman continue to be met with incredulity, often plain incomprehension, even by most of their philosophical readers.

257

Kants transcendental idealism was indeed revolutionary, a Copernican revolution in philosophy. It called for a radical change of perspective. So did Goodmans irrealism. But in the case of generic statements there is no perspective in need for change, no back to the things themselves! exhortation is necessary. Unless we confuse generic statements with the corresponding universal statements, we immediately see that no additional, distinctive facts make them true. No revolutionary considerations are needed to see that Dutchmen are good sailors does not say what is said by All Dutch people are good sailors or Some Dutchmen are good sailors or a conjunction such as Lodewyk is a good sailor and Rykaard is a good sailor and Hansel is a good sailor. No revolutionary considerations are needed, therefore, to acknowledge that there is no distinctive fact to which it might correspond the only relevant facts are those to which the latter statements would correspond. One who rejects generic facts need not also reject universal facts. Nor need one who rejects general facts, whether universal or generic, also reject compound facts. And the rejection of general and compound facts does not imply rejection of atomic facts. But that the intellectual activity of generalization is crucial to cognition and that the truth of general statements requires extralinguistic entities was taken for granted by Frege and Russell, though they differed regarding what these entities might be. As we have seen, both argued vigorously against the reductionist view in effect, antirealism with respect to generality according to which universal statements are just disguised conjunctions of their singular substitution-instances, a view still commonly held. General propositions such as All men are mortal stand (if true) for general facts, Russell held, and there must be primitive knowledge of some of them. But Russells argument applied to general universal statements, and he held that there are general universal facts. General generic
258

statements, much less generic facts, were totally ignored by both Frege and Russell, as they have been by virtually all philosophers. To be sure, Aristotle did note that the statement Man is white, or, as J. L. Ackrill suggests, Men are white, allows both that some men are white and that some men are not white, and acknowledged that such indefinite statements have no place in the syllogism.283 (Ackrill complains that they lack an explicit quantifier and for this reason he says, somewhat presumptuously, it is a pity that Aristotle introduces [them] at all.) Kant sharply distinguished what he called strict universality from assumed and comparative universality (through induction), which is therefore only an arbitrary increase in validity from that which holds in most cases to that which holds in all.284 Most universal statements indeed express only assumed and comparative universality, and perhaps Kant would have agreed that they are best understood as only generic. John Dewey did write about generic and universal propositions, but explained that by the former he meant propositions about kinds (general in the sense of generic), which have existential import, and by the latter abstract hypothetical propositions, which are nonexistential in import.285 Quine in effect dismissed generic statements as involving ambiguities of syntax. He wrote, Sometimes the plural form of a general term does the work merely of the singular form with every; thus Lions eat red meat..Sometimes it does the work rather of a singular with an or some, but with an added implication of plurality; thus Lions are roaring.286 It was twentieth century linguists and some legal scholars, not philosophers, who explicitly and seriously devoted attention to generic statements.

259

3. The Irreducibility of Generic Statements. Some logicians and philosophers of language have acknowledged the existence of the nonstandard quantifiers many, few, and most.287 None is reducible to the standard quantifiers all or some. Presumably, statements employing many and few defy a realist interpretation their truth value obviously depends, at least in part, on our interests and attitudes, not on any distinctive facts about what is many and what is few. Statements employing most would allow a realist interpretation, if most is taken to mean more than half. Unlike generic statements, such nonstandard general statements, though useful and common, are hardly indispensable for generalization or for cognition.

Generic statements resemble statements employing many and few by defying a realist interpretation. They also resemble them by defying reducibility to statements employing all or some. Indeed, generic statements are not reducible to any other kind of statement. Nicholas Asher and Jacques Morreau have remarked that the puzzling thing about generics [is that] their truth conditions connect them at best only very loosely with particular facts about the world, and that they entail and are entailed only by other generic statements.288 The latter is not quite true. Dutchmen are good sailors does entail Some Dutchmen are good sailors, and it is entailed by All Dutchmen are good sailors and there are Dutchmen.

But Dutchmen are good sailors does not entail All Dutchmen are good sailors, and is not entailed by Some Dutchmen are good sailors. It also neither entails nor is entailed by All Dutchmen who are sailors are, always or usually, good sailors, which was the
260

analysis Arnauld seemed to favor.289 If only two Dutchmen are sailors, their both being good sailors would not be enough to make Dutchmen are good sailors true. The statement neither entails nor is entailed by Most Dutchmen are good sailors. Most Dutchmen are not even sailors, good or bad. And, if they were, but only 52% of Dutchmen while 70% of Italians, 80% of Germans, and 90% of Norwegians are good sailors, this might not be enough to make Dutchmen are good sailors true. 52% of Americans are women, but it is not true that Americans are women. However, even if only 10% of Dutchmen are good sailors, this might be enough, as long as 2% of Italians, 3% of Germans, and 4% of Norwegians are good sailors. That the word enough is needed here indicates that we take generic statements to be true not because we find generic facts in the world that make them true but partly because of our interests and attitudes. In the 21st century Dutchmen can read and write would not be true if only 45% could read and write, but in the 17th century perhaps it was.

Dutchmen are good sailors does not entail that more Dutchmen than people of any other nationality are good sailors, absolutely or proportionally. We do not and need not compare Dutchmen with all other nationalities in order to make or accept the statement. If comparison does take place (usually implicitly), it is largely, though not wholly, up to us with whom to compare them. Instead of Norwegians and Italians, we might pick Germans and Spaniards. But perhaps we would not pick Hungarians and Mongolians, because Hungary and Mongolia are landlocked, and we might think the comparison would be unfair. At any rate, if only four Dutchmen and only two persons of any other nationality are good sailors, we are not likely to say that Dutchmen are good sailors. If only four Dutchmen and only two persons of any other nationality are graduates of the
261

Dubuque College of Cosmetology, we would not say that Dutchmen are graduates of the Dubuque College of Cosmetology.

It has been suggested that adverbs such as usually, typically, and in general are closest in meaning to the generic operator.290 This would be trivially true of in general if inserting it in Dutchmen are good sailors merely makes explicit that the statement is generic, and perhaps of typically if it is used as a synonym of stereotypically (see below). Not so of usually. How usual must it be for a Dutchman to be a good sailor if the statement Dutchmen are good sailors is to be true? It might be true even if only 10% are good sailors, as long as only 6% of Italians, 7% of Germans, and 8% of Norwegians are.

Nor, contrary to another suggestion, need the statement be saying that all Dutchmen are normally good sailors. What being a good sailor involves, say, hanging onto a swaying line in raging seas, might be abnormal for all people, Dutch or not, though admirable. Even becoming a sailor might be abnormal, in some legitimate sense of this vague word. It might conflict with emotions that are normal, such as fear of drowning. At any rate, as Gregory Carlson has conclusively pointed out, generic statements can also be made about normal kittens and abnormal drunk physicians. Are we to take such statements to be about normal normal kittens and normal abnormal drunk physicians?291

Shall we say, instead, that Dutchmen are good sailors means that all Dutchmen are good sailors in normal circumstances? But what are these circumstances? Sailing on merchantmen or sailing on warships? Serving under demanding or serving under easy262

going shipmasters? Short or long voyages? Perhaps people even become sailors mainly when the economic circumstances are abnormal. Being a sailor might be attractive mainly in such circumstances. But, again, what are these circumstances? High unemployment in the Netherlands, or high unemployment just in its coastal areas? What such examples show is that no brute fact makes Dutchmen are good sailors true. But this does not mean that the statement is subjective. Although it was of special importance to shipmasters, its truth was not dependent on their personal wishes. It might have been accepted by all other people in a position to know, e.g., first mates and ship owners. Its truth was objective in the proper sense of being intersubjective, agreed to by competent judges, people with knowledge and open mind about navigation. It was not what Kant called mere fancy. As Carlson says, we know that not all dogs bark, but also we know that Dogs bark is true. He adds, the knowledge that there are three-legged rabbits does not falsify the statement that rabbits have four legs.292 If we say that nonetheless Dutchmen are good sailors is not really objective, we must mean that it did not admit of a realist interpretation, that it did not correspond to a fact. But this is exactly what I have argued.

Generic statements have been called vague, but their vagueness is unlike that of predicates. According to Peirce, A proposition is vague when there are possible states of things concerning which it is intrinsically uncertain whether, had they been contemplated by the speaker, he would have regarded them as excluded or allowed by the proposition. By intrinsically uncertain we mean not uncertain in consequence of any ignorance of the interpreter, but because the speaker's habits of language are indeterminate.293 Perhaps generic statements are indeed vague in this sense. But what Peirce had in mind was
263

vagueness of propositions due to the presence of vague predicates like bald the quantity of hair on a persons head may be such that it is intrinsically uncertain whether a speaker would apply the predicate to it. Generic statements are not vague for such a reason. Their vagueness is not due to a vague predicate. Good sailor may be a vague predicate, but this would not be the main reason Dutchmen are good sailors is vague. It is vague because of its logical form. It would be vague even if we replaced the predicate good sailor with a predicate that is not vague. This is why generic statements are useful, indeed indispensable. Predicates such as bald are also useful and perhaps indispensable because they are vague. But their vagueness is different from that of generic statements.

Generic statements have been said also to be inexact, imprecise. Again, this is true, but how we understand it calls for caution. The inexactness of a generic statement is not due to the presence in it of an inexact word. The statement Jack is here is inexact, but if we wished we could state Jacks location with reasonable precision by saying, e.g., Jack is in the kitchen, and might be happy to replace the former statement with the latter. In the case of Dutchmen are good sailors, however, an attempt at precision is likely to yield a statement that, whatever its merits, we would not put in place of the original. Either it would significantly differ in truth value, as All Dutchmen are good sailors would, or it would not be even a general statement, as a conjunction of statements of the form x is Dutch and x is a good sailor would not.

Carlson distinguishes inductively established correlations from real rules or regulations, associating generic statements with the former and universal statements
264

with the latter.294 It is unclear what he means by real rules or regulations. But his phrase inductively established correlations is reasonably clear. Its use implies that, as Kant might have put it, generic statements possess at most assumed universality. Of course, Kant had in mind universal, not generic, statements, and, as we saw earlier, contrasted those possessing only assumed and comparative universality, through induction with statements possessing strict universality, meaning that they are also necessary and a priori .295 But generic statements, however they are established, lack even assumed universality this is why they are generic.

Arnauld would have said that universal statements established inductively are only morally universal. Russell and most other epistemologists in effect have agreed: they are only probable. According to Russell, even if the sun rose every day in the past, it is only probable that it will rise tomorrow. (He wisely avoided assigning a numerical value to that probability!) This was the problem of induction. Indeed, reasonable people seldom expect inductive reasoning to yield more than a generic statement, unless it also involves causal information. We appeal to what history teaches e.g., in predicting election results, hurricanes, and the gyrations of the stock market precisely when we lack such information. In both everyday and scientific reasoning, induction unsupported by causal information, what might be called primitive induction, is usually taken to justify only generic statements, as the frequent occurrence of the phrase ceteris paribus shows. This is why scientific writing routinely includes caveats such as The precise mechanism through which fluticasone propionate affects allergic rhinitis symptoms is not known. The closer scientists are to field or lab, the less willing they are to venture universal statements.
265

Nevertheless, though based only on induction, Dutchmen are good sailors does not mean that all Dutchmen are probably good sailors, as Russell might have said. If only 10% of Dutchmen are sailors, it would be false that all Dutchmen are probably good sailors, whether in the statistical or in the epistemic sense of probably. But it might still be true that Dutchmen are good sailors if, say, only 6% of Italians, 7% of Germans, and 8% of Norwegians are good sailors. Nor does the statement mean that all Dutch sailors are probably good sailors. It might be true even if only 40% of Dutch sailors are good sailors, as long as, say, only 15% of Norwegian, 14% of German, and 13% of Italian sailors are good sailors.

Asher and Morreau say that it is reasonable to infer from Fs are G that something is G given that it is F. But inferring from Dutchmen are good sailors that Maarten is a good sailor given that Maarten is Dutch would not be reasonable if, as surely is the case, less than 50% of Dutchmen are sailors.

I noted earlier that the suggestion that Fs are G means All Fs are typically G might be acceptable if typically is understood as a synonym of stereotypically. Indeed, a common complaint about generic statements is that they involve stereotyping, misrepresentation or at least exaggeration of the facts. The complaint targets mainly generic statements that, like Arnaulds examples, concern nationality, gender, race, age, or religion. These are sensitive matters, and people care deeply how statements about them might be intended or understood. Many resent, even find insulting, that such statements are made at all.
266

Stereotyping is a pejorative today, and it does apply to generic statements involving abuse of conceptualization or classification. But it is misplaced if applied to all generic statements. Like most conceptualization and classification, most generic statements are innocent. New York Times columnist David Brooks writes, I ... believe most of human thought consists of stereotypes. Im not against stereotypes; Im against crude stereotypes. If we say that all generic statements involve stereotyping, then we must say that so do almost all universal statements, since almost all are intended and understood as though they are only generic. That the conceptualization generic statements involve is sometimes abused counts against them no more than the frequent abuse of inductive reasoning counts against induction. Indeed, abuses of generic statements, including those charged with stereotyping, are usually just abuses of induction. One is more likely to be struck by lightning than to be attacked by a shark, resort managers in the Bahamas say, and this is true, but it does matter whether one is swimming in the ocean or sleeping in a hotel bed. The type of inductive reasoning exemplified in We dont need fire insurance because weve never had a fire is unfortunately familiar.

Deductive reasoning, too, is frequently abused. The validity of a deductive argument is not a matter of choice, but its premises are. One can prove deductively any proposition if free to choose the premises. It is not the validity of the standard proofs of the existence of God that usually has been questioned if not already valid, they can be rendered valid by adding suitable premises but the truth of their premises, e.g., that existence is a property (perfection, real predicate), that the universe has a cause, or that the human eye manifests intelligent design.
267

The variety of antirealism defended in this paper is modest and measured. It does not deny the reality of individual things or even the reality of atomic, compound, and universal facts. It denies only the reality of generic facts. This would hardly cause common sense to rebel. Even philosophers have not claimed that there are such entities. Yet, if cognition of the world requires the intellectual activity of generalization, and generic statements are the chief vehicles of generalization, then this modest and measured variety of antirealism has much of the bite of traditional antirealism. Moreover, since it is based solely on the ubiquity and irreducibility of generic statements, it is more plausible than standard antirealism. It is also less obscure.

Nevertheless, we must end this chapter with two important caveats. The first is that while generic statements cannot be given a realist interpretation, they resist an unqualifiedly antirealist interpretation. There is no fact to which Winters in Iowa are severe corresponds, but the statement is ordinarily taken to be true while Winters in Florida are severe is not. The reason, of course, is that, though equivalent neither to universal statements nor to singular statements or conjunctions of singular statements, generic statements would not be regarded as true were it not for the truth of some singular statements. But they are not equivalent to the corresponding particular statements. What Winters in Iowa are severe says is quite different from what Some winters in Iowa are severe says. Yet it does entail the truth of the latter. Of course, the converse does not hold: Some winters in Iowa are severe does not entail Winters in Iowa are severe. Even if Some winters in Florida are severe were true, Winters in Florida are severe would not be. Generic statements have no relationship to generic facts because there are
268

no such facts, but they do have a relationship to singular facts, even though they are not reducible to statements asserting those singular facts. The relationship between generic statements and the relevant singular statements is not as close as that of universal statements to their singular instances. As Wittgenstein pointed out, while the former do not say what the latter assert, they do show it. But even this much cannot be said in the case of generic statements. Even a semirealist interpretation of them cannot be defended. All winters in Iowa are severe entails The 1903 winter in Iowa was severe, while Winters in Iowa are severe does not. But neither can an unqualifiedly antirealist interpretation be defended. Unqualified antirealism has application only to statements that are unqualifiedly false. And many generic statements, like Winters in Iowa are severe, are true, not false, for obvious empirical reasons. But these reasons do not include experience of generic facts.

The second caveat is that while generic statements are the common means of generalization, we understand them nonetheless as approximations of universal statements. The word all remains the paradigmatic expression for generality. This is why the primacy of generic statements may be said to be pragmatic, not logical or semantical. Patients with prior strokes benefit from taking Lipitor is indeed different from All patients with prior strokes benefit from taking Lipitor, but the latter remains our guide to understanding the former. The same can be said, respectively, of Patients with prior strokes do not benefit from taking Lipitor and No patients with prior strokes benefit from taking Lipitor, even though both are false. Generic statements are not universal statements but we necessarily understand them as secondary to universal statements.
269

Chapter Nine: THE PHENOMENOLOGY OF LOGIC

1. Logical experiences.

Logical cognition seems to correspond to no distinctive entities. In the case of universal, particular, and molecular statements we have found that at most semirealism is warranted. In the case of generic statements, full-fledged antirealism seems inescapable. But, surely, there is more to logical cognition than words. Surely, there is a connection between logical expressions and what we perceive or think. We employ logical expressions, even if only the verb to be, in all statements we make, in science as well as in everyday life, not just in logic. Few, if any, statements fail to contain logical expressions. Yet we also take for granted that all true statements are true because the world is as they say it is, and that all false statements are false because the world is not as they say it is. To acknowledge this need not be to accept a theory of truth as correspondence to fact. It is just to acknowledge what seems obvious, that as Aristotle said, To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, and of what is not that it is not, is true.296 How does logical antirealism, whether full-fledged antirealism or semirealism, deal with this fact?

We face here a predicament analogous to the one Wittgenstein faced in his later philosophy after he denied that the use of words for sensations depends on anything they
270

refer to. Such words refer neither to anything outer such as behavior, nor to anything inner.297 But Wittgenstein was aware that so far he had told us only what their use does not depend on. What does it depend on, then? The answer that it depends on nothing is deeply unsatisfactory, even if true, as most readers of Wittgenstein would testify. Wittgenstein offered an answer that appealed to his notion of a defining criterion, as contrasted with a mere symptom. It replaced the notion of reference in the case of words for sensations, yet it did stand for a connection between their use and something else.

Holding ones cheek in a certain distinctive way, Wittgenstein wrote, may be a criterion for saying that the person has a toothache, in the sense of grounding the learning and then governing the use of the expression.298 On the other hand, the persons facial expression may be just a symptom. Holding ones cheek and having a toothache, of course, are not the same. They are only connected in some way. But we do not infer this connection from our own past experience such inductive reasoning would not be serious, since it would necessarily be based on only one case.299 Nor do we infer the connection from other persons past or present experience we are not aware of other persons toothaches. Perhaps holding one's cheek is not a criterion for the use of toothache (Wittgenstein mentioned it only as a possible example), but surely some patterns of behavior, however complex, are such criteria, in the sense that, though logically neither sufficient nor necessary for the correctness of the use of toothache, if they were not at least occasionally satisfied that use would be bewildering and presumably never mastered. A person can have a toothache without holding a cheek or doing anything else connected with having a toothache, and a person can smile, sing, and dance despite an aching tooth,
271

but if this were usually or even often the case we would have doubts not only about the truth of the assertion that the person has a toothache but about its being a proper use of language.

The logical antirealists predicament is even closer to one that Kant faced: There seems to be no connection between what Kant called the pure concepts of the understanding (e.g., the concept of causality), which according to him are a priori and necessary for cognition of the world of experience but stand for nothing in experience, and the world of experience to which nonetheless they apply. To deal with this problem, Kant proposed his doctrine of the schematism of the pure understanding. He wrote, [P]ure concepts of the understanding being quite heterogeneous from empirical intuitions, and indeed from all sensible intuitions, can never be met with in any intuition. For no one will say that a category, such as that of causality, can be intuited through sense and is itself contained in appearance. How, then, is the subsumption of intuitions under pure concepts, the application of a category to appearances, possibleWe must be able to show how pure concepts can be applicable to appearancesObviously there must be some third thing300 What was this third thing? A few pages later Kant explainrd that, in the case of causality, The schema of cause, and of the causality of a thing in general, is the real upon which, whenever posited, something else always follows.301 This, of course, is what Hume had called constant conjunction. The pure concept of causality must be understood as related to the concept of constant conjunction if it is to have application to experience. Thus Kant agreed with Hume that constant conjunction is what we must depend on in applying the concept of causality, that it mediates the application of the concept, but of course he sharply denied that it is the same as causality. Indeed, Hume
272

also denied this. He insisted that the idea of causation involves the idea of necessary connection, which is not the same as constant conjunction.

Such were Wittgensteins and Kants ways of dealing with certain difficulties they faced. Logical antirealism faces an analogous difficulty: there seems to be no connection between the use of logical expressions and what we perceive or think. We cannot deal with this difficulty by following Wittgenstein or Kant, for they were concerned with very different topics and relied on psychological and linguistic assumptions that are irrelevant to our topic. Instead, I shall appeal to certain distinctive experiences as the third thing that seems to mediate between the use of logical expressions and what we perceive or think. We may call them logical experiences (though not in the sense Russell used the phrase in Theory of Knowledge.302) They are associated with the expressions and govern their use, consciously or unconsciously. They anchor that use. But they are not experiences of logical objects, or of any other objects. They need not occur whenever the expressions are used, just as one need not be holding ones cheek whenever one has a toothache. Language is too subtle to conform to such rigid requirements. They are experiences in the ordinary, natural, and innocuous sense of experience, not the philosopher's or the introspective psychologists technical and suspect sense. In that ordinary sense the paradigms of experience are joy and misery, pains and pleasures, itches and shivers, not anything allegedly exemplified by all perception such as sensations or sense data. When one sees a tree, one does not experience a tree.

What are these logical experiences? We should not assume that there must be a different one for every logical expression. We need not be logical reductionists in order to allow
273

that some logical expressions are genuinely reducible to others. Nevertheless, several logical experiences are familiar and arguably also central because they are associated with logical concepts that are central.

Consider the concept of identity expressed by the verb to be in one of its senses, e.g., in This is the dog I saw yesterday. It is essential to our cognition of what we perceive or think but stands for nothing in what we perceive or think. If identity were something that we perceive or think, presumably it would be a relation, but, as Hume, Hegel, Wittgenstein, and many others have pointed out, no such relation is observable or indeed imaginable, even when supposed to hold between observable or imaginable things.303 You saw a dog yesterday, you see a dog now, and you correctly judge that they are identical, that they are one and the same dog, but you dont see or remember seeing a relation of identity between them. Nor do you establish that every property that one has the other also has and then infer that they are identical from their indiscernibility. Nevertheless, to say that there are only the linguistic expressions for identity, whether the plain is and same or the fancy identical and =, would be misleading because identification in the form of recognition occurs in humans even before the acquisition of language, indeed it seems to occur in animals incapable of language. Even if we suppose that in nonverbal or preverbal recognition we must apply some concept to the object recognized, e.g., the concept of dog, this would not be the same as applying the concept of identity. If a prelinguistic child applies the concept of dog to what the child sees, the child need not also be applying the concept of identity to the dog seen now and some dog seen earlier.

274

However, though standing for nothing, the concept of identity is associated with a well known experience, the experience of familiarity, which moreover precedes the acquisition of that or any other concept. Not coincidentally, familiarity is also the experience essential to recognition. And it is the bridge between cognition as mere recognition and cognition as verbal description. It serves as a steadying anchor for our use of the identity expressions same, identical, and =, and like expressions they would not have been learned without it. (Even if logical concepts are innate, the expressions for them in language are not.) The experience of familiarity is not itself identity, it is not what the expressions for identity would denote if they did denote. Nor is it consciousness of some relation or property identity, for there is no such relation or property to be conscious of. And it is not the mere application of a concept or the utterance of an expression, even when it accompanies such an application or utterance. Familiarity is an experience, something one feels or has, not something one does.

Indeed, because of its role in recognition the experience of familiarity enjoys special dignity. It is involved in the acquisition and application of all concepts, not just that of identity. To acquire and then apply the concept of dog we must be able to find certain objects familiar, whether dogs or pictures of dogs. To learn and then use the word dog we must find certain phonemes familiar, we must recognize them.304 So there is every reason to believe that the experience of familiarity occurs earlier than conceptual cognition and the acquisition of language, even though it bears intellectual fruit only when concepts are acquired and expressed in language. Moreover, a sophisticated command of a language requires more than recognizing and learning words it requires also being able to make identity judgments about the words themselves, most obviously
275

when a word is pronounced or inscribed in different ways. The word dog is the same word as DOG, even though on this page at least two of the letters look very different. Of course, the level of sophistication that knowledge of such matters involves is not lofty. It is what we call literacy.

Unfamiliarity, strangeness, is also a distinctive and familiar experience. It is associated with the use of identity expressions in negative identity-statements. It is not just the nonoccurrence of familiarity. It is a genuine (often profoundly disconcerting) experience in its own right. Its importance ought to be evident but is often overlooked. Indeed, there could be no cognition of a world in which nothing is familiar, as Plato pointed out in the Theaetetus when arguing against Protagorean skepticism. But also there could be no cognition of a world in which nothing is unfamiliar. A world is something we explore, and strangeness both prompts the exploration and sometimes is faced in its results. In general, we must allow that logical experiences have opposites, contraries, which too are logical experiences and may be no less important.

There seem to be experiences associated also with at least some of the other logical expressions. They too serve as steadying anchors in our use of those expressions. There seems to be such an experience associated with the expressions for implication (ifthen, ). In his noteworthy discussion of what he called the logic of signcognition, H. H. Price wrote of a feeling of if, and suggested that it arises through the experience of questioning or doubting.305 And Sartre famously held that questioning is what introduces nothingness in the world.306 There is important insight in both views, which complement each other, and I shall return to them in
276

connection with the concept of negation. But questioning is too intellectual for our present context, though neither Price nor Sartre understood it as intellectual. I suggest, instead, that expressions for implication are associated at the deepest level with the experience of expectation, of which also cats, rats, and numerous other species are capable. Expectation is not predicting or imagining a future event it is better described as a feeling, the feeling of expectation. It surely occurs in infants as well as nonhuman animals. Whether disappointed or fulfilled, expectation includes consciousness (perception, thought, imagination) of what is expected, and often also of its ground or basis. Expectation of rain includes thinking of rain, and may also include seeing dark clouds. But the feeling of expectation is distinct from from both the thinking and the seeing. One may see or think about dark clouds but expect nothing. This becomes clearer when we distinguish (1) pure expectation, which does not even appear to have a ground (I just know it will snow tonight), (2) inferential expectation (its cold, so it will snow), in which the ground is explicit, and (3) conditional expectation (if its cold, then it will snow). The experience, feeling, of expectation is palpable in (1), but not in (2) or (3). Yet (1) is usually a guise of (2), which seems to be the original phenomenon of expectation. Of course, it is (3) that explicitly involves the logical concept of implication, and often if not always it is purely intellectual in character, empty of feeling. But (3) is implicit in (2): corresponding to every inference, there is a conditional proposition of which the antecedent is the premise(s) and the consequent is the conclusion of the inference.

Negation seems associated with disappointment, an experience of a striking and much discussed character, vividly described by Sartre in a famous example. Looking in a caf
277

for someone you eagerly expect but fail to find need not involve the intellectual performance of making the negative judgment that the person is not there, which one could do even if not expecting to find the person. But commonly it involves a distinctive and much too familiar experience, a feeling usually the feeling of disappointment, in one or another of its many degrees and forms.307 On the other side of the English Channel but about the same time, H. H. Price wrote, Disappointed expectation is what brings NOT into our lives.308 The feeling of disappointment is not the same as the feeling of expectation. Disappointment is associated with negation, while expectation is associated with implication. Nevertheless, they are obviously and intimately related.

Looking for a certain book on a bookshelf but not finding it is often accompanied by a characteristic experience that is commonly described as seeing it is not there or, less commonly, as seeing it is not one of the things there. Of course, it is not an experience of an object that might be called negation. To suppose that it is would be crude and simplistic. Since it is an experience, its occurrence is not independent of us. But surely it is independent of our use or even possession of language, indeed even of concepts. Presumably, the lower animals are also capable of it. It is not independent of sensibility insofar as it involves a form of sense perception, which it need not. Instead of looking for a book, I may be looking for an answer to a philosophical objection, and failing to find one I may have the intellectual experience of there not being one. Heidegger and, later, Sartre made central use of an experience associated with both negation and implication. They called it anguish, but meant by this awareness of ones future, at least the future one expects or desires, as something that is not, but will be if one makes a certain choice now.

278

Disappointment and expectation constitute the experiential core of sentential logic: negation and implication are the naturally (not necessarily formally) primitive sentential connectives. Like familiarity and strangeness, they also seem essential to any life deserving to be called cognitive. But so does also surprise, which presumably is the opposite of expectation. If the world is the world we live in (which other world might it be?) then we may say that these experiences are essential also to the world as it is for us. But though we may say this, it is not entailed by what preceded it. Nor is it, of course, a tautology. Those who disagree with what I have said, or with Kant, are not guilty of inconsistency.

There seem to be distinctive experiences associated also with other propositional connectives. In the case of disjunction, there is the experience, often painful, of indecision, of facing a choice, especially when the choice, the decision, must be made, sometimes heartrendingly. As to conjunction, Gustav Bergmann drew attention to an experience associated with it by remarking that there is an obvious distinction between two simultaneous awarenesses, one of this, the other of that, on the one hand, and a single awareness of this-and-that, on the other.309

Are there distinctive experiences associated with generality and thus constituting the experiential core of the theory of quantification in logic? In the case of the particular, existential, quantifier (some, there is, ) we may be tempted to say that it is the experience of existence. But even if there were such an experience, this would be a misunderstanding of the quantifier, despite what conventional philosophy tells us. To say that there are many things Jack fears is not to say that they exist, that they are real. Often,
279

just the opposite is the point of saying it. Some of the things we fear are real but (fortunately) many are not, though (unfortunately) we do fear them nonetheless. To say that there are many things Jack imagines is not to imply that these things exist; what is more likely to be implied is that they do not. But whether or not we follow convention and restrict quantification to existent objects (beings, entities), there does seem to be a characteristic experience associated with the particular quantifier. It is the experience of presence, real or imaginary. An example might be the experience of the presence of the Times on the rack when I look for it and find it. It should not be confused with my seeing the newspaper or with seeing that it is there. I may see many newspapers on the rack and perhaps also see that they are there, but usually I would experience nothing related to them. Another, rather depressing, example would be the stubborn presence before the mind of something one fears, imaginary though it usually is, e.g., a fatal automobile accident involving ones child.

The experience of presence, associated with the particular quantifier, is the opposite of disappointment, which is associated with negation. This suggests that the experience associated with the universal quantifier is the experience of absence, for example the absence of the Times from the rack when one looks for it but fails to find it. It need not be the experience of disappointed expectation (one might have hoped but not expected to find the Times), but surely the two experiences are closely related. This may be why we find plausible the interdefinability of the universal quantifier and the particular quantifier by way of negation, or at least the equivalence of (x) x and ~( x) ~ x. (E.g., if everything is material then it is not the case that something is not material, and if it is not the case that something is not material then everything is material.) Obviously, the
280

experience of absence should not be confused with seeing the newspapers that are there. Nor should it be confused with the judgment that the Times is not among the newspapers that one sees there, which is an intellectual performance one can engage in with respect to newspapers one never looks for.

When using the words presence, absence, and being-there, my intention was not to allude to such philosophers as Heidegger and Sartre. I have relied on the ordinary senses of those words. For example, in its ordinary sense presence is not confined to the present. We may speak today of a persons presence at a meeting yesterday or of the persons presence at the meeting tomorrow. But, obviously, much more needs to be said about the logical experiences with which the logical expressions are associated. I have only scratched the surface. No attempt has been made here to provide an exhaustive taxonomy. Fortunately, much has already been done by others, with Heidegger and Sartre indeed obvious examples but also many of those they influenced. We should not ignore, however, the trail blazing philosopher-psychologists whose work preceded continental phenomenology. For example, William James dwelt in detail on the richness of what he called the stream of thought, the place of language in it, the role in it of relations and not just of substantives, and on the inadequacy of both sensationalism and intellectualism. He acknowledged the occurrence of a feeling of and, a feeling of if, a feeling of but, and a feeling of by, and pointed out the dependence of the thought of something as existent extra mentem on repeated experiences of the same.310

The fact that there are experiences associated with logical expressions makes evident that the role of language in cognition is limited in ways that were ignored by the philosophers
281

who took the linguistic turn. Indeed, those expressions correspond to no objects. But it is not true that all there is to their use, and thus to logical cognition, is language. There are also the associated experiences. This may be why it strikes us as incredible that logical, indeed any, cognition might be just language. There are also experiences, and these are not language. This is why the linguistic turn in philosophy ought to be even more confined than I urged earlier. It ought not to be purely linguistic even where it is most plausible in logic. Logical experiences are essential to essential not only to the uses of logical expressions but to human life and the world in which we live. The world is not the earth viewed from the moon. We are immersed in it. It is a world of action. It is a world of familiarity and strangeness, expectation and disappointment, presence and absence. It is to this world that the logical experiences are essential. Is this not enough to explain why they serve to anchor the logic of our cognition of what we perceive or think in what we perceive and think, to keep logic down to earth?

I have argued that identity is a logical concept that is not necessarily linguistic, because it is presupposed in recognition and thus in the acquisition of language itself. Although there is no relation or anything else in what we perceive or think that can be called identity, identity can be said to correspond in a special way to all entities that we perceive or think about. If existence is indefinite identifiabilty, as I have argued elsewhere,311 then it is in virtue of identity that any objects of awareness, whether perception, imagination, or thought, are conceptualized into entities, objects that exist or have being. This is why identity is the fundamental concept, the one presupposed by the possession of any other

282

concept. It is presupposed by the concept of existence, which I have argued must not be confused with the concept of generality, which is expressed by the particular quantifier.

2. The Need for Language. We have now arrived at a measured, cautious, and discerning version of metaphysical antirealism, which unlike Kants is almost thoroughly linguistic (the concept of identity being the exception) and unlike current versions is limited to the logical structure of the world. Because of its recognition also of the essential role in logical cognition of certain experiences of the world, it can be only a version of antirealism. Nevertheless, it has very much the bite that any properly motivated antirealism might have. Logical structure, though not the substance, is hardly an accident of the world. The things structured, the objects, are the substance, as Wittgenstein wrote in the Tractatus: Objects make up the substance of the world (2.021), Empirical reality [Realitt] is limited by the totality of objects. (5.5561). This is why Aristotle charged the science of being qua being first with the study of the most certain principles of all things, the principles of the syllogism. What is true of logic directly affects what is true of being, or, in the mundane terminology of current philosophy, of the world.

This is why the word antirealism is too strong for the view we have reached, though there is no convenient alternative to it. Our antirealism holds that logical expressions stand for nothing in the world, but it also holds that they are intimately associated with certain experiences of the world. In the previous chapter we found the position regarding
283

universal statements that Wittgenstein and Gustav Bergmann reached was similarly nuanced, though also very different.

Our antirealism acknowledges the virtual tautology, eloquently emphasized in recent philosophy by Wilfrid Sellars, that nothing unconceptualized can be the content of judgments or statements, and thus serve as evidence or enter in other epistemic relations.312 But, unlike most current versions of antirealism, it does not deny the need for a distinction like Kants between things-in-themselves and things-for-us. It avoids what might be called linguistic creationism, the heady view that there is nothing that is not conceptualized or verbalized. As Kant remarked, the view that there can be appearances without anything that appears is absurd.313

Nor does our antirealism deny, on the side of things-for-us, the difference between what Kant called sensibility and understanding. That there is such a difference is evident, however difficult it may be to state. We might say that understanding is up to us, while sensibility is not, but this, though pointing in the right direction, would be misleading or at least excessively vague. Let us say, instead, that we have some idea of how we may choose to conceptualize differently the things we find, but not of how we may choose to find different things. The only objective criterion of reality is coerciveness, in the long run, over thought, William James wrote.314 The logical experiences of unfamiliarity, disappointed expectation, and absence make this coerciveness especially vivid. They occur in the coercive context of things as we find them, not of things as we make them.

284

The occurrence of logical experiences may prompt us to revisit the question whether a person could be engaged in logical cognition but lack the use or even grasp of logical expressions. To answer this question responsibly we must be able to describe such cognition clearly and in detail. We have no such description. Speculation that God, angels, or extraterrestrials might enjoy it does not suffice. But by the very nature of the method I have advocated, we must allow for its possibility. For we are engaged in offering not definitions or conceptual analyses but descriptions. And these descriptions ultimately are empirical and thus capable of being disconfirmed.

Our antirealism is limited to the dependence of logical cognition on language. And on this topic relatively little empirical work has been done. One reason perhaps is the relative obscurity and unfamiliarity in science of the philosophical concept of logic. This is why, even if we agree that cognition of the world involves much that is not in the world, we may be unable to say why it requires language. There is just our sheer inability to think of another alternative, once we have acknowledged that logical objects are not objects of any sort of consciousness, and have also refused to sweep difficulties under the rug by appealing to what is in the mind, as Kant did, or in the brain, as current naturalism does. That logical cognition depends on language seems to be the only available hypothesis. To be sure, it is an empirical hypothesis. It can be rejected, but first one must state what the alternative would be and then finds evidence for it. We cannot place ourselves, whether in imagination or thought, much less perception, in a situation that allows for a test of the hypothesis. We know what it is like to employ language, even if only subvocally or in imagination, when we confront, say, situations requiring description in general statements, but we do not know what it would be like to confront
285

them without language. To know the latter, we would need the innocence of infancy, and we are neither innocent nor infantile. Our ignorance in this respect is comparable to our ignorance about the cognitive life of nonhuman animals, perhaps even the cognitive life of other adult humans. We need not be solipsists in order to acknowledge the severe limitations of what we know about other minds, nonhuman or human. Such knowledge must rest on observation only of their behavior or their brains. but knowledge of our own cognitive life rests on neither.

As we saw in chapter 5, even though psychologists and linguists have ignored the specific question of the dependence of logical cognition on language, they have investigated the general topic of the dependence of thought on language extensively. But we found the results even of that investigation disappointing. I suggested that the main reason is that what is especially needed is empirical study of language acquisition in children, and this is severely constrained by the impossibility of serious but ethically acceptable experimentation. This is why psychologists and linguists have often limited themselves, if not to quasi-philosophical speculation, to bare observation and memory reports by special adults, e.g., Jamess Mr. Ballard. And these are difficult to evaluate. In chapter five I considered some examples. But I ignored the cases, also found in the literature, of adults (e.g., Genie315) who were deprived of normal language acquisition as children and never mastered a language adequately enough to provide us with useful reminiscences. Such cases are in various respects too unusual to support generalizations, even though they seem to confirm the view that syntax, though not semantics, has an inherited biological basis. The adults in question often seem to exemplify rather sophisticated thought even in the absence of grammatical competence. Observation of
286

nonhuman animals may lead to a similar conclusion. Cannot a cat see that there is no milk in the saucer, or that all of the cats kittens are in the room? Even if a cat cannot do this, it seems absurd to claim that Genie or a normal prelinguistic child could not.

To admit that there can be cases of logical cognition without language is to allow a major qualification of the chief claim of this book. But it would not affect its main point: that logical cognition corresponds to no relevant objects in the world. At any rate, there is little room for philosophical judgments regarding such cases. They make vivid the modest character of any philosophical claim about the dependence of cognition, logical or not, on language. Empirical claims about it, of course, can lay no claim to invulnerability to exceptions. The topic is empirical, though some aspects of it admit of, indeed require, philosophical treatment, namely, those concerning the existence of logical objects. The modesty of our conclusions is not incompatible with perseverance in further empirical research.

The title of this chapter is The Phenomenology of Logic. The term phenomenology, of course, is notoriously ambiguous. If taken to stand for an independent discipline, that discipline would be a part of empirical psychology, and therefore anthropocentric. If it stands for a particular philosophical method, phenomenology would coincide with the whole of philosophy. The latter was the rich but nondoctrinaire sense of phenomenology expressed by Husserls slogan Back to the things themselves! i.e., back to what we actually find before us, rather than what philosophical or scientific theory, or even common sense, says is there. So understood, phenomenology would be anthropocentric only in the trivial sense that, like all cognition we engage in, including
287

the thesis of antirealism itself, it is human. This is why my appeals to logical experiences in this chapter, and phenomenological appeals in philosophy generally, are not incompatible with the rejection of anthropocentrism. They encroach on the province of no science.

Chapter Ten: FACTS AND TRUTH

1. Realism and antirealism regarding facts. The position we have reached so far does not question realism regarding things (ontological realism), and it accepts semirealism (as Wittgenstein and Bergmann did) regarding atomic, compound, universal, and particular statements. It is antirealist only regarding generic statements. Nevertheless, given the ubiquity of generic statements in any cognition of the world, the position does seem committed to cosmological antirealism. But it is not an all-embracing cosmological antirealism like Kants. It says nothing about the physical structure of the world, or about space, time, and causal order. Nor is it even unqualified logical antirealism: It acknowledges the association of logical expressions with certain experiences. The resulting position is thus moderate and inoffensive to common sense, the mature and thoughtful judgment we all share and must respect even if not accept. Neither semirealism regarding compound and universal facts nor antirealism regarding generic facts is or entails antirealism regarding anything dear to common sense.
288

Nonetheless, as I noted in the Introduction, sentences themselves, not just the sentential connectives and the quantifiers in them, are logical expressions. The logical objects for which they supposedly stand are facts. Now, in addition to the sorts of logical antirealism we have considered so far, there is antirealism with respect to facts as such, with respect to all facts. It questions the legitimacy of the very category of fact, and thus of any theory of truth as correspondence to fact. It is the topic of this chapter.

The case for accepting the reality of facts is straightforward, though seldom properly made by realists. It is essential to cosmological realism, even if not to ontological realism. My example of four worlds containing the same things, including Jack, Jill, and the relation of admiring, and different only in respect to facts about Jack, Jill, and the relation of admiring, should have made this evident. If Jack admires Jill but Jill does not admire Jack, only the fact that this is so, not the mere presence of Jack, Jill, and admiration in the real (actual) world, would distinguish that world from a world in which Jill admires Jack but Jack does not admire Jill, a world in which they admire each other, and a world in which neither admires the other, if these four possible worlds differed in no other respect. If the world is the totality of facts, as Wittgenstein thought, we cannot be cosmological realists yet deny the reality of facts. If there were no such entities as facts, there would be also no world. There might still be things, but realism regarding things does not entail realism regarding the world.

Facts are thus indispensable for our very understanding of the dispute between realism and antirealism, as Richard Fumerton has forcefully argued.316 Defenders of realism often
289

rest their case on the obvious reality of material objects: the existence of ones hands, of a tree in the park when no one sees it, of the sun and the earth before there were human beings, and so on. But the realism they defend and the antirealism they oppose are ontological, not cosmological, they are realism and antirealism regarding things, not regarding the world. Israel Scheffler objected to Goodmans Ways of Worldmaking by saying that surely we didnt make the stars.317 Schefflers objection was simplistic, and he failed to notice the very title of Goodmans book. To acknowledge the mindindependence of stars is not to acknowledge the mind-independence of the world. We may remain realists regarding things, including stars, even if we abandon realism regarding the world. We may be both ontological realists and cosmological antirealists.

Ontological antirealism is certainly repugnant to common sense, but cosmological antirealism is not because it questions only the reality of the world conceived as the totality of facts, a conception purely philosophical and too recherch for common sense, which in any case seldom contemplates the world and when it does finds it rather mysterious. Had cosmological antirealism questioned the reality of the world conceived as the totality of things, which is the commonsense conception, insofar as common sense has such a conception,cosmological antirealism would have been no less repugnant than ontological antirealism. To deny that there are things would be to accept a metaphysical antirealism that denies even the reality of a childs animals, vegetables, and minerals.318 It is a position, as Hume saw, doomed to produce no conviction. It is so implausible that the few philosophers, perhaps Berkeley, who seem to have held it are viewed by nonphilosophers with astonishment.

290

Are there facts? Is there a strong case for realism regarding facts? In defense of his ontology of facts, Reinhardt Grossmann wrote, Among the furniture of the world, there are quite obviously facts. It is a fact that the earth is round, a fact that whales are mammals, and a fact that 2 plus 3 equals 5.319 But Grossmann also held that facts are abstract entities, like properties and relations, which he regarded as universals: While an apple may be red at time t, the fact that this is so is not itself located at t or any other time, nor is it located in space where the apple is located or anywhere else.320 Grossmann did add: There is only one world, a world in which universals and particulars dwell side by side in harmony, and that what keeps this world togetheris the unity of the fact, but he also cautioned: The problem is not how universals can be kept tied to particulars, but how facts are to be analyzed into their constituents.321 David Armstrong has claimed that Grossmanns position involves absurd lack of economy.322 I suggest, however, that it involves subtlety and sophistication. It bears important, though unintended, similarities to Wittgensteins ontology of facts in Tractatus Logico-Philosophicus. Like Wittgenstein, Grossmann was sensitive both to the special need for the category of facts in ontology and to the special nature of facts.

Metaphysical realists grasp and defense of their position are seldom clear or unambiguous. For example, In his spirited defense of realism, Michael Devitt writes that the sentence a is F is true because it has a predicational structure containing words standing in certain referential relations to parts of reality and because of the way that reality is.323 Yet, in the same paragraph, he denies that truth requires mysterious entities such as facts. One wonders what Devitt could mean, if not a fact, by a way that

291

reality is, or by the word situation, which he uses several pages later in speaking of pairing of sentences with situations.324

Devitt is not alone in taking such a puzzling stand on facts. Hilary Putnam writes that a state of affairs (he could have said possible fact) is a kind of ghostly double of the grammarians sentence. But he then says, Whether a descriptive sentence is true or false depends on whether certain things or events satisfy the conditions for being described by that sentence.325 Presumably, this page satisfies the conditions for being described by the sentence This page is white. How do those conditions differ from the state of affairs or fact that this page is white?

Paul Horwich defends a minimal theory of truth, which he thinks avoids commitment to Russellian facts. He claims that his theory is perfectly consistent with the intuitions that whenever a proposition or an utterance is true, it is true because something in the world is a certain wayFor example, (1) <Snow is white>s being true is explained by snows being white.326 Horwich, too, appeals to ways something is without telling us how these ways might differ from Russellian facts. Moreover, snows being white would not explain <Snow is white> s being true unless snows being white was a synonym of that snow is white, which it is not. The genitive expression snows being white obviously refers to a being white, a color, and would refer to a way snow is only if by way we mean a property, in this case the property of being white. And this property is only part of what explains the truth of Snow is white. It is that snow has (exemplifies, instantiates) that property, not the property by itself, which explains the truth of Snow is white. Speaking of snows being white is quite
292

different from saying that snow is white. Consider the analogy with the genitive expression Jacks coat. It is not a synonym of the phrase that Jack has a coat, and so it does not fully explain the truth of Jack has a coat. Jacks coat refers to an individual thing, a coat, just as snows being white refers to a property, and that coat is only part of what explains the truth of Jack has a coat. It is that Jack has a coat, not the coat itself, that explains the truth of Jack has a coat.

William P. Alston calls his theory of truth minimalist realism, and in a circumspect defense of it he writes, I see no reason to suppose that facts are not objectively real, and as such capable of rendering true propositions true in a nontrivial sense. He adds, however, that the mode of reality of facts is quite different from that of substances, states, properties of substances, and events, as Strawson and others have been at pains to point out.327 Alston does not tell us what the difference is or what he means by mode of reality. However, William Vallicella, also a defender of realism, does. He argues that true propositions require truth-making facts. But he astutely points out that facts could be truth-making only if they are proposition-like, structured in a proposition-like way only if a fact has a structure that can mirror the structure of a proposition.328 Though his formulation of it may invite deflationist attacks, Vallicellas view is in the spirit of Wittgensteins position in the Tractatus, where we find the subtlest and deepest, even if brief, accounts of the notions of fact and correspondence to fact. But Wittgenstein applied these notions, respectively, to atomic facts (Sachverhalte, states of affairs), which are configurations of simple objects, and atomic (elementary) propositions, which consist of names of simple objects. Not surprisingly, he could give no examples of simple objects and therefore no examples of atomic facts or of atomic propositions. Wittgenstein
293

famously argued his notion of correspondence to fact had no application to the statements of mathematics, logic, ethics, and religion, which, according to him, say nothing even though some show much, including the higher. But he saw that it had no application even to the simple, compound, and general statements of everyday talk.

The importance of the category of facts (or states of affairs) is seen clearly and discussed at some length by Jan Westerhoff in his recent book Ontological Categories. He relies on two arguments. The first he calls semantic. Westerhoff writes, Sentences are indispensable for a comprehensive description of the worldtherefore states of affairs must be parts of the world.329 The premise, of course, is true, but the conclusion does not follow. Compare Numerals are indispensable for a comprehensive description of the worldtherefore numbers must be parts of the world. Westerhoff calls his second argument for states of affairs cognitional. Gestalt, not associationist, psychology provides the more plausible account of perception, he says. We always perceive wholes (complexes); we do not first perceive and then put together their parts.330 But to admit that this is so is not to admit that what we perceive is states of affairs. For this to be true, propositions of the form S perceives x must be reducible to propositions of the form S perceives that p. Clearly, few if any are. I now see my hand, but not necessarily that it is my hand or even a hand, that it has five fingers, that it is about a foot in front of my face, or anything else. Westerhoff says that states of affairs provide the primary epistemic point of contact between us and the world.331 Indeed, one of Hegels most influential contributions was to argue that there is no direct consciousness of objects that is unmediated by judgment.332 I noted earlier that in American philosophy Wilfrid Sellars held much the same view. But infants do seem to begin by perceiving things, e.g., their
294

mothers, long before they make any judgments. Perceiving that what they see is Mother, or anything else, is much too intellectual for them. Both wholes and states of affairs have constituents, but it does not follow that all wholes are states of affairs. Perhaps, as Wittgenstein remarked in his Notebooks, all complexity is propositional, but this is not self- evident.333

So far I have considered the case for realism regarding facts. But there is also a strong case for antirealism. The ontology of facts is usually considered in relation to the theory of truth as correspondence to fact. And Strawson writes, The only plausible candidate for what (in the world) makes a statement true is the fact it states; but the fact it states is not something in the world.Of course, statements and facts fit. They were made for each other. If you prise the statements off the world you prise the facts off it too; but the world would be none the poorer. Facts are not broken or overturned, interrupted or prolonged, kicked, destroyed, mended, or noisy.334 Strawson took the latter point to be relevant because he believed that the whole charm of talking of situations, states of affairs or facts as included in, or parts of, the world, consists in thinking of them as things, and groups of things335

Donald Davidson concurred: The realist view of truth, if it has any content, must be based on the idea of correspondence, correspondence as applied to sentences or beliefs or utterances, to entities that are propositional in character; and such correspondence cannot be made intelligible. He went on to add: The real objection [to correspondence theories] is that such theories fail to provide entities to which truth vehicles (whether we take these to be statements, sentences, or utterances) can be said to correspond.336 Quine
295

agreed: Objects in abundance, concrete and abstract, are indeed needed for an account of the world; but facts contribute nothing beyond their specious support of a correspondence theory.337 I have already mentioned Hilary Putnams remark that a state of affairs is a kind of ghostly double of the grammarians sentence.

Devitt was right that, if understood as Russell and Wittgenstein did, facts are mysterious entities. There seem to be no such entities. Alan White pointed out that facts are not even properly describable as states of affairs, for [F]acts, but not events, situations, or states of affairs, can be disputed, challenged, assumed, or proved. 338 Indeed, facts seem to be man-made fictions. The general argument against them is that to serve the purpose for which they are introduced they must be a sort of items that the world cannot be plausibly supposed to contain. They cannot be said to be entities in the sense in which animals, vegetables, and minerals, or even colors and spatial relations, are said to be entities. They bear uncanny resemblance to sentences, in particular the sentences they supposedly make true, and thus appear to be no more than reified, hypostatized sentences. The fact that Dick likes Jane much too conveniently has exactly the distinctive features of the sentence Dick likes Jane: three constituents, configured in the same order as the words, one of them (the relation of liking) categorially different from the other two (the individuals Dick and Jane), much as the verb in the sentence is syntactically different from the names. The fact seems tailor-made for the sentence!

296

That there are such entities as facts is especially implausible in the case of the subcategory of relational facts. Indeed, there is a close connection between the category of fact and the category of relations, since a fact is conceived not as the mere collection of its constituents bbut as a complex entity with constitents related to one another. Now, if there are facts, at least some relational facts about observable things presumably would themselves be observable. But, clearly, none are. The reason is straightforward: Relations themselves are not observable. You can see the left and the top edges of this page, but surely you do not see an item, which might be called the relation longer-than, that holds between them. If you think that you do, what is its location in your visual field and what are its color and shape? After all, surely one cannot see an item that is not in ones visual field, and has neither shape nor color. But if you do not see the relation longer-than, then you also do not see the relational fact to which the true statement The left edge of this page is longer than the top edge is supposed to correspond. Of course, you see the page, as well as its edges, and you also see that the left edge is longer than the top edge. But the question is whether when you see all this you see an item that is the relation between the two edges of being longer than, and therefore whether you can also see a relational fact of which that relation is a constituent. If you supposed that the relational fact is visible while the relation is not, would would be so because the relation is somehow hidden? Or because the relation is too small to see? And you could be asked about the location in the visual field, and the color and shape, of the relational fact itself, not just of the relation, and surely the answer would be that it has none.

To ask these questions is not display sectarian empiricism. Perhaps there are many unseen and unseeable, unobserved and unobservable, entities. But the relational fact that
297

the left edge of this page is longer than the top edge should not be one of them, since the two edges of this page the page you see in front of you at this moment are quite visible, as is the page itself. If there were such entities as facts, surely facts about observable things should themselves be observable. But they are not. Of course, you can see that the left edge of the page is longer than its top edge, but this merely shows that expressions like that the left edge of the page is longer than its top edge must not be taken to stand for entities. It is no part of common sense or any science to think that they do.

Bertrand Russell was perhaps the most astute champion of relations. What he eventually said about them is therefore instructive: For my part, I think it is as certain as anything that there are relational facts such as A is earlier than B. But does it follow that there is an object of which the name is earlier? It is very difficult to make out what can be meant by such a question, and still more difficult to see how an answer can be found. There certainly are complex wholes which have a structure, and we cannot describe the structure without relation-words. But if we try to descry some entity denoted by these relation-words and capable of some shadowy kind of subsistence outside the complex in which it is embodied, it is not at all clear that we can succeed.339 As we have seen, however, the case for the complex wholes that are relational facts is no better than the case for relations. The fact that a is earlier than b is no easier to descry than is the relation earlier. Indeed, this should be expected. How could the former be descried if the latter could not? Pace Russell, there is no special difficulty in making out what is meant by the question whether there is an object of which the name is earlier. What he found difficult was how to avoid giving it a negative answer.
298

Russell in effect admitted that there is no object of awareness, whether perceptual or intellectual, to which the relational expressions in relational statement. Exactly the same can be said also about the expressions for predication, usually some form of the verb to be, in subject-predicate statements. To say that this sheet of paper is white presupposes mastery of the concept expressed by the word is, but surely there is no relevant object of awareness, in addition to this sheet of paper and its color, to which is corresponds. It would be absurd to suggest that there really is such an object but it is not perceived. Is it not perceived because it is hidden from us in the way atoms are hidden?

The interpretation of relational expressions in the history of philosophy has usually been antirealist. Hume and Leibniz held that relations are just the product of the comparison of ideas, not independent denizens of the world. According to Kant, relations such as causality and inherence are a priori, nonempirical concepts of the understanding, imposed on what we perceive by us but themselves representing nothing. The nonempirical status of relations was vigorously defended, and celebrated, by the British idealists, especially T. H. Green and F. H. Bradley. Frege, who resolutely rejected the category of fact, did assume that relational expressions stand for what he called concepts, and held that all expressions standing for concepts to be incomplete, unsaturated, fundamentally different from names, and thus incapable of serving as logical subjects of sentences.

The metaphysical notion of fact is grounded in our use of declarative sentences, and the supposition that there are facts in the world depends at least in part on the assumption that sentences must correspond to something in the world, that somehow they must be like
299

names. But sentences are not even nouns, much less names. They cannot serve as grammatical subjects or objects of verbs, which is the mark of nouns. Despite appearances, logic does not need them for such a service. Regarding the paradox of material implication, Quine pointed out that This controversy would not have arisen if the notion of statements as naming had been carefully avoided, and the variables p, q, etc., had been treated explicitly as standing in positions appropriate to statements rather than names. For example, Quine said, the natural reading of p q is p only if q, rather than the misleading p implies q. 340 Notoriously, p is true, if taken literally, is gibberish. Snow is white is true is grammatically ill-formed. Snow is white is true is not, but its subject-term is not a sentence it is the name of a sentence. The category of fact is of little use in contemporary logical theory, which usually adopts a Fregean approach to the semantics of sentences by taking their denotations to be their truthvalues, though it does not hypostatize truth and falsity as objects, as Frege did. The paradigms of expressions that stand for entities, whether individuals or predicates, are words assigned to them by convention, explicit or implicit. This is why the number of names and predicates is finite, in principle determinable for any language. Not so with sentences. They are not introduced in language by convention. And their (potential) infinity is one of the most important data in linguistics.

Of course, as we saw earlier, there are nominalizations of sentences, such as Snows being white. They indeed are nouns and often do serve as subjects or objects of verbs. But they are not sentences and could not be taken to stand for facts even if there were facts. Consider the phrase John's whistling as it might occur in the sentence John's whistling awoke her. It stands for a whistling, which is an action, a doing, perhaps just a
300

noise, not for a fact. Hence the tendency to refer to facts instead with what Fowler called fused participles, like Snow being white or John whistling awoke her. (Fowlers example was Women having the vote reduces men's political power.) One of the two words in John whistling is presumably intended to modify the other; otherwise, they would be just a pair of unrelated words. But if whistling is intended to modify John, then the grammatically proper sentence would be Whistling John awoke her or, better, John, who was whistling, awoke her. Both are obviously different from John's whistling awoke her they could be true when it is false and false when it is true. And if John is intended to modify whistling, then the grammatically proper sentence would be the original John's whistling awoke her, the subject-term of which refers to something John does or makes, an action or a noise, not a fact. Contrary to Roderick Chisholm, Fowler's objection to fused participles was not a mere stricture.341 They are bad logic, not just bad grammar.

In chapter 9 I drew attention to certain experiences associated with our use of the logical connectives and the quantifiers. Since I have counted as logical expressions not only sentential connectives and quantifiers but also sentences, we may now ask, Are there distinctive logical experiences associated with sentences as such? The obvious example would be belief, in the sense of an experience or feeling of conviction, confidence, perhaps faith. But assertions of sentences, i.e., statements, usually are not accompanied by such feelings they are just made, correctly or incorrectly. Contrary to current philosophical opinion, they usually express no beliefs if belief is something that sentences might express but is neither faith, conviction, confidence, nor or anything else. Statements only seldom, if ever, express faith. They do sometimes express judgments, if
301

by a judgment we mean a decision about truth or falsehood, and decisions are sometimes made with conviction, even a feeling of confidence. But, of course, not always or even usually. It is common in current philosophy to speak of peoples systems of beliefs, and even to say that theories are such systems. But people are at most just confident in some judgments they make. Theories are, at most, systems of propositions, which in the present context means no more than that they are systems of sentences. And it is unlikely that anyones totality of beliefs deserve to be called a system many are haphazard, unrelated to the persons other beliefs.

2. Semirealism regarding facts. In the previous section I considered the case for realism and then the case for antirealism regarding facts. But there is a third way of understanding facts, neither realist nor antirealist. It is semirealist. In general, if a proposition is in dispute between realism and antirealism, with the realist asserting and the antirealist denying it, the semirealist differs from both by holding that it is an improper proposition, perhaps even that there is no such proposition, and thus that both asserting it and denying it are improper. There is an analogy here with sophisticated agnosticism. The theist asserts the proposition God exists and the atheist denies it, but the sophisticated agnostic questions, for varying reasons we need not consider here, its propriety or sense and neither asserts nor denies it. The unsophisticated agnostic, of course, takes its propriety or sense for granted but claims ignorance of its truth-value.

Now the semirealist way of understanding facts relies on Wittgensteins distinction between saying and showing, and applies the resolution he proposed of the puzzles
302

regarding universal statements to the puzzles regarding facts. Let us recall that fact was on Wittgensteins list of formal concepts, together with object, complex, function, and number. According to him, when something falls under a formal concept as one of its objects, this cannot be expressed by means of a proposition, but instead is shown in the very sign for this object (4.126). This is why it is nonsensical to speak of the total number of objects. The same applies to the words 'complex', 'fact', 'function', 'number', etc. They all signify formal concepts, and are represented in conceptual notation by variables, not by functions or classes (as Frege and Russell believed) (4.1272). It cannot be said that what the sentence This page is white asserts is a fact, for the same reason it cannot be said that this page is an object. Indeed, as we have seen, to say p is a fact would be worse, because if we replaced the sentential variable in the phrase with a sentence, as we should be able to do, the result, e.g., This page is white is a fact, would be grammatically ill-formed, which This page is an object is not. Of course, we may say That this page is white is a fact, but That this page is white is not a sentence and thus not a value that p can take.

Wherever fact is used as a proper concept-word, nonsensical pseudo-propositions would be the result, just as wherever object is used as a proper concept-word. We can no more intelligibly talk about facts than we can talk about objects. But this does not mean that Wittgenstein was an antirealist regarding facts and objects. When an item, e.g., what the true sentence This page is white allegedly corresponds to, falls under the formal concept fact, this indeed cannot be expressed by means of a proposition, but neither is it nothing. For it is shown in the sign for the item. In the case of a fact, the sign in onceptual notation is the sentential variable p, the variable reserved for
303

sentences as its values. In ordinary language it would be the sentence itself, though here I go beyond what Wittgenstein says. To be sure, a sentence does not show a fact, for it might be false, but it does show that if it corresponds to anything it could correspond only to a fact, not an object, complex, function, number, etc., just as the name Peter does not show Peter but rather that if it stands for anything it stands for Peter. Wittgnsteins showing is not some supernatural intuition, whether of objects or of facts and thus truth.

If the concept of fact is formal, then it also cannot be said that there are facts. Nor can it be said that a true sentence (or statement, proposition, judgment, thought, belief, etc.) corresponds to a fact. To this extent, antirealism regarding facts is right. Yet, as we saw earlier, the presence in the world of the individuals, properties, or relations mentioned in a true sentence does not suffice for its truth. Jack admires Jill mentions Jack, Jill, and the relation of admiration. But their presence in the world does not suffice for the truth of the sentence. It does not distinguish the world in which the sentence is true from worlds just like it in this respect but in which the sentence is false. Surely, however, there is such a distinction to be made. To this extent, realism regarding facts is right.

Semirealism regarding facts differs from realism by denying that true sentences stand for special entities that are additional to and categorially different from the entities normally mentioned in the sentences, and which can be referred to, described, and analyzed independently of those sentences. But it differs also from antirealism by acknowledging that there is more to truth than the sentences (statements, propositions, judgments, thoughts, beliefs, etc.) that are true. It may even label this additional feature of truth correspondence to fact. The label would be reasonable since in everyday talk we do say
304

that a sentence is true because it fits the facts. The terminology of facts thus need not be shunned by the semirealist. There is more to the truth of a sentence than the sentence that is true, there is also what makes the sentence true, but we cannot say what that is except by repeating the sentence or uttering a synonym of it. We can say that this page is rectangular, white, made of paper, and that we are holding it in our hands. None of these, or anything like can be said of the fact that this page is white.

Insofar as facts cannot be referred to or described independently of the sentences that supposedly stand for them, antirealism regarding facts is plausible and the correspondence theory of truth is implausible. Nevertheless, especially in the case of observation sentences (I see a cat there, not a dog), obviously a sentence is true not just in virtue of the words it contains or some relation it bears to other sentences or to the speaker, a culture, or an institution. This is why also realism regarding facts and the theory of truth as correspondence to fact are also plausible. If the correspondence theory is defective, the reason is not that it distinguishes the sentence from the fact that makes it true, but that it fails to acknowledge the special nature of facts, the special status of the concept of fact.

Wittgensteins notion of what cannot be said but can be shown was applied to matters that are formal, of global import, unobservable, and thus not literally picturable, yet also essential to both language and the world. We need not follow him everywhere. Nor need we claim a special insight into what he really meant by the distinction between saying and showing in order to see, as we did in chapter 4, that it fits the traditional conception of meaning and thought as representation. It is clearest when applied to formal concepts,
305

where the futility of statements such as that A is an object is obvious. Its plausibility in ethics rests on the claim, examined in chapters 3 and 4, that ethical statements are partly, but necessarily, about the whole world, the totality of facts. Arguably, its plausibility in religion has a similar explanation. And so does its plausibility when applied to the category of facts.

The distinction has immediate consequence for the realism/antirealism debate, at least insofar as it concerns cosmological realism and antirealism. If the world is the totality of facts and fact is a formal concept, then world is itself a formal concept. Things in the world may be observable and picturable, but the world as such certainly is not. Trivially, the world has global import. And, trivially again, the world is essential to itself, even if perhaps not to language. Now the key proposition in dispute between the cosmological realist and the cosmological antirealist is There is a world that is independent of our cognition. The realist asserts it, and the antirealist denies it. But the semirealist refuses to allow that it is a genuine proposition. The semirealists reasons are not those that led Carnap to the same conclusion, but the conclusion nonetheless enjoys ample plausibility. There is a world that is independent of our cognition is not a genuine proposition because it cannot be said that there is a world. And this cannot be said because the world is the totality of facts and it cannot be said that there are facts.

But what cannot be said in this case can be shown, as it can in many other cases, ranging from generality to morality. Were one to ask whether what cannot be said exists, the answer would be that it does not exist, but also that it is not nothing. A universal statement would not be true if some of its singular instances were false, but we cannot
306

say that all of them are true except by repeating the statement. Similarly, there is more to truth than the statements that are true. But we cannot say what that is, what makes a statement true, except by repeating the statement or a synonym of it. What makes a statement true, according to the semirealist, is indeed something other than the sentence, and we may call it a fact, but it cannot be described except by using the statement or a synonym of it.

Metaphysical semirealism thus does not deny that there is more to reality, the world, than words. Our statements are not all that there is because some of them are true. Nor does metaphysical semirealism deny that there is a reality, things-in-themselves, independent of our cognition. To this extent it agrees with metaphysical realism, though also with transcendental idealism. But unlike the latter it also does not assert that there is such a reality. To this extent it agrees with metaphysical antirealism. Instead of appealing, as the antirealist does, to the tautology that there is no cognized reality that is not cognized, the semirealist explains that we cannot say that there is a reality that is independent of our cognition, but also we cannot say that there is no such reality.

A comparison between Wittgensteins and Heideggers use of show would be instructive here. Heidegger argued that phenomenology lets things show themselves, even though as a philosophical discipline it, of course, requires statements. He relied on the fact that in Greek the terms for saying and showing are related. Indeed, there is a sense of show in which showing something does require saying. To show that Plato was the author of the Republic we must say a great deal about Greek philosophy. Show also has other senses that are compatible with saying. You can show, without speaking, a
307

new toy to a child, or an unusual item to an adult. But such showing ve to the primitive levels of cognition and is not what Heidegger had in mind. Wittgensteins sense of show was completely different. In that sense, what is shown can never be said, and showing is never phenomenal, like showing a toy. It is logical, because formal concepts are involved. This is why Wittgensteins sense is relevant to the dispute between logical realism and antirealism. But what is shown in that sense need not be an esoteric logical feature. It could be just the banality that true statements assert facts.

Wittgenstein applied his distinction between saying and showing to all logical objects, not just facts. The inaccessibility of facts except in sentences is due to the status of fact as a formal concept. It can be understood exactly as we understand the inaccessibility of an objects being an object, or the inaccessibility of the totality of the singular instances of a universal statement. Facts show themselves only in language, they are shown only in language. Our cognition of what makes our statements true is thus dependent on language, on something that is ours, on us. This is why Wittgensteins position on facts is not realist. But that dependence of the world on language does not consist in the reducibility of statements about the world to statements about language. The world is the totality of facts, and there are no genuine statements abou facts. This is why Wittgensteins position is not antirealist. It is semirealist.

Metaphysical semirealism is a moderate position. It is not ontological antirealism. It leaves the reality of things untouched. It does not, absurdly, hold that all there is to animals, vegetables, and minerals is words. Nor does it, equally absurdly, hold that all there is to the world is sentences. The world is the totality of facts. To this extent,
308

metaphysical semirealism is not cosmological antirealism. But it also holds that, since fact is a formal concept, the relationship between a true sentence and the fact it asserts can only be shown. This may seem mysterious, but it is no more so than the futility of saying that this page is an object. If the world is the totality of facts, then the distinction between true sentences and the world can also be only shown, not asserted. But for the same reason neither can it be denied. Facts are not entities that true sentences name, and if truth were a relation between sentences and facts, as the correspondence theory of truth holds, it would not be an ordinary relation like admiration or earlier-than. This does not mean, however, that there are no facts, or that truth is not correspondence to fact. What it means is that we must free ourselves from the confines imposed by thinking of truth as correspondence to fact and see that the question about what makes true sentences true ignores the complexity, subtlety, and wide range of the use of the word true.

3. Truth. Only a poor philosophy of language would hold that to have meaning an expression must stand for an entity, but it is sound metaphysics to identify what entities there are by first asking what entities, if any, various expressions appear to stand for. Realism about Jack and Jill need not infer Jacks and Jills reality from the existence of their names, but it might identify what it is about as the people those names stand for. Realism about properties (universals) need not hold that there are such entities on the grounds that there are predicates, but it might identify what it is about as what predicates seem to stand for. Realism about facts identifies what it is about as what true sentences seem to stand for, though it need not hold that sentences would have no meaning or sense if there were no such entities as facts.
309

In chapter 8 I argued that generic statements are essential to cognition. Many generic statements, in everyday life as well as in science, are considered obviously, unquestionably, true. Yet I also argued for antirealism regarding generic facts. It might have been asked, if there are no generic facts, what makes generic statements true? But the use of the word make in this context, though common in philosophy, is metaphorical and misleading. We should rather speak of the features, relational or nonrelational, in virtue of which sentences are taken to be true, and acknowledge that these features are many and diverse.

According to Frege, Truth is the object that is the reference (Bedeutung) of all true sentences. He did not claim to have discovered a new object, as an astronomer might claim to have discovered a new planet. He explained that the reasons for his position were, first, the general distinction between the sense and reference of expressions names as well as predicates other than sentences and, second, the obvious connection between the reference of those expressions and the truth-value of the sentences in which they occur. (Whether She is young is true obviously depends on who she is.) Frege mentioned two alternatives to his view, but argued that neither is credible. Regarding the alternative that truth is the relational property of correspondence to fact, he remarked, as we saw earlier, What is a fact? A fact is a thought that is true.342 And regarding the alternative that truth is a nonrelational property of sentences or thoughts, he wrote, One can, indeed, say: The thought that 5 is a prime number is true. But closer examination shows that nothing more has been said than in the simple sentence 5 is a prime number. Unlike contemporary deflationist theorists of truth, however, Frege also
310

explained why this is so: Subject and predicate (understood in the logical sense) are just elements of thought; they stand on the same level for knowledge. By combining subject and predicate, one reaches only a thought, never passes from a sense to its Bedeutung, never from a thought to its truth-value.343 According to Frege, the primary bearers of truth-value are thoughts, not sentences, but he did not mean by thought a subjective, psychological state or event: By a thought I understand not the subjective performance of thinking but its objective content, which is capable of being the common property of several thinkers.344 Frege did not say that one never passes from a thought to the world. He said that one never passes from a thought to its truth-value. Russell and Wittgenstein needed a world of facts. Frege needed just plain Truth.

Anyone who thinks that Freges focus on truth, rather than on facts or the world, is obscurantist may be reminded that the notion of a fact and that of a world (whether understood as the totality of facts or the totality of things) are not clearer. It would be absurd to identify the world with the totality of Fregean thoughts, since according to him thoughts belong in a realm that is neither physical nor mental, but not more absurd than to identify the world with the totality of true sentences. What would not be absurd, however, is to identify the world with truth it would only be uncommon. Anything essential to the actual world must be included in our knowledge of it, and knowledge requires truth. Perhaps knowledge requires also belief, but it would be grossly misleading if I said that I believe that I am typing now, even though of course I know that I am.345

311

Identifying the world with truth is not unprecedented in the history of philosophy. Freges view resembled the medieval view of Truth as one of the transcendentalia. Both views allow for, and explain, a wide diversity in the application of the word true. Neither allows for a simplistic definition of truth. Frege, of course, did not attempt to provide one. And according to the medieval view, the transcendentalia are not even categories of things, much less things: they enjoy supreme abstraction by ranging across the highest genera, i.e,, the ontological categories, and therefore could not be defined per genus et differentiam.346 In addition to Truth, the transcendentalia included Being, Good, One, and Beautiful. The presence of both Truth and Good among them may explain why we find plausible the suggestion made in our time by Putnam, Goodman, and Dummett that truth is a sort of goodness (Putnam), rightness (Goodman), or correctness (Dummett).347 Putnams suggestion is preferable, since rightness and correctness are plausibly themselves regarded as kinds of goodness. The presence of both Truth and Being among the transcendentalia may account for the important use of true in such phrases as true friend and true art, where true has the meaning of real or genuine, contrasted with apparent. And the familiar diversity in the uses of good may render the diversity in the uses of true less surprising.

The several theories of truth emphasize different uses of true, and each is plausible in its own way. Some sentences, most notably those reporting observation, are taken to be true in virtue of surviving directly what Quine called confrontation with sense experience. Their case lends support to the standard correspondence theory of truth, especially when held by empiricists. Some simple mathematical sentences perhaps are taken to be true in virtue of surviving, also directly, confrontation with intellectual
312

intuition, a priori insight, or reason. But most mathematical sentences and all theoretical sentences in science are taken to be true mainly, if not solely, in virtue of their membership in systems that survive confrontation with experience or with reason. We may say that such sentences survive confrontation with experience or with reason indirectly. Their case lends support to the theories of truth as coherence and as idealized warranted assertability. Sometimes sentences are accepted only because they belong to theories judged more beautiful or elegant than their alternatives. They could have given rise to an aesthetic theory of truth. And sentences like I will be alive tomorrow are taken by the speaker to be true mainly, if not solely, for practical reasons. Their case lends support to the pragmatist theory of truth. They call for further explanation.

In ordinary, not all, circumstances, my taking the sentence I will be alive tomorrow to be true rests neither on experience or reason, nor on its coherence with other sentences. If I thought it needed to rest on any of these, or indeed on anything, I might not take it to be true and as a result perhaps suffer disastrous consequences today or at least cancel my life insurance policy. Rather, its truth is presupposed by virtually everything I do today. My life would be radically different if I did not have unquestioning faith that it will continue for at least one more day. Acceptance of the sentence I will be alive tomorrow is thus practically necessary, while the acceptance of some sentences of mathematics and science may be said to be theoretically necessary and the acceptance of some observation sentences (Its hot!) perhaps palpably necessary.

313

The word true is versatile enough to allow without equivocation for such diversity in what makes sentences true. It does resemble the word good in this respect. Gustatory pleasure, knowledge, compassion, right conduct, and justice are all standard examples of good things, but they have little else in common, except in the eyes of philosophers with penchant for straightjackets. Yet there is no equivocation in calling all of them good.

The ways of knowledge and truth are thus not neat and tidy. This is especially evident in ethics, as we saw in chapters 3 and 4. The moral there, however, was not that we should accept noncognitivism. And the moral here is not that we should stampede into a theory of truth that renounces correspondence to fact altogether. A doctors orders are imperatives, but usually their legitimacy and authority are unquestionably cognitive their ground is usually cognitive. Theology often grounds the authority of God in his omniscience, not his status as our creator, but ordinary religious thought usually holds the latter to be a sufficient ground. It is in this second way, presumably, that the authority of what Kant called practical reason (Vernunft) and his description of ethical judgments as both imperatives (Imperative) and cognitions (Erkenntisse) should be understood.348 Then, perhaps, we should call ethical judgments valid rather than true, as Kant indeed often did. They can be valid in the sense in which we call valid both a doctors orders and the propositions grounding them. But ethical judgments can be valid also in the sense in which a traffic policemans telling us to move to another lane can be called valid. We may follow Nelson Goodman and just use right for all four: ethical judgments, a doctors orders, the propositions grounding those orders, and the traffic policemans orders.

314

Goodman indeed admitted that his irrealism closely resembled Kants transcendental idealism. And Kants ethics cannot be detached from his transcendental idealism this is why he added the third section of the Groundwork, making clear that the first two cannot be properly understood apart from the third. It follows that if we cannot understand his transcendental idealism then we cannot understand the moral yet nonepistemic authority of a noumenal self. It should be clear, however, that nothing merely human could enjoy such authority. The autonomy that Kant thought essential to morality required membership in the intelligible, not just the sensible, world. He wrote, The causality of [actions which can be done by disregarding all desires and incitements of sense] lies in man as intelligence and in the laws of such effects and actions as accord with the principles of an intelligible world[H]e is there his proper self only as an intelligence (while as a human being he is merely an appearance of himself).349

In the case of generic statements, neither correspondence to fact nor coherence or practical utility suffices for truth, yet all three are relevant. Dutchmen are good sailors is taken to be true because a sufficient number of singular statements of the form x is Dutch and x is a good sailor are taken to be true, each in one or more of the senses mentioned earlier. How many such statements suffice would depend on how the generic statement coheres with various other statements about sailors, ships, shipping lanes, marine weather, piracy, the presence of men-of-war protecting merchantmen, and so on. And surely its truth would depend also on practical considerations, best known to shipmasters and ship owners, such as the purpose and length of most sailings, the value of cargoes, the availability and cost of labor at the docks, and so on. If such diversity in what is relevant to the truth of generic statements seems excessive, we should recall that
315

generic statements are not restricted to just one area of cognition, whether science, prudential planning, or personal know-how. They are ubiquitous and indispensable wherever we engage in generalization.

One of the objections to the correspondence theory of truth is that it does not provide a criterion of truth. The usual reply is that it is intended not as a criterion but as an account of truth. This reply misses the point of the objection, which is that the account fails to even provide room for a criterion. This is why Kant, who called the correspondence theory a nominal definition (Namenklrung) that proffers no criterion of truth,350 also called it a mere verbal explanation351 and, paving the way to Hegel, held that the proper and sufficient criterion of empirical truth presupposes the idea of the systematic unity of nature.352 For Kant, there were no Russellian or Wittgensteinean facts, or anything else that was not already epistemic and thus, by implication, alethic, to which true judgments would correspond. It is not our cognition that must conform (richten) to objects; rather, objects must conform to our cognition.353 Wittgenstein did endorse the correspondence theory, but only for impossibly impoverished sentences about configurations of simple objects, of which therefore no example could be given, not for the sentences of logic, mathematics, ethics, or even everyday discourse. The failure of an account of a concept to provide a criterion for its application is an important reason for rejecting the account. Where it is not a decisive reason, we expect a special explanation. Theologians often provide such an explanation for the apparent absence of a criterion for the application of the concept of God. Wittgenstein in effect provided one in the case of the concept of truth as correspondence to fact: the formal status of the concept of fact.

316

Just as the presence of Being and Good among the transcendentalia does not preclude us from saying a great deal about the various kinds of being or of goodness, the presence of Truth does not preclude us from saying a great deal about the various kinds of truth. On the contrary, as we have seen, it suggests that the kinds of truth are numerous, lke the kinds of being and of goodness. Truth must not be confused with any one of its kinds, just as being and goodness must not be confused with any one of their kinds. Being is not just matter, nor just mind. Justice is a good but so are pleasure and happiness. It should go without saying that being should not be confused with any of the kinds of matter or mind, for they themselves must not be confused with any of their own kinds. Mind is not just cognition, nor is it just emotion. Justice is not just paying ones debts, it is also telling the truth. That there are many kinds of truth, being, and goodness does not imply that the words truth, being, and goodness are equivocal, just as that there many colors does not imply that the word color is equivocal.

This is why the various theories of truth I have mentioned are all legitimate. The truth of the sentence I have a headache is plausibly viewed as correspondence, if not to a fact then to a thing: the ache that is felt in the head. But even if some sentences can be paired off with bits of the world in this way, most cannot, including many that especially interest us. Richard Rorty offered as examples the sentences constitutive of a theory like Marxism.354 More obvious examples are counterfactuals, e.g., If Hitler had not invaded the Soviet Union, Germany would have won the Second World War, and complex, largely dispositional sentences such as She liked him, admired his intelligence, was attracted to him, but did not love him because he reminded her of her father. The truth of such statements is better viewed as their coherence with a vast number of other sentences.
317

And many sentences about the future, like my example I shall be alive tomorrow, are surely taken as true because the speaker must accept them in order to engage in normal activities. No account of truth should fail to note such differences between true sentences, or such kinds of truth. We must not seek a definition of truth, just as we must not seek a definition of being or of goodness.

Part Three: METAPHYSICS DEHUMANIZED Solipsism, when its implications are followed out strictly, coincides with pure realism (Wittgenstein)

Chapter Eleven: I AND THE WORLD

1. The paradox of antirealism.

318

The general thesis of this book is that there is no place for anthropocentrism in philosophy. I have argued that epistemology and ethics can be dehumanized directly, without qualification. In metaphysics, however, anthropocentrism is present in the form of antirealism, including Kants transcendental idealism and contemporary antirealisms. Part Two was devoted to showing the virtues of antirealism. Therefore, we should not dehumanize metaphysics by just rejecting antirealism. We must not just return to a preKantian metaphysics. Our goal should be, not to reject antirealism, but to free it of anthropocentrism. This means that we must understand the central claim of antirealist metaphysics as making no reference to humans. Can this be done? This will be the topic of the present Part Three.

Let us first review what has been accomplished so far. In the Introduction, after noting the implausibility of the traditional arguments for antirealism, I promised to bypass them and start afresh, from specific and readily understandable truths, not abstract and obscure philosophical assumptions. I hoped to arrive in this way at a version of antirealism that is modest and credible. The antirealist position defended in Part Two does not deny the reality of things. It is cosmological, not ontological, antirealism. It denies only the reality of generic facts, and regarding other facts it accepts, at most, semirealism, like that defended, though in different ways, by Wittgenstein and Bergmann. But it retains the metaphysical bite of traditional antirealism. For generic statements are essential to cognition of the world, even if not to cognition of things. Such an antirealist position would hardly be opposed by common sense. Even philosophers have not claimed that there are generic facts. As to semirealism regarding other facts, common sense would find the distinction between it and realism too technical to worry about.
319

The thesis of antirealism is that the world, at least insofar as it is cognized, depends on our cognition of it. But any cognition that is at all advanced requires statements. There may be no generic facts, but generic statements are ubiquitous and unavoidable in cognition. And regarding other facts, even if we accept Wittgensteins semirealism, our position allows them only insofar as there are statements. What shows itself but cannot be said is not nothing, it shows itself, but it does so in what we do say in statements. Even a mere conception of the world requires statements, be they true or false, if it is to count as conception of a world, rather than of a mere collection of things. And statements are uses of language. Insofar as cognition of the world, whether knowledge or mere conception, requires statements, it depends on language and therefore so does the world insofar as it is cognized. But the only language we know or can even conceive is our language, a human language, or at least one translatable into ours. Could there be languages that are nonhuman and in principle untranslatable into a human language? There is no clear sense in which they would count as languages. As Quine remarked, illogical cultures are indistinguishable from ill-translated ones.355

It seems, therefore, that the world, insofar as its known or knowable, still depends on humans, though more specifically on their language. The antirealist position explained and defended in Part Two still implies that the world is dependent on a tiny part of itself if not just on me, P.B., then on us, the members of the human species. In either case, we face cosmological humanism, indeed human creationism. Both, of course, are absurd. They are anthropocentrism at its worst. The absurdity of supposing that the world

320

depends on me is obvious. The absurdity of supposing that it depends on us may seem less glaring. But the supposition remains absurd.

Contrary to appearances, however, antirealism need not imply that the world depends on me or on us. It need not imply a bizarre solipsism that says I made the stars, nor a human creationism that says We made the stars. The first-person pronouns I (me, my) and we (us, our), as they are used in the realism/antirealism debate, indeed in most if not all philosophical contexts, can and must be understood as impersonal. If the first-person singular pronoun I could be so understood, then the consequence would be the demise of the narrowest form of anthropocentrism, subjectivism, in all three of its forms: solipsism, skepticism, and egoism. If the first-person plural pronoun we could be understood as impersonal, then the consequence would be the demise of the broader form of anthropocentrism exemplified by traditional antirealism. And, as we shall see in the next chapter, the question whether antirealism is a form of idealism subjective, transcendental, or absolute would also have been answered. Our focus in the present chapter will be on the first-person singular pronoun I.

Philosophical attention since Descartes has focused on the self, the ego. The validity of Descartes argument I think, therefore I am has often been questioned, but its premise has remained, even if only tacitly, a central assumption of modern philosophy. The Latin synonym of the English first-person pronoun I is Ego. Both ordinarily refer to oneself, the speaker using them. Thus the focus on the self is not only anthropocentric, it leads anthropocentrism to its logical limit by limiting it to just one human being. In ethics that limit is egoism, in epistemology it is skepticism about the external world
321

(including other minds), and in metaphysics it is solipsism. Indeed, in the history of philosophy, ethics has been largely preoccupied with battling egoism, and epistemology with battling skepticism. There has not been much of a battle against solipsism, perhaps because no one has seemed to hold it.

In chapter 2 I considered how subjective epistemology would face the challenge of anthropocentrism. I concluded that, on pain of begging the question against the skeptic, it must renounce having a subject matter at all. It must be understood as concerned, not with entities, whether all humans or just oneself, but rather with the necessary conditions of thought and talk. In this and the following two chapters I shall consider how the challenge of anthropocentrism would be faced by metaphysics. I shall argue that in the relevant philosophical contexts first-person pronouns, both singular and plural, refer to cognition of the world and thus to the world itself, not the human beings to which they refer in nonphilosophical contexts. In nonphilosophical contexts I refers to, say, P.B., and we refers to, say, P.B. and Jack. P.B. and Jack are both located on the planet earth. The world, of course, is not located on any of its planet.

2. First-person singular pronouns In chapter 2 I argued that when engaged in Cartesian doubt I must use the pronoun I, not the noun P.B. or any definite description such as the speaker of this sentence. That pronoun must be used without an antecedent noun, it must be a dangling pronoun. Descartes could not have offered Descartes exists, instead of I exist, as the first truth
322

he discovered. The reason is that Descartes was the name of a certain 17th century Frenchman, who was part of the external world the existence of which Descartes had not yet proved.

Many philosophers have held that the primary reference of the indexical I is the philosophical self, the ego, das Ich. Presumably, it was such an entity that Descartes meant when later he asserted that he is a thinking thing. There is a difference between using I and using ones name. John Searle gives the following example. If I make a mess in a supermarket by spilling a bag of sugar on the floor, I may be ashamed, look to see if anyone saw me, and worry about what to do. Whether I use I or my name does not affect the truth-value of saying that I made a mess, yet there is an important difference. As Searle says, what is essential to the case is that it is me that is making a mess.356 There is a difference between my making a mess and P.B.s making a mess. The connection with the experience of shame is evident and direct only in the case the former. For less Anglo-Saxon examples of this sort, such as hearing steps behind me while I am peeping through a keyhole, we may go to Sartre.

If I use I in an autobiography, P.B. could replace I throughout without change in truth-value there would be a change merely in literary genre, from autobiography to biography. There would be no difference in reference, though there might be subtle differences of the kind Searle noted. If we say that there would be a difference in meaning or sense, our use of these terms would be technical and therefore requiring extensive, inevitably controversial explanation of how they differ from reference, the sort of explanation I have tried to avoid in this book and is notably absent from Searles
323

example. But in a Cartesian context none of the statements in the biography would be allowable because all would beg the question against the skeptic, who questions the existence of a physical world and therefore the existence of all of its inhabitants, including human beings such as Descartes, or P.B..

In chapter 2 I pointed out that it would have been futile for Descartes to say at the initial stage of his reasoning, as he did later after having proved his existence, that in his argument I think, therefore I am, I referred only to a thinking thing, not to a 17th century Frenchman. Presumably, Louis XIII also was or had a thinking thing, but Descartes did not hold that the argument proved the existence of that thinking thing. It would have been also futile for him to say that I referred to this thinking thing. He might have been using the phrase this thinking thing to refer to Louis XIIIs thinking thing. Russell held that the demonstrative pronoun this can refer only to objects of acquaintance, meaning direct awareness, but surely he was mistaken. I can speak about a man mentioned in a newspaper story I just read by saying This man is horrible, though I am not acquainted with him even in the ordinary sense of acquainted. At any rate, even if Descartes had held Russells view, he would still have needed to say who was acquainted with this thinking thing, Descartes or Louis XIII? Of whose direct awareness was the thinking thing an object, Descartes or Louis XIIIs?

It would not have helped if Descartes had replaced I with TT, a name he might have given to the thinking thing the existence of which he thought he had proved with his argument. The reason is not that there there would be no such thing as TT, or that the premise would be false or the argument invalid, but that referring to TT would again beg
324

the question against a sophisticated skeptic. If Descartes inferred that TT exists, which thinking thing did he name TT? Surely, not Louis XIIIs, for Louis XIII, a human though royal being, would not exist if the physical world did not. But for the same reason Descartes could not have named Descartes thinking thing TT.

Could Descartes have just said, It is my thinking thing? No, because he would be using the first-person singular pronoun my. To speak of my thinking thing is to identify it as the thinking thing related to me by a relation, whatever it might be, say, the relation of having, possessing, or owning, and so I must be able to identify the reference of me, and thus of I, independently of that relation. Moreover, contrary to Descartes intentions, to speak of my thinking thing would imply that I am not that thinking thing, that I am only the one who bears that relation to it, which I bear also to my car and my nose, I being the owner, and the thinking thing, car, and nose being the owned. And my bearing that relation to TT would be just as questionable as my bearing a relation, say, the relation of admiration or distaste, to Louis XIIIs thinking thing. Moreover, any skeptic worth his salt would forthwith ask, How do you know that there is this thinking thing to which you say you bear the relation? The skeptic would ask the question just by being true to skepticism, not because of metaphysical prejudice against thinking things or subtle semantic opinions about pronouns. It would be useless to say that TT is the thinking thing to which only I have access. Even if this were so, and we ignored this unacceptably metaphorical sense of access as well as the implied crude distinction between inner and outer, saying it would presuppose an independent answer to the earlier question about the reference of I. Who is the one who alone has access to that thinking thing?

325

Indeed, not even the most frugal cogito, in which instead of I am, Descartes would infer from I think only There is a thinking would be acceptable. Which thinking would that be? For the already given reasons, we cannot say that it is Descartess, Louis XIIIs, P.B.s, or even just mine. If we bite the bullet and say the thinking is no ones, we face irrelevance in addition to syntactical absurdity. There may be thousands of such orphaned thinkings. The existence of which one did Descartes infer? Could he just say: this thinking, and perhaps name it T? But which thinking is this thinking? Why suppose that another thinking is not also a this thinking? Pace Russell, this is not a logically proper name but a demonstrative pronoun, and many things, near and distant, past, present, and future, observable and unobservable, can be and often are referred to with it, even by the same person and at the same time, e.g., when the person is texting while chatting. And, presumably, the existence of some such orphaned thinkings would be just as questionable for the skeptic as the existence of Louis XIIIs thinkings. The existence of which thinkings would not be questionable? To say Those of which I am directly aware would take us back to where we started. We did not begin with dangling pronouns in order to end with orphaned thinkings.

It is not a principle of logic but surely true that Someone is F readily follows from I am F only if the indexical term I could be replaced with a name or a nonindexical definite description, even if it is never so replaced and even if we do not, perhaps cannot, know with which name or description to replace it. We readily accept statements employing indexicals because we take for granted that in principle they can be replaced with nonindexicals, even if sometimes we find ourselves unable to do so. We accept statements employing here and now since usually we are willing to replace them
326

with names or definite descriptions of places and times. This is true also of I in ordinary contexts. P.B. could replace I in I am writing this book, even if we agree that, as in Searles and Sartres examples, the replacement would be more than just a change of words. In some legal documents it might even be required. But in a Cartesian context it is prohibited. And, as we have seen, it is only an illusion to think that it would be allowed if we used TT, or even just T, in place of Descartes or P.B.

In the argument I think, therefore I am, the pronoun I is profligately used twice in the premise and the conclusion. Does it have the same reference, be it to a man, a thinking thing, or a mere thinking? If it does not, the argument is invalid. Does it have reference in the premise at all? If it does not, the argument would lack a premise. But if it does, the question against the skeptic is begged.

When the reality of the whole world is at issue, whether in epistemology or in metaphysics, the reality of none of its parts, whether Descartes, P.B., or all humans, may be presupposed. Descartes could not have claimed that I exist was the first truth he had discovered if by I he referred to Descartes, the French philosopher. But neither could he have begun his argument, as he did, by writing Let us suppose, then, that we are dreaming if by we he was referring to humans. Like Descartes, humans are inhabitants of the external world the existence of which Descartes had proposed to doubt. What was distinctive of Descartes epistemology was not his initial commitment to the existence of only one entity (himself), or the nature of that entity (a thinking thing), but his decision to presuppose reference to nothing that might be doubted, and therefore the requirement,

327

which he in fact met, that initially he use only indexical expressions in the relevant contexts.

Such a decision, however, has far-reaching implications of which Descartes seemed totally unaware. When I say that now it is cold here, I am not saying that on November 7, 2007, it is cold in Iowa City, even though on November 7, 2007, I am in Iowa City. Yet neither am I explicitly referring to a time other than November 7, 2007, or a place other than Iowa City. This causes no geographical or historical problems as long as I can replace now with November 7, 2007 and here with Iowa City, or some other date and name there would still be a time and place I am talking about. But if I cannot do so, then I would be like a geographer only of here, or a historian only of now.

There is a difference between indexical and nonindexical expressions. There is a difference between saying It is cold here and saying It is cold in Iowa City, though it is neither the sort of difference there is between It is cold here and Il fait froid ici, nor the sort of difference there is between It is cold in Iowa City and It is cold in the university town on the Iowa River, even though all four seem to be saying the same thing. It certainly is not just a difference of words. Perhaps this is why some have thought that there is an entity or property such as here or hereness, which would be the primary reference of the indexical here. Perhaps the difference between It is cold now and It is cold on November 7, 2004 is why some have thought that there is an entity or property such as the present or presentness, which would be the primary reference of the indexical now.357 The latter view is more plausible because of the seeming connection between the notion of time and the notion of existence. Augustine noted that what does
328

not exist now seems to not exist at all, which is not the case with what does not exist here. Kant held that while space is the pure form of outer sense, time is the pure form of both outer and inner sense. And Heidegger named his classic work Being and Time, not Being and Space.

I suggested in chapter 2 that, to be consistent, a subjective epistemology like Descartess would be like a geography of here or a history of now that is in principle unable to say where is here and when is now. It would be an epistemology of dangling pronouns and without nouns. Of course, our interest here is not in exegesis and criticism of Descartes. Considerations similar to those that apply to Descartes epistemological project apply also to the metaphysical dispute between realism and antirealism. I must not understand distinction between the world and me as holding between the world and P. B. There is no place for the ordinary use of first-person pronouns in the context of debating the dependence or independence of the world on our cognitive faculties, just as there is no place for it in the context of Descartess project of doubt. The pronouns do have a use in both, indeed an essential one. But it cannot be their ordinary one. It must be impersonal. That there is such a use is evident from the intelligibility of Cartesian doubt and the realism/antirealism debate, even if they are misconceived or defective in some other way. If in such contexts I cannot refer to the speaker, obviously, for the same reasons, it cannot refer to any other part of the world. Though grammatically a personal pronoun, the use of I in such contexts must be understood as impersonal. How is this use to be understood, then? What could I refer to?

329

Like all indexicals, I serves to indicate or refer, not to name. Unlike a name, what an indexical indicates depends on the circumstances of its utterance, in particular, on the speakers spatial, temporal, perceptual, conceptual, and linguistic perspective in those circumstances. As etymology suggests, a perspective is a view, a view from a standpoint, a point of view. But it need not be any particular persons view. If a travel agent praises the view of Manhattan from the top of the Empire State Building, it would be absurd to ask Whose view is it? as absurd as asking Whose truth is it? when a proposition is described as true, or asking Whose knowledge is it? when physics is described as a body of knowledge. Nor need the view be optical, or even perceptual. It might be conceptual. Despite its etymology, view often refers to understanding or thought, as in political (scientific, educational, philosophical, religious) view. And it could also be linguistic, in the straightforward sense of involving a language, a respect in which a normal adults view of something surely differs from a neonates and is in part a function of how the thing is, would, or could be described in the adults language. In this broad sense, a view can be called a cognition. I is an indexical, rather than a name, because its use is inseparable from a perspective, a view, a cognition.

When used in philosophical contexts where reference to the speaker or any other inhabitant of the world would be question begging, there is nothing I could indicate except that perspective, view, or cognition itself. It cannot indicate a self, an ego, or a thinking thing, but obviously it does indicate something, it does have reference. Only the perspective, view, or cognition that even in ordinary contexts is essential to its use remains. In the nonordinary, strictly philosophical, contexts of Cartesian doubt and the realism/antirealism debate, the view could not be a view of Manhattan or of any other
330

particular object. In those contexts it can only be a view of the world, a worldview. But a view is necessarily a view of something, real or unreal. To indicate a view is to also indicate its object. To indicate a view of Manhattan is also to indicate Manhattan. Therefore, in those philosophical contexts I also indicates the world itself.

Both in the narrow, optical sense, and the wider sense of cognition, a view is a state of consciousness. We thus achieve contact wiith Kants transcendental idealism and Hegels absolute idealism, as well as Husserls phenomenology, but we need not presuppose any particular theory of consciousness. We must insist, however, that to count as advanced cognition the consciousness must receive linguistic expression.

The use of I coheres with and usually presupposes the use of five other indexicals: here, now, thus perceived, thus understood, and thus described. (Thus is etymologically related to this and that.) The particular cognition these six indexicals indicate in the philosophical contexts that concern us here is not defined by just a geometrical point, a position in space (here). It is defined also by position in time (now), perceptual modality (seen, heard, felt), conceptualization (thus understood), and linguistic expression (thus described). The world inseparable from that cognition is the world thus perceived, understood, and described here and now. If we regard understanding as the core of cognition, we may aggree with Hegel that the I is the pure Notion.358 As we shall see in the next chapter, by Notion [Begriff] he meant comprehension, grasp.

Despite its standard use, I sometimes refers even in everyday speech to a view, perspective, cognition, not the speaker. One may say, for example, I cant believe you
331

left your children unattended! in order to provide information not about ones beliefs or anything else about oneself but to indicate the abstract view that leaving children alone as grossly imprudent. In scientific discourse I occurs rarely. We often find, instead, impersonal declarative sentences. When We know that... is found, it usually serves the same role as It is known that..., the point of which is precisely to avoid reference to anyone. If it involves reference, the reference is likely to be to the current state of knowledge, whether generally or in ones specific discipline.

Our conclusion in this section allows us to move closer to understanding otherwise baffling assertions such as Hegels Reality is spirit and Wittgensteins I am my world. These assertions are antirealist. They are also nonanthropocentric. I hall return to them repeatedly.

3. The Self The impersonal use of the first-person singular pronoun allows us to proceed with our task of freeing antirealism of anthropocentrism. It involves four steps. The first is to show that there is no such entity as the metaphysical self or ego. It will be taken in the present chapter. The second step, closely related to the first, is to show that there is no such entity as consciousness conceived as something that the metaphysical self has or engages in. The third is to show that the notion of oneself presupposes the notion of others.. It will be taken in the next chapter. The fourth step is to explain the antirealist thesis as holding that the dependence of the world on our cognition of it is a case of identity, not a causal or logical relation. It is idenity not because, as ordinary idealism claims, the world is consciousness or mind, but because there is neither a self nor a consciousness from which
332

it might be distinguished. This fourth step will be taken in chapter thirteen. I should remark that to suggest that it is the brain, or some part of it, that engages in consciousness is a case of ignoratio elenchi.359 It may be convenient shorthand in describing facts such as the alterations in and sometimes total cessation of consciousness when certain changes in the brain take place, but it is not the brain that is conscious, the person is conscious. To say this is not to contradict serious neuroscience.

In his Treatise of Human Nature Hume described the self as that to which our several impressions and ideas are suppos'd to have a reference,360 essentially what, much later, Wittgenstein was to call it the philosophical self or the metaphysical subject. Hume agreed that there are ideas in Lockes sense, though he called them perceptions, but unlike Locke he held they were the only knowable or even conceivable entities. Locke had written that Experience convinces us that we have an intuitive knowledge of our own existence.If I know I feel pain, it is evident that I have as certain perception of my own existence as of the pain I feel.361 Hume vigorously disagreed. When looking within himself he found no such entity: when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe any thing but the perception.362 Hume then proposed that the self is merely a collection of perceptions, perhaps the collection of all perceptions, which in effect would be the whole world that he thought was knowable by him.

333

Kant had not read Humes Treatise, relying for his knowledge of Hume on Humes later work An Inquiry concerning Human Understanding, which contained no discussion of the self. But he reached a similar conclusion. The Self was one of the three topics of the transcendental dialectic in the Critique of Pure Reason, the other two being the World-asa-Whole and God. None is an object of experience, whether of sense perception (outer sense) or of introspection (inner sense), and thus none is an object of knowledge. Kant proceeded to replace the notion of the self with the notion of what he called the transcendental unity of self-consciousness, or apperception. But the term selfconsciousness (Selbsbewusstein) is viciously ambiguous. It may mean (1) consciousness of a self, (2) consciousness of one state of consciousness by another state of consciousness, or (3) conscious mental states, in contrast, say, with the unconscious desires postulated by Freud.

A century and a half later, Sartre argued that there is no self-consciousness in sense (1). He called consciousness in sense (2) positional, thetic, reflective, and pointed out that only occasionally is consciousness self-conscious in this sense. And a major tenet in his philosophy was that all consciousness is self-consciousness in sense (3). Only in that third sense is self-consciousness essential to consciousness. The French language does not have one word for self-consciousness as German and English do, conscience de soi being the natural translation of Selbsbewusstein. It thus seems to allow only senses (1) and (2). Sartre was compelled to invent the phrase conscience (de) soi in order to express sense (3).

334

Kant was explicit that in his phrase unity of self-consciousness he was using selfconsciousness in sense (3). The I think must be able to accompany all my representations, he wrote, but he went on to explain that this is so because otherwise something would be represented in me that could not be thought at all, which is as much as to say that the representation would either be impossible or else at least would be nothing for me.363 One would have no access to such a representation. We may note that this indeed is the case with Freuds unconscious desires, though Kant, who lived a century before Freud, knew nothing about them. It is because of this requirement of accessibility that Kant classified the unity of self-consciousness as transcendental, a fact about consciousness that is required for the very possibility of consciousness, rather than an empirical and therefore contingent fact that might be due to the presence in all consciousness of an intuition, outer or inner, of an object that thinks, for which I might stand.

Kant declared in the first edition of the Critique that we do not and cannot have the least acquaintance with the constant logical subject of thinking,364 and he did not change his mind in the second edition, where he wrote that the representation of an I is wholly emptya mere consciousness.365 It is needed only insofar as [the manifold representations of intuition] must be capable of being combined in one consciousness.366 Representations must be so combined if they are to be cognitively relevant. They are combined, however, not as representations for the same subject, whether a phenomenal or a noumenal ego, but as belonging in the same consciousness. As Kant remarked in the first edition, no cognitions can take place in us, no connection and unity among them, without that unity of consciousness which precedes all data of intuition and in relation to
335

which alone all representations of objects is possible.367 In the second edition he pointed out that inferences from [I think] can contain a merely transcendental use of the understanding, excluding every admixture of experience.368 Descartes inference of I exist, meaning by I a thinking thing, which its owner intuits or is directly aware of, was therefore illegitimate.

Kant did appeal to a noumenal self in his ethics, because he thought that the commands of practical reason required a free agent, in violation of the principle of causality. But he made it clear that the existence of that self, like the existence of God and the freedom of the will, which were also required by his ethics, is a matter of faith, not knowledge: How a law can be by itself and immediately a ground of determination of the will (which is, after all, the essential feature of all morality), that is for human reason an insoluble problem and the same as how a free will can be possible.369

In the closing years of the 18th century, Fichte, who rejected Kants distinction between things-in-themselves and things-for-us, declared that the self does exist but only insofar as it is conscious of itself.370 About a century later, however, Nietzsche commented that a thought comes when it wishes, and not when I wish; so that it is a perversion of the facts of the case to say that the subject I is the condition of the predicate thinks.371 It was Hegel, however, who, in the intervening years, refined Fichtes position and prepared the ground for Nietzsches.

In the Encyclopedia of the Philosophical Sciences Hegel claimed that Kants transcendental idealism may be styled subjective idealism, because Kant holds that
336

both the form and matter of knowledge are supplied by the Ego or knowing subject the form by our intellectual, the matter by our sentient ego372 But he congratulated Kant on emancipating philosophy from the soul-thing [Seelending].373 Earlier, in the Phenomenology of Spirit, Hegel had rejected the assumptions that there is a difference between ourselves and cognition, and that the Absolute [reality, the world] stands on one side and cognition on the other, independent and separated from it, and yet is something real.374 He remarked in the Encyclopedia that I is the vacuum or receptacle for anything and everything: for which everything is and which stores up everything in itself.375 A couple of pages earlier, Hegel wrote, in point of contents, thought is only true in proportion as it sinks itself in the facts; and in point of form it is no private or particular state or act of the subject, but rather that attitude of consciousness where the abstract self, freed from all the special limitations to which its ordinary states or qualities are liable, restricts itself to that universal action in which it is identical in all individuals.376 This sentence in effect encapsulated the four steps mentioned earlier as needed if antirealism is to be freed of anthropocentrism. Cognition of the world is not private and particular but rather identical in all individuals and thus abstract. And it does not correspond to the facts, it is in the facts, sunk in them. I shall return to this Hegelian view in chapter 13.

Early in the 20th century, G.E. Moore also expressed doubt about the existence of a self: It is quite possible, I think, that there is no entity whatever that deserves to be called I or me.377 Elsewhere, he agreed with William James that The present thought is the only thinker.378 At roughly the same time, in his Logical Investigations Husserl explicitly denied the existence of the ego, though later, in Ideas, he accepted it. Further
337

thought on the topic had to await Wittgensteins writings, both early and later, as well as Sartres. Both endorsed Humes rejection of the self, and each offered a novel account of the role of the first-person singular pronoun.

In Tractatus Logico-Philosophicus Wittgenstein asked sarcastically, much as Hume had, Where in the world is a metaphysical subject to be found? (5.633), and answered by saying, The philosophical self is not the human being, not the human body, or the human soul, with which psychology deals, but rather the metaphysical subject, the limit of the world--not a part of it (5.641). Wittgenstein s account was contained in an allimportant but seldom discussed sequence of terse sentences, some I have already cited in other contexts, that included I am my world (5.63), There is no such thing as the subject that thinks or entertains ideas (5.631), The subject does not belong to the world (5.632), [S]olipsism, when its implications are followed out strictly, coincides with pure realism. The self of solipsism shrinks to a point without extension, and there remains the reality co-ordinated with it (5.64).

As 5.633 (Where in the world is a metaphysical subject to be found?) makes clear, Wittgenstein's reason for 5.631 (There is no such thing as the subject that thinks or entertains ideas) was not idiosyncratic. It was at least partly phenomenological, like Hume's and, later, Sartre's. The metaphysical subject is not to be found. A simple experiment might confirm the claim. Let the reader focus, not on this page or something remembered or imagined, nor on the readers nose or eyeglasses or anything else the reader happens to see, but on what sees this page. Few have claimed success in such an experiment. Few have claimed to find what sees this page, the subject of the seeing. Nor
338

was the positive view of the self that Wittgenstein offered in 5.63 (I am my world) idiosyncratic. His description of the relation between the self and the world was not entirely unlike Humes (a heap or collection of different perceptions) or Sartres (an absolute, impersonal consciousness that constitutes the world).

If one finds these views absurd, one probably thinks of the self as oneself, say, P.B. Indeed, the assertion that P.B. does not belong to the world, much less that P.B. is the world, is absurd. But P.B.s existence is not in question. What is in question is the existence of the self to which our several impressions and ideas are suppos'd to have a reference that Hume looked for but could not find, of the subject that thinks or entertains ideas that Wittgenstein also could not find. P.B.s existence is not denied, jusy as Hume dd not deny Humes existence and Wittgenstein did not deny Wittgensteins existence. What is denied is the existence of that philosophical self. That there is no such entity is phenomenologically evident. It should be evident also from the grammatical monstrosity of expressions like the I (das Ich), I-hood (die Ichheit), and I-like (ichlich), which have marred some philosophical writings.

Although there is an obvious nonphilosophical distinction between P.B. and the world, there is no philosophical distinction between me and the world. The reason is that in the latter case each must be understood in terms of the other. The point without extension to which the self shrinks, according to Wittgenstein, is what determines the limit of the world, of the reality co-ordinated with that point. To say this indeed is to express a sort of solipsism, but a solipsism that coincides with pure realism. Such solipsism is not alarming, because the I in I am the world refers, not to Wittgenstein, P.B., or any
339

other thing in the world, whether physical or mental, but to the world itself, even if we add, which Wittgenstein did not, that it refers to the world as cognized, viewed, in a certain way. It is solipsism because it says that only the world exists. It is not the traditional solipsism that might say that only P. B. exists. There is no such thing as the philosophical self, there is just that world. To say this, as Wittgenstein saw, is indeed to endorse a sort of realism, a pure realism. It is also to show how antirealism may be freed of anthropocentrism. The impression of absurdity plaguing antirealism vanishes when the implications of its most extreme variety, solipsism, are followed out strictly.

There is no room left for any sort of subjectivism, be it traditional solipsism in metaphysics, skepticism in epistemology, or egoism in ethics.379 There is no room for traditional solipsism because the self is only a point from which the world is viewed, and has no independent ontological status. There is therefore nothing relevant with which to contrast the world, nothing that would be left if the world did not exist. Without the contrast between me and the world, the solipsists assertion Only I exist becomes empty. It could mean either that only the world exists or that nothing exists, neither of which is what the solipsist intended. Solipsism thus loses whatever content it seemed to have. It seemed to say that only I exist, not that either the world or nothing exists.

There is no room for skepticism about the external world: Scepticism is not irrefutable, but obviously nonsensical, when it tries to raise doubts where no questions can be asked (6.51). It is nonsensical because, again, there would be nothing with which the external world could be contrasted, nothing to which it would be external. And no room is left for egoism in ethics. We saw in chapters 3 and 4 that the implication for ethics of
340

rejecting the philosophical self goes beyond the rejection of ordinary egoism. A complete rethinking of the subject matter of ethics becomes required: The sense of the world must lie outside the world. In the world no value exists... (6.41).

The fatal weakness of all subjectivism whether solipsism, skepticism, or egoism is its implicit commitment to a pre-Humean view of the self. Without a subject, there cannot be a subjectivism. Hume, Wittgenstein, and Sartre saw this clearly. Their rejection of subjectivism, though highly technical, was in agreement with common sense, which firmly disapproves of egoism, rejects skepticism without hesitation, and dismisses solipsism as madness. But what is directly relevant to us here is that if Hume, Wittgenstein, and Sartre were right, then the distinction between me and the world could not be taken at face value. To be employed in philosophy, it must receive a radical, nonanthropocentric reinterpretation.

Chapter Twelve. WE AND THE WORLD

1. Consciousness. The rejection of the self is the first of the four steps to free antirealism of anthropocentrism. The second step, which Hegel and, much later, also Moore, Wittgenstein, and especially Sartre in effect combined with the first, is the rejection of
341

consciousness if understood as an entity, presumably a relation between a self and its objects, that is distinct from those objects. The first step is a rejection of subject-object dualism, the second is a rejection of act-object dualism. Indeed, the former entails the latter. If consciousness is a relation between subject and object, and if there is no subject, then there is also no such relation. But, like the first step, the second step can be defended also on phenomenological grounds.

Just as the ubiquity of the pronoun I, which, as Kant noted, encourages the philosophical thought that there is a self, an ego, a special entity that engages in activities such as perceiving, thinking, and imagining, the ubiquity of the verbs perceive (and more specific verbs like see and hear), think, and imagine encourages the philosophical thought that there is consciousness, awareness, a special entity of which perceiving, thinking, and imagining are species, which is the selfs intending, its being directed upon, objects. Taken together, these two thoughts encourage a picture of consciousness as an arrow shot by the self and aimed at objects external to it. The first two steps reject this picture. Of course, they do not deny that there are people or that people perceive, think, and imagine. They reject the philosophical conception of thinking, perception, or imagination that rests on that picture.

There is no phenomenological ground for believing that there is such an entity as consciousness, just as there is not for believing that there is a self. The reader is invited now to attempt to find, not the entity that sees this page, but rather the readers seeing it the seeing itself, the visual consciousness. I believe few would claim success. Few would claim to find anything relevant in addition to the page. Hegel wrote that the truth of
342

[self-certainty] contains nothing but the sheer being of the thing [Sache].380 Heidegger glossed this as the truth of sense-certainty is not concerned with the knower, consciousness, or the I.381

Wittgensteins views, mentioned earlier, about solipsism and skepticism were grounded in his rejection of the philosophical self, but they gained independent plausibility from the conception of consciousness defended earlier by G. E. Moore in the article The Refutation of Idealismand, much later, by Sartre in The Transcendencce of the Ego and especially Being and Nothingness. Wittgenstein presumably was aware of Moores article almost everyone in British philosophy during Wittgensteins first stay at Cambridge was. Moore argued in it that consciousness is diaphanous, translucent: [T]he moment we try to fix our attention on consciousness and to see what, distinctly, it is, it seems to vanish: it seems as if we had before us a mere emptiness. When we try to introspect the sensation of blue, all we can see is the blue: the other element is as if it were diaphanous.382 In general, Moore wrote, that which makes the sensation of blue a mental fact seems to escape us: it seems, if I may use a metaphor, to be transparent we look through it and see nothing but the blue.383

Several decades later, Sartre eloquently described the role of consciousness in constituting the world as the revelation of the objects conceptualized as a world. Consciousness exhausts itself in its objects, he wrote, precisely because it is nothing but the revelation of them: Consciousness is outside; there is no within of consciousness.384 Indeed, as we saw earlier, Sartre held that consciousness is always also self-consciousness, but not in the sense of being consciousness of a self, or even of
343

another state of consciousness, but rather in the sense of being nonpositionally (nonthetically) conscious of itself by being, so to speak, a conscious consciousness, e.g., a conscious seeing of this page or a conscious, as contrasted with an unconscious, desire. Sometimes a state of consciousness is positional, thetic, self-consciousness, meaning by this that it has a consciousness as its object, not itself but another state of consciousness. Even then, however, consciousness, though positional, exhausts itself in its object, which in turn exhausts itself in its own object.385 Whether perceptual or conceptual, consciousness is not a thing. One may even go so far as saying that it is nothing. To use a word Heidegger had applied, consciousness is only the lightening of its objects, like the coming of dawn, which lightens, reveals, the rocks, bushes, and hills that had been invisible in the darkness of the night, but is not itself an object.386

If Humes and Wittgensteins denial of the existence of a self, and then Moores and Sartres conception of consciousness, are accepted, we can grasp more clearly the conception of ourselves needed for understanding the key thesis of antirealism that, as Nelson Goodman put it, we make the world.

We make the world is, of course, a metaphor. It means that what the world is for us depends, at least in part, on our cognition of it. It depends not only on how we perceive but also on how we conceptualize and describe it. As I suggested in the Introduction, the thesis of antirealism, so understood, is not entirely foreign to common sense. We can easily understand what would be meant by saying that the world of the fly is quite different from our world, and that the world of the octopus is even more so, as long as we are aware that the sense organs of a fly or an octopus differ radically from ours and that
344

therefore a fly or an octopus perceives the world not at all as we do. We can easily understand what would be meant by saying that our world is probably quite different, perhaps indescribably so, from the world of an intelligent extraterrestrial life form, as long as we are aware that almost certainly it would be not only perceived but conceptualized very differently. Even radically different human cultures are often said to have, live in, different worlds because the concepts through which they understand the world are different.

While Kant held, roughly, that the world is dependent on our consciousness of it, Hegel held, also roughly, that the world is that consciousness. Their positions are likely to be found more plausible if we do not limit consciousness to perception, and then if we suppose that perception necessarily involves conception. The defect of empiricism, Hegel wrote, is that it makes sense-perception [Wahrnehmung] the form in which fact is apprehended, but the process of knowledgeproceeds to find out the universal and permanent element in the individual apprehended by sense. This is the process leading from simple perception to experience [Erfahrung].387 For, in propositions where the subject-matter is due to the senses e.g. This leaf is green we have such categories introduced, as being and individuality.388 More than a century and a half later, Wilfrid Sellars argued against the myth of the given, and held that all cognition presupposes econceptualization, that sense perception as such grasps no facts.389 To say that perception involves conception need not mean that perception is propositional, that statements of the form S perceives x entail statements of the form S perceives that p. It need only mean that statements of the form S perceives x entail statements of the form S perceives x as F. But surely this is false of neonates, and perceiving-as must not
345

be confused with perceiving-that. A neonate first just perceives (sees, feels) Mother, and neither perceives Mother as the neonates mother nor that Mother is the neonates mother. The latter are later stages of the chilkds cognitive development. Presumably perceiving Mother is first followed by perceiving Mother as the chilss mother, and perceiving that Mother is the childs mother comes later, perhaps never if it presupposes the acquisition of language.

Following Heidegger, I suggested earlier that consciousness is like the coming of dawn, the lightening, revealing of its objects. Had those objects been there, had they existed, before the coming of dawn? Or do they exist only if revealed? In other words, are they mind-dependent? If by mind we mean, as Moore and Sartre did, a consciousness that is not a thing but merely the revelation of things, then the mind cannot be meaningfully said to enter in causal, logical, or any other relations, and thus to depend on anything or to have anything depend on it.

The image of the coming of dawn has a clear, natural, application to perception. At dawn things become visible, knowable. Hence the initial plausibility of an antirealism such as Berkeleys: to be is to be perceived. The image does not have a clear or natural application to conception. For the plausibility of antirealism with respect to conception we must go beyond Berkeley to Kant and especially Hegel.

The distinction between perception (Kants sensibility) and conception (Kants understanding) gives rise to important, already mentioned, questions, many dealt with in detail by Kant. One was just mentioned: whether objects can be perceived without
346

being conceptualized. Clearly, even if they can, they would not be objects of advanced cognition. But another, arguably more fundamental, question is whether existence and nonexistence are themselves outcomes of conceptualization, of the application of the concepts of existence and nonexistence. If they are, then the answer to the question whether the objects revealed by the coming of dawn, i.e,, before being conceptualized, existed before being revealed would be that they neither did nor did not.

The status of the concept of existence has been one of the most controversial topics in philosophy. Kant claimed that existence is not a real predicate, and Russell attacked Meinong for holding that it is a predicate and that so is subsistence. Kants claim was motivated partly by doubts about the validity of the ontological argument for the existence of God, and Russells attack on Meinong by Russells invention of the theory of definite descriptions. According to the latter, existence is properly expressed only by the particular (existential) quantifier. Pegasus is a horse seems true but in fact is false, according to Russell, because it really says There is one and only one x such that x pegasizes. (The verb pegasize was Quines later invention roughly, it is a synonym of is identical with Pegasus.) But, to be relevant, the theory of definite descriptions must stipulate that the individual variable x in Russells analysis takes as values only things that exist. Otherwise, Pegasus is a horse would be true. Therefore, the theory could not be used consistently to argue against Meinong. Moreover, by making that stipulation the theory itself would be employing exist as a predicate.

2. The I that is We and the We that is I.

347

Antirealism faces the paradox of seeming to say that the world is dependent, at least for its nature if not also existence, on some of its zoological parts whether just on me or on us. I have argued that there is no special entity self, ego, subject denoted by the firstperson singular pronoun. I have also argued that to avoid the paradox we must understand first-person singular pronouns (I, me, mine), when used in philosophical contexts such as Cartesian doubt and the realism/antirealism debate, as impersonal. I shall argue now that in such contexts the first-person plural pronouns should also be understood as impersonal.

The choice between I and we is often stylistic. A politician may use the latter in order to avoid showing conceit. It is often stylistic even in philosophy. In epistemology we speak of skepticism regarding the limits and extent of our knowledge, though what we actually consider is usually the extent and limits of my knowledge. But the choice is not stylistic in ethics, which from Plato to Hobbes to Sidgwick to Rawls has been preoccupied with egoism, the view that advocates the pursuit of my own good rather than of our good. Nor is the choice stylistic in stating the antirealist thesis and understanding the metaphysical doctrine of antirealism.

I have argued that in philosophical contexts such as Cartesian epistemology and the realism/antirealism debate the first-person singular pronoun should be understood as indicating a perspective, a view, a cognition, not a person. Would that cognition be my cognition? In such contexts, indeed in any advanced context, a cognition is inherently social. My worldview is the same as our worldview, just as my view of Manhattan is the same as our view of Manhattan when both are the view of Manhattan from the Empire
348

State Building, the one that, say, a tourist guide describes. In the philosophical contexts just mentioned, there is no difference between the role of first-person singular pronouns and the role of first-person plural pronouns. The distinction between me and us lacks significance in such contexts.

Needless to say, this is not true in most nonphilosophical contexts. In everyday life as well as in science, usually I indicates, say, P.B. and we indicates P.B. and you. The difference is obvious. P.B. weighs 160 pounds, P.B. and you together weigh 300 pounds. P.B. visited Charleston last year, we did not. But in doubting the existence of the world or considering the question of its mind-independence, there is no relevant difference between me and us, between my doubt and our doubt, between my stand and our stand on the question. There are such differences only when we assume that P.B. and you are parts of the world, a natural assumption in other contexts but one that cannot be made when doubting the existence of that world or considering whether it is mind-independent.

In the philosophical contexts in which the use of the first-person singular pronoun I is impersonal, the use of the first-person plural pronoun we is also impersonal, and for the same reason. But, though required in those contexts, the impersonal use of we is not limited to philosophy. It is common in science, as in We know that the speed of light cannot be exceeded, the impersonal sense of which becomes explicit when restated as It is known that the speed of light cannot be exceeded, Physics has found (established, discovered, confirmed) that the speed of light cannot be exceeded, or According to physics, the speed of light cannot be exceeded. An impersonal use of we is also common in moral contexts, as in some occurrences of We dont torture prisoners, the
349

impersonal sense of which becomes explicit when restated as Torturing prisoners is unacceptable in this country or Morality prohibits the torture of prisoners. It is common also in etiquette, as in We dont wear flip-flops at weddings, which means that wearing flip-flops at a wedding is a violation of proper dress rules, custom, or expectation. Clearly, in such cases, we does not refer to any particular persons knowledge, moral attitude, or dress rules. This is made explicit when the speaker is asked Who knows that the speed of light cannot be exceeded? or Who doesnt torture prisoners? or Who does not wear flip-flops at weddings? and replies: I didnt mean anyone in particular.

The deeper explanation of the occasional interchangeability of I and we is that there cannot be a drastic divergence between my sense perception and concepts and those of other persons. The other is taken by me to be capable of judgments that I must consider only if taken to perceive and understand the world at least roughly as I perceive and understand it. Otherwise, not only I would not understand what the other says, I may not recognize it as language. Verbal communication with humans presupposes that we are not speaking about different things or a different language.390 Nonverbal communication with our pets presupposes that they perceive, however differently, the same things that we perceive, e.g., the food tray or the door.

The distinction between ones own and others cognition of the world has little application when cognition is advanced. Its application to perception (including introspection) is indeed clear, but beyond early childhood ones cognition of almost anything includes also conception, understanding. And that depends on what one has
350

learned from others, in the common, public language mastered after infancy. The same is true, of course, of anyone elses cognition. Therefore ones advanced cognition is dependent not on the aggregate of others cognitions but rather on a systematic impersonal whole that seems to have life and properties of its own. This is familiar in the case of scientific cognition, but it is true also of everyday cognition of country, town, automobiles, and virually everything else. We usually acknowledge that in any cognition above that of infants one is dependent on others: ones parents, teachers, the authors of the books and articles in magazines and newspapers one has read, the people one has talked with at home, in school, or in the street, or has heard in the classroom or on radio or television. If one has knowledge that is strictly ones own, underived from and unaffected by others, it is of the sort enjoyed by neonates and qualifies as knowledge only barely, namely, mere sensation. Educated persons explicitly or implicitly take physical things to be as physicists say they are, the past to be as historians say it was, and mathematical truths to be those vouchsafed by mathematicians. And physics, history, and mathematics themselves are inherently social. This is especially evident today. The requirement that experiments be repeatable by others as well as by oneself is sacrosanct. The 2009 status report by the Large Hadron Collider listed 2900 authors. The Wikipedia on the Internet is familiar to most of us. Even in everyday life, individual claims to knowledge count for little if they do not survive scrutiny by others.

The cognitive dependence of oneself on others is not only causal. It is also logical. It was an essential thesis of Wittgensteins later philosophy, as well as Sartres. Much of the history of 20th century Anglo-American philosophy was epitomized by Wittgensteins
351

move from the thesis in the Tractatus that the world is my world to the thesis in the Philosophical Investigations that that a private language is logically impossible and that understanding a public language presupposes agreement in judgment. And many consider the highlight of 20th century continental philosophy to be Sartres argument that only through the Others look can I see myself as an object, as the entity in the world that I am that it is the Other who defines me as a person.

But it is to Hegel that we must go for the original account of the logical dependence of oneself on others. He insisted on the need to move from the I (das Ich) of the colourful show of the sensuous here-and-now to the I that is We and We that is I (Ich, das Wir, und Wir, das Ich ist). He called the latter Spirit, and described it as this absolute substance which is the unity of the different independent self-consciousnesses which, in their opposition, enjoy perfect freedom and independence.391 This was a description of what we may call society. As Terry Pinkard puts it, Hegel reconceptualized the unity of thought and being as an intersubjective unity.392 And J. N. Findlay explains: Hegel holds that the understanding of other minds, far from being more obscure than the understanding of things, is the model and paradigm in terms of which intercourse with things can assume a limited clarity. In all intercourse with things we are striving towards the complete penetration and lucidity of social intercourse.393 But Pinkard and Findlay are downplaying the metaphysical significance of Hegels view. Human cognition may be social, but to say this of reality, of the Absolute, would be nonsense. Recognition of the primacy of society over the individual is only one step, however necessary, to grasping the Absolute.

352

Hegels central argument was that A self-consciousness exists for a self-consciousness. Only so is it in fact self-consciousness; for only in this way does the unity of itself in its otherness become explicit for it.A self-consciousness, in being an object, is just as much I as object.394 To exist is to be an object, and self-consciousness must earn its objecthood and therefore existence by being an object of another self-consciousness. The two self-consciousnesses recognize themselves as mutually recognizing one another. At first, of course, such recognition appears as a split between two extremes, one selfconsciousness being only recognized, the other only recognizing. Each is indeed certain only of its own self, but not of the other. But its own self-certainty would have truth only if its own being-for-self had confronted it as an independent object.[T]his is possible only when each is for the other what the other is for it.395 That selfconsciousness is a social phenomenon had been suggested by Fichte a decade earlier in the Foundations of Natural Right.

According to Hegel, Spirit (mind, Geist) develops from the primitive stage of subjective spirit sensuous cognition to its second stage of objective spirit, by which he meant the normative customs and traditions of everyday life [Sittlichkeit], the family, the state, and institutions such as corporations and professional guilds, which would include what today we call the academic disciplines, and culminates in its third stage of absolute spirit, which includes art, religion, and the complete, perfect, knowledge Hegel called philosophy. If by mind we mean cognition, then at the stage of sensation the mind is primitive, undeveloped, and becomes developed only at the social stage. Society can be said to be a stage in the development of mind in the sense that it would not exist if the perceptions, thoughts, and judgments of its individual members did not. Indeed, academic disciplines,
353

including philosophy, can be said to be mental in the sense that they would not exist if the perceptions, thoughts, and judgments of their individual members did not.

It may still seem that we are left with anthropocentrism. Wittgensteins public language, after all, is a human language, Sartres Other is a human being, and Hegels objective spirit is a society of humans, not of ants or Martians. (Hegel explicitly held even that the Absolute achieves self-knowledge in human beings, those he regarded as philosophers, his position thus appearing to be not just anthropocentric but anthropomorphic.) However, though all this is true, it is not the whole truth. Statements about a public language, Wittgensteins agreement in judgment, or Hegels social institutions are not reducible to statements about human beings, even though the language, agreement, or institutions would not exist if human beings did not. The public that speaks the public language, the parties to the agreement in judgment, and the institutions of society are not mere collections of human beings. Nor, of course, are they themselves human beings. In an important sense, they are impersonal. They are not mere collections of the particular individuals they include, just as universal statements are not mere conjunctions of the singular statements that instantiate them even though they would not be true if any of those singular statements was not. Hegel and Wittgenstein could still be charged with anthropocentrism in an extended and rarified sense, but their views also suggest how anthropocentrism can be avoided, which will be the topic of the next chapter.

Hegel famously thought that the discipline of philosophy itself was an example of the priority of society over the individual. Philosophical views are historical achievements. They are inseparable from the history of philosophy. Earlier philosophical systems,
354

despite their diversity, are preserved in those developed later.396 [T]he diversity of philosophical systems is the progressive unfolding of truth, not simple disagreements. 397The history of philosophy is either one philosophy at different degrees of maturity; or the particular principle, which is the groundwork of each system, is but a branch of one and the same universe of thought. In philosophy the latest birth of time is the result of all the systems that preceded it. 398 Hegel was fully aware that a philosophers thought, however original, is rooted in the thought of other philosophers, past and present, read or heard, as well as in the culture to which it belongs and the language ii employs. The content of a philosophical work is incomprehensible in abstraction from the content of previous works. The same is true, of course, of the other disciplines. Einsteins work was strikingly original but inseparable from that of his predecessors. Instruction in philosophy has always been mainly instruction in the history of philosophy, even if only the history of its latest period or in a single country.

Commentators sometimes say that when Hegel described philosophy as the most advanced stage of the development of spirit he meant his own philosophy. Indeed, he did hold that the Absolute achieves self-knowledge through humans. But Hegel would have insisted that no individual human beings knowledge could be identified with the Absolutes self-knowledge. Discussing Spinoza, he wrote, Such a standpoint [as Spinozas] is not to be regarded as an opinion, a subsjective, arbitrary way of thinking of an individual, as an aberration of speculation; on the contrary, speculative thinking in the course of its progress finds itself necessarily occupying that standpoint and to that extent the system is perfectly true; but it is not the highest standpoint.399 Perhaps Hegel

355

thought, or hoped, that the standpoint of his syetem was the highest, though I doubt that he did. If he did, he would hardly have been unique among philosophers in this respect.

That advanced cognition is social is obvious in the case of cognition that necessarily involves the use of language. Some cognition is possible without language. A neonates and a dogs sense perception are examples. But distinctively human cognition, like that exemplified by the sciences and mathematics, obviously is not. Perhaps neonates and dogs have an innate language of thought that is needed even for sense perception, and perhaps God could not know the world without employing a divine language, but neither possibility will be discussed here. Our concern is with distinctively human cognition, the sort of cognition neonates and dogs clearly lack and God is believed to infinitely surpass. Wittgensteins perhaps most persuasive reason for holding that language is necessarily public, social, was his description of an imaginary private language for keeping track of ones sensations. Even if possible, such a language would be so impoverished as to be pitiful. A public language is indispensable to any cognition above that of infants. And it does presuppose agreement in judgment at least regarding which sounds or marks to count as words and which words they happen to be.

Until Hegel, epistemologists from Plato and Aristotle to Descartes, Hume, and Kant took for granted that cognition is individual, personal. In the Introduction I called this view cognitive individualism and contrasted it with what I called cognitive socialism. Hegel saw that it is true only on the primitive, subjective, level, and that advanced, truly objective, cognition is necessarily social. It is the nature of humanity, he wrote, to press onward to agreement with others; human nature only really exists in an achieved
356

community with others.400 Elsewhere he described civil society (brgerliche Geselschaft) as a universal family.401 Self-sacrifice in defense of ones community, whether country or infantry squad, is not an empty ideal but a fact. It is vividly present among front-line combat soldiers: the survival of the squad, of ones buddies, is a goal overriding any other goal.402 Hegel became famous partly because of the moral and political implications these views were taken to have. Cognitive socialism, however, by itself has no such implications. One of its elementary tenets is that advanced cognition requires a public language. This is hardly a moral or political tenet. Cognitive socialism has been acknowledged in the sociology of knowledge403 and in social epistemology,404 though not by appealing to metaphysical considerations like Hegels.

Initially, as in Kants version, antirealism insists on the dependence of the world on my cognition. In Hegels version, it insists on its dependence on our cognition. In Goodmans, Dummetts, and Putnams versions, it insists primarily on the dependence of the world on our language. Heidegger assertion that it is language, not we, that speaks, was not a continental idiosyncrasy.

To say that the world depends on me seems absurd. To say that it depends on us may appear to multiply the absurdity. But what is at issue is dependence on cognition, and the dependence of the world on our cognition is less jarring. Its dependence on me would be abhorrent to common sense because it makes the world, the paradigm of what is objective, appear subjective. Its dependence on us would be less alarming. Not only would it be not dependence on a solitary person, it would not be dependence even on a mere collection of persons. It would be dependence on a cognition that is irreducibly
357

institutional. There is no mystery about what such a cognition might be. An example is the cognition, knowledge, embodied in any developed discipline, say, physics. As cognition, a body of presumed knowledge, physics is not reducible to the collection of individual physicists cognitions. And while it would be glaringly absurd to hold that the physical world is what I say it is, it would not be absurd to hold that the physical world is what physics says it is. The move from I to we does not signify total abandonment of anthropocentricity. But it does signify abandonment of subjective anthropocentricity.

Berkeleys paradigm of cognition was perception, not conception. (He argued vigorously against abstract ideas.) Indeed, this is the immediate, natural, way to think of cognition. It explains why we also think of our cognitive faculties as personal, not social, and of knowledge as a personal, not social, achievement. This is why we find cognitive individualism more plausible than cognitive socialism and why antirealism usually has been understood as subjective. Indeed, it is more plausible in the case of perception. Cognitive socialism comes into its own when we recognize the role in cognition of conception, the understanding, and especially of description. While perception is inherently individual, personal, and subjective, conception and especially description are inherently social, public, and objective. A neonates cognition of Mother is entirely perceptual and dependent on no one elses perception of Mother. But the childs later conception of Mother as mother is totally dependent on others, obviously so when expressed in language. Even if Berkeleys idealism could be defended on the level of perception, it was Kants transcendental idealism and Hegels absolute idealism that were plausible on the level of conception. If we accept Kants and Hegels view that perception presupposes conception, then conception is primary.
358

3. Idealism. Traditional idealism is the original version, indeed several versions, of antirealism, in the broad sense of antirealism explained in the Introduction. According to the idealist, everything is mental. According to the antirealist, everything is dependent on our cognitive faculties, at least insofar as known or knowable. Cognitive faculties are paradigms of what is mental, even if understood neurologically. The idealist and the antirealist thus agree that insofar as the world is known or knowable, it is minddependent. Moreover, the idealist is likely to support idealism with some of the same arguments that the antirealist employs. But idealism avoids the difficuly faced by antirealism of explaining the difference between cognition and its object by denying that there is no difference, claiming that the perception of a tree and the tree perceived are no more distinguishable than the feeling of pain an the pain felt. Berkeley remarked that nobody will pretend that real pain either is, or can possibly be, in an unperceiving thing, or without the mind, any more than its idea.405 Of course, to avoid this difficulty much more is needed than insisting that to be is to perceived, and the challenge for antirealism, beginning with Kant, has been to provide the detailed qualifications, explanation, and argumentation needed to make Berkeleys claim at all plausible..

There are many varieties of idealism, but Berkeleys, Kants, and Hegels have been crucial in its history. They are the only varieties I shall consider here, and will focus on Hegels. Berkeleys made idealism a technical philosophical position. Kants transcendental idealism inaugurated modern antirealism. Hegels absolute idealism freed
359

antirealism of anthropocentrism. Berkeleys idealism rested on the assertion that to be is to be perceived, which is a version of the thesis of an antirealism limited to our perceptual cognitive powers. This is why, though it seems incredible, there is no puzzle about what it says similar thoughts have occurred to many teenagers. Kants transcendental idealism acknowledged the essential place among our cognitive powers of conception, the understanding. It is not likely to occur to teenagers. Hegels absolute idealism avoided the anthropocentrism latent in Berkeleys as well as Kants versions, and as a consequence has baffled even philosophers.

Despite their obvious similarities and historical connections, idealism and antirealism differ in important ways. Understood as holding that everything is mental, spiritual, idealism is sometimes either greeted as a pleasingly warm view of the universe, which its 20th century critic G. E. Moore said he wished were true, or rejected as utterly absurd .406 Antirealism, by contrast, stays in the bloodless realm of philosophical technicality and arouses neither enthusiasm nor alarm, even though, as we have seen, it can seem to have absurd implications, for example, that there was no earth before humans evolved.

As stated above, idealism does not entail antirealism, and antirealism does not entail idealism. The idealist says that all reality is mental, the antirealist usually does not. What is mental need not be dependent on our cognitive faculties, and what is dependent on our cognitive faculties need not be mental. Idealism is a theory about the nature of reality. Antirealism is not. It says nothing about what reality consists of, or about its nature. Its thesis that at least knowable reality depends on our cognitive faculties says nothing about the nature of that reality. In this respect it resembles theism: few theists hold that the spatiotemporal world is divine even though all theists hold that it is dependent for its
360

existence and nature on a divine being. Whether reality is mental remains for the antirealist a further question. How it is answered, and how idealism is understood, would depend on what meaning is attached to mental. Presumably the idealist does not claim that rocks, firewood, and puddles of water are mental in any familiar sense of mental. In everyday discourse they are paradigms of what is not mental, just as perceiving, thinking, and feeling are paradigms of what is mental. If we use mental in accord with those paradigms, then we must say that there would be no rocks, firewood, or puddles of water if everything were mental.

In his classic paper The Refutation of Idealism, which marked the end of the dominance of idealism in British philosophy and inaugurated 20th century realism, G. E. Moore defined idealism as the view that the Universe is spiritual.407 This in effect means that reality is either a spirit or consists of states of one or several spirits. Moore used spirit as a synonym of mind, a spiritual state being what today would be called an irreducibly mental state, in particular, a state of consciousness. According to Moore, the idealists thesis, he had in mind Berkeleys, that to be is be perceived seems plausible only because it confuses the object of perceptual consciousness with its content, a confusion made possible by the fact that consciousness itself has no content, is diaphanous, transparent. Moore thus anticipated Sartre, who decades later wrote that consciousness has no inhabitants and exhausts itself in its objects. But Sartre also proposed that the activities traditionally attributed to the ego are performed by an impersonal consciousness, which constitutes the world, and that the ego is an object in the so constituted world, not the subject that constitutes it. He thus combined the rejection of Berkeleys idealism with espousal of a quasi-Kantian transcendental idealism. Moore
361

would not have endorsed the latter. However, he did agree with Kant that the idealist, presumably Berkeley, employs the verb perceive in a sense that excludes conception, thus impoverishing idealism.

Berkeley could not consistently regard the subject-term of perceive to be the name Berkeley, for Berkeley, the 18th century bishop, had a material body even if also an immaterial soul. The subject-term that his use of perceive would take therefore could only be I, understood as standing, not for Berkeley, but for a self or ego, which may belong to but is not identical with a human being. Hume, Wittgenstein, and Sartre looked for such an entity but failed to find it. Without it, however, there would not be the perceiving to which Berkeley appealled as the measure of what there is. For there would be nothing to engage in that perceiving, even if there were something to be perceived.

Idealism became dominant in philosophy after philosophy took the new way of ideas in the 17th century, and its development was dependent on the sense attached to idea. Descartes and Locke proposed that the primary and perhaps only objects of knowledge are our ideas, though neither was an idealist. Descartes wrote, Of my thoughts some are, as it were, images of things, and to these alone properly belongs the name IDEA.408 Locke used the term in a similar sense: the term [idea] stand[s] for whatsoever is the Object of the Understanding, when a man thinks.409 This sense was too broad, however, because it implied that if material objects were directly perceived, which both Descartes and Locke denied, they would count as ideas, which both would have found absurd. Locke explained further that by idea he meant, not only concepts (which he called abstract ideas) and mental images, but also sensations and feelings (passions).
362

This use of idea was therefore entirely different from Platos use of

, to

which nonetheless it must be traced, as must also the medieval philosophers use of it for what they regarded as the forms of divine perception.

Berkeley used idea in Lockes sense, but earned his classification as an idealist by holding that minds and their ideas were all that there is, a view Descartes and Locke would have found appalling. Hume used perception in Lockes sense of idea, but distinguished within perceptions between impressions and ideas: Those perceptions, which enter with most force and violence, we may name impressions: and under this name I comprehend all our sensations, passions and emotions, as they make their first appearance in the soul. By ideas I mean the faint images of these in thinking and reasoning.410 Both Descartes and Locke distinguished sharply between ideas and their owner, Descartes thinking thing or Lockes self. German philosophers called it ego, or just plain I (das Ich). But by deniying that he could find such an entity within himself, Hume originated a totally different line of thought, which culminated in Wittgenstein and Sartre.

Kant used Vorstellung, commonly translated as representation, roughly in Lockes broad sense of idea, but he made a sharp distinction between intuitions (Anschauungen), which are immediately related to the object and are singular, and concepts (Begriffe), which are mediate, related to objects by means of marks that can be common to several things. A concept, he held, is either empirical or pure: Pure concepts have their origin solely in the understanding, not in images of sensibility, and are called notions. Kant used also idea (Idee), and like Descartess and Lockes his
363

use was rooted in Platos

. But he described ideas very differently: A concept

made up of notions, which goes beyond the possibility of experience, is an idea or a concept of reason.411 Unlike the concepts of the understanding, the ideas of reason have no application to experience. We have no knowledge of God, freedom (of the will), and immortality, but we do have ideas of reason of them, he famously held. We have no experience of God, freedom, or immortality, but God, freedom, and immortality are unavoidable objects of reason, and as such are essential to morality.

Hegel distinguished between the several modes of feeling, perception, desire, and will, which are in general called ideas (mental representations [Vorstellungen]), and thoughts [Gedanken], categories, or, in more precise language, adequate Notions [Begriffe], which philosophy puts in the place of the generalized images [Vorstellungen] we ordinarily call ideas.412 But he acknowledged that the mind [consciousness, das Bewusstsein] makes general images of objects, long before it makes Notions of them413 While in the case of Kant the usual English translation of Begriff has been concept, in the case of Hegel it is Notion, in order to mark Hegels distinctive metaphysical understanding of the nature of Begriffe. Hegel was aware that No complaint is oftener made against the Notion than that it is abstract, but remarked Of course it is abstract, if abstract means that the medium in which the Notion exists is thought in general and not the sensible in its empirical concreteness.414

Though Hegel described Notions as thoughts, he so described also judgments and syllogisms.415 Gedanke in German is as ambiguous as thought is in English. It can refer to (1) a thinking, a subjective mental action, an event, whether a judging, a
364

supposing, or reasoning, (2) the objective content of a judgment (as it does in Freges writings), or (3) a component of that content. For Hegel, it referred to a Notion only when used in the third sense. This is also the main sense of concept in contemporary AngloAmerican philosophy. But by the Notion of x Hegel meant the comprehension, grasp [Griff] of x, which is a sense much stricter than the usual sense of concept. What is usually so named by no means deserves the name of Notion, Hegel wrote.416 He called the Notion the soul [Seele] of objective reality.417 It is in this sense that Hegel held that the Absolute, reality, is Thought, and thus earned his classification as an idealist.

He was also literally an idealist because of his use of Idea (Idee), which was quite different from Kants as well as Descartess and Lockes. Hegel did not make Kants sharp distinction between Begriff and Idee: The Idea, he wrote, is the adequate Notion, that which is objectively true, or the true as such[S]omething possesses truth only in so far as it is Ideareserving [thus] the expression Idea for the objective or real Notion and distinguishing it from the Notion itself and still more from mere pictorial thought418 Elsewhere Hegel wrote, The thought, which is genuine and selfsupporting, must be intrinsically concrete; it must be an Idea; and when it is viewed in the whole of its universality, it is the Idea, or the Absolute. The science of this Idea must form a system.419 The absolute Idea is the rational Notion that in its reality meets only with itself [I]t contains all determinations within it. It is the only object and content of philosophy, of perfect science.420 This is why it must be within itself a circle in which the first is also the last and the last is also the first,421 a consequence sometimes cited today as an objection to the coherence theory of truth. An Idea, in this sense, may be thought of as a Notion that includes its logical relations to all other Notions, and thus
365

coincides with their system. Hegel sometimes also called it the whole Notion.422 The Notion followed Being and Essence as the third part of the exposition of the Absolute Idea in the Science of Logic. The Notion is not yet complete, but must rise to the Idea, which alone is the unity of the Notion and reality.423

Hegels phrases the rational Notion and the whole Notion do not stand for kinds of Notions, nor do his phrases absolute Idea and logical Idea stand for kinds of Ideas. They make explicit or emphasize certain characteristics of Notions or Ideas. Indeed, this is true also of the three stages of the development of the Absolute: Idea, Nature, and Spirit. They are not three distinct realities. We may think of the Idea as truth, of Spirit as cognition, and of Nature as the antirealist advocates and Hegel of course accepted, namely, as dependent on, inseparable from, no more than an aspect of cognition.

In general, Hegels distinctions are not between entities but between ways of regarding or thinking of entities and the world. This would be true even of distinctions among things in the world, which would be deceptive if taken at face value because to comprehend, understand, one of them requires comprehending or understanding the rest. To be sure, we can perceive one thing without perceiving another, which was the phenomenological fact supporting the atomism of the British empiricists, and later of Russells logical atomism, but Hegel concern was with cognition as understanding, not mere acquaintance, and his standards for understanding were very, perhaps impossibly, high. But in accepting these standards his concern was not with the empirical facts about human understanding but with the nature of understanding, what Plato might have called

366

its perfect, eternal form, of which a humans understanding is at most an imperfect image.

The absolute Idea. Hegel wrote, is the rational Notion that in reality meets only with itself. It alone is being, imperishable life, self-knowing truth, and is all truth[N]ature and spirit are in general different modes of presenting its existence, art and religion its different modes of apprehending itself Philosophy has the same content and the same end as art and religion; but it is the highest mode of apprehendimg the absolute Idea, because its mode is the highest mode, the Notion.424 Unlike art and religion, philosophy aims at understanding, grasping the absolute Idea. It is the Notion that can have only itself as object. It is a grasp of all truth. It develops into Nature and then Spirit. (Art, religion, and philosophy are the three stages of development of Spirit.)

There can be only one whole Notion, since there can be only one system of all Notions, as indeed any plausible theory of language and truth would hold. While the comprehensions of different things are themselves different, a full comprehension of one would involve comprehension of all, since, Hegel held, they are all related essentially. For example, while one can comprehend the French revolution to different degrees, a full comprehension of it would involve comprehension also of French history before and after the revolution, and of course much more than French history, indeed much more even than history. The truth of one statement, if fully understood and justified, presupposes the truth of all other statements.

367

Hegel did not mean by the Absolute some mysterious object or part of reality. He meant reality itself, which of course must be understood as relative to nothing else, as unqualified, unconditional, all-encompassing. M. J. Inwood, one of Hegels most astute commentators, remarks, this expression [the absolute] is meant to convey no more than reality as a whole and even a materialist, for example, might agree that reality as a whole cannot be dependent on, or conditioned by, anything distinct from itself.425 The Absolute is infinite in the sense of being unlimited. But to understand this sense we must resist thoughts of infinite space and infinite time, or appeals to mathematical truths such as there is no greatest number. We must think rather of the infinite in the literal sense as what is limited by nothing (unlimited, indefinite, indeterminate), self-contained, complete, and of the finite as what is limited (definite, determinate), incomplete, a part of something else. And the image of infinity on which we must rely would not be that of the straight line, as it might be when thinking of infinite space or time or of the infinity of the series of natural numbers, but rather that of the circle.426

Comparison with Spinoza, to whom Hegel often refers, saying for example that to be a follower of Spinoza is the essential commencement of all philosophy,427 might be instructive. Corresponding to the Notion of the absolute, he wrote,is the notion of substance in Spinozism.428 Hegel regarded his phrase the Absolute as co-referential (not synonymous) with God. Spinoza defined God as Being absolutely infinite, that is to say, substance consisting of infinite attributes, each one of which expresses eternal and infinite essence. He defined substance as that which is in itself and is conceived through itself, thus in effect following Aristotles definition of primary substance in the Categories as that which is neither predicable of a subject nor present in a subject.
368

Hegel called substance the final unity of essence and being, a subsisting in and for itself.429 Spinoza defined an attribute of substance as what the intellect perceives of substance as constituting its essence. The two attributes of substance that are known to us are extension and thought, the order and connection of the modes of which are the same because substance thinking and substance extended are one and the same substance, which is now comprehended under this attribute and now under that. 430 The similarity to Hegel is striking. Hegels Absolute corresponds to Spinozas God, Hegels Geist to Spinozas attribute of thought, and Hegels Nature to Spinozas attribute of extension. It would not be far-fetched to regard Spinozas view of the relation between extension and thought as analogous to Hegels view of the relationship between Nature and Spirit, and Spinozas view that, like all attributes of substance, thought is perceived by the intellect as constituting the essence of substance, as analogous to Hegels view that the Absolute is Thought. Spinoza and Hegel can therefore be called, as indeed both often were, atheists as well as theists, materialists as well as idealists.

Finite things are dependent on the Absolute for both their nature and their existence, but the dependence is neither causal nor logical. It is better understood as analogous to the dependence of the parts of a biological organism on that organism. And the Absolute is itself dependent on its finite parts, much as an organism is dependent on its parts. Hence Hegels characterization of the Absolute as concrete, not a mere abstraction. Knowledge of it is itself an organic whole, in the sense of being systematic. Indeed, we may note, this is true of any advanced cognition, e.g., a theory in a developed science. The statements in it depend on each other for both their content and truth. Its concepts presuppose each other. Another useful analogy would be the words in a language: they seldom if ever can
369

be understood in isolation from the language or from each other. Hegel famously demanded that philosophy be systematic. His work usually met that demand.

Indeed, the Absolute can be thought of as ideal or absolute knowledge. This is what Hegel meant by saying that it is Thought. But it is not anyones knowledge or thought. Nor is it knowledge of anything other than itself. For there is nothing to which it is relative, whether as subject or as object. The Idea is truth [das Wahre] in itself and for itself the absolute unity of the Notion and objectivity [des Begriffs und der Objektivitt] . Its ideal content is nothing but the Notion in its detailed terms: its real content is only the exhibition which the Notion gives itself in the form of external existence Every individual thing is one aspect of the Idea431 A notion or idea of something is often said to express its essence or nature, but according to Hegel it does not merely express but actually is that essence: [I]n point of contents, he wrote, thought is only true in proportion as it sinks itself in the facts.432 These assertions may be puzzling, but would be less so when understood to be elaborations of the thesis of antirealism.

What Hegel meant by Geist need not be different from what we have meant in this book by cognition. Geist can be translated as mind and as spirit. The latter may seem quaint today, though it is the choice of most Hegel scholars. It has the virtue of not appearing to stand for the subject matter of psychology. It also occurs in common phrases like the spirit of our age, which signifies not some component or quality of our period of history, like its beginning or duration, but rather what distinguishes it most deeply from other periods in history and thus constitutes its character, nature, or essence. In the Encyclopedia of Philosophy Hegel held that Spirit is the third stage of the unfolding of
370

the Absolute, the first being the Logical Idea and the second Nature. Our word mind would be too vaporous to help express this, and in any case we usually understand it as standing for what Hegel meant by subjective spirit, which he considered only the first and most primitive stage in the development of spirit.

When Hegel says that Every individual thing is one aspect of the Idea,433 this need mean only, as I suggested earlier, that all individual things are dependent for their nature and existence on the whole of reality. For, as I also suggested, the parts of reality were taken by Hegel to depend on each other and on the whole in the way the parts of an organism depend on each other and on the organism, or in the way the concepts in an ideal theory depend for their content on each other and on the theory.

Hegel did call his system absolute idealism, but it would be superficial to interpret him as holding that everything is mental. His idealism bore little resemblance to Berkeleys. It was closer to the antirealism sketched in this book. [I]n thinking, Hegel wrote, the object does not present itself in picture-thoughts but in Notionsconsciousness being immediately aware that this [the object] is not anything distinct from itself. What is pictured or figuratively conceived .has, as such, the form of being something other than consciousness [but in the case of a Notion] consciousness remains immediately aware of its unity with this determinate and distinct being.434 In the case of mental images, we can distinguish between the object imagined and the imagining of it. No similar distinction can be made in the case of Notions. If antirealism did not allow for Kantian things-inthemselves, as Hegel did not, then the antirealist thesis that the world depends on our cognition of it need not be different from Hegels thesis that they are the same. As we
371

shall see shortly, they are the same in the sense in which a headache, which should not be confused with its physical cause in or outside the brain, and the feeling of that headache, even if ultimately reducible to that cause, are commonly believed to be the same. Berkeley thought that what is true of unfelt pains is true also of unperceived trees. If he was wrong, the reason was not his belief that trees are independent of cognition but his refgusal to acknowledge modes of cognition other than perception that provide access to trees but not to pains.

The difference between Berkeley and Hegel is commonly said to be that while Berkeleys idealism was subjective, Hegels was objective. These characterizations are not wrong, but they do require deeper explanation and major qualifications. We can say that according to the subjective idealist the world depends on me, while according to the objective idealist it depends on us. But this might not allow us to describe Berkeleys idealism as subjective, because he insisted that something would exist even if perceived only by God. And it is not clear that it does justice to Hegels idealism. Hegel did not hold that the world depends on human beings. He did insist on the social nature of developed human cognition, and insofar as he also held that the world depends on such cognition his idealism was objective, in the straightforward sense in which, say, physics is objective while personal fancy is subjective. But Hegel described his idealism as absolute, as holding that the Absolute is identical with the Idea. And the Idea is neither subjective, if this would mean that it belongs in subjective spirit, nor objective, if this would means that it is belongs in objective spirit. It belongs nowhere.

372

Chapter Thirteen: MIND AND THE WOR 1. The Ghost in the Machine. The antirealist thesis seems paradoxical, absurd, because it seems to say that the whole world from the page you are reading now to the remotest known galaxies, and since the Big Bang to the farthest conceivable future depends for its existence and nature on the minds, cognitive capacities, of humans, a species in one of its planets fauna. This would be a zoological thesis, and as such has no place in philosophy. Of course, it would also be absurd zoology, as well as absurd physics and astronomy. But antirealism is not a theory in zoology, nor in physics or astronomy. As we saw in the preceding two chapters, the mind relevant to the philosophical topic of the dependence or independence of the world on the mind is not a thing in the world, human or nonhuman, nor is it a part or property of such a thing. This is why the antirealist does not say that the world is dependent on the human mind. There is no human mind on which the world might depend. There is only the world. It is cognized or at least cognizable perceivable, conceivable, describable but there is no other world that is relevant, whether in philosophy or in everyday life, even though there might be a world that is neither cognized nor cognizable. Of course, there are humans in the world, just as there also are whales and chimpanzees. But there is nothing in or about humans, just as there is nothing in or about whales or chimpanzees, on which the world might depend.

To describe that world as cognized or at least cognizable is not to refer to some persons mind and thus to imply its dependence on that person. The idea of mind usually employed in careless talk about mind-dependence is deeply defective. It rests on a powerful but misleading picture of the mind as something in us, a self or a soul, often
373

fancied as located somewhere or somehow behind the eyes, perhaps even felt there. To be sure, there is a brain behind the eyes, but it would be both poor neurology and poor philosophy to identify it with the mind that is at issue in the realim/antirealism debate. Years ago, attacking Descartess dualism, Gilbert Ryle aptly called that idea of mind the dogma of the ghost in the machine. But the word dogma suggests that the idea was a philosophical invention, which it was not. We do point to our heads when speaking of our thoughts, and the belief that they are located within the skull is ancient, not acquired from philosophy or science. We also point to our hearts, however, when speaking of our feelings, especially love, and the belief that feelings are located somewhere within the chest is also ancient. But few if any educated people today share it. At any rate, as we have seen, even if thoughts were in the head, pointing to our heads in philosophical contexts like Cartesian doubt and the realism/antirealism issue would be no more admissible than pointing to our toes. Thought is no more in the head when considered in those contexts than love is in the heart when considered in biology. To say this is not to question anything that neuroscience says. It is to acknowledge that, without flagrant question-begging, we cannot take the mind relevant in such philosophical contexts to be a part or feature, mental or physical, of a thing, human or nonhuman, in the world. Neuroscience says nothing about this question.

Hume denied that there is a philosophical self, a metaphysical subject, in or outside the world. If he was right, it would follow that there is nothing with which the world might be contrasted, and on which it might depend. It would also follow that there is no distinction relevant to the realism/antirealism issue between me and the world. Antirealism would become what in the Tractatus Wittgenstein called pure realism.
374

Hegel denied that individual, personal cognition, if at all advanced, is independent of societal, public cognition. If he was right, and if there is no distinction relevant to the realism/antirealism issue between me and the world, there is also no distinction relevant to the realism/antirealism issue between us and the world. If by we or us is meant societal, public, cognition, and by world its object, the distinction between us and the world is rather like the distinction between feeling a headache and the headache felt. We attach little sense to talk about unfelt headaches or to talk about feelings that are feelings of nothing. If we bypass our obsession with the mind-body problem, we can say that headaches are nothing but feelings and that feelings of a headache are nothing but headaches. Yet we have little inclination to reject the distinction and to speak only of headaches. Headaches are the same as feelings of headache, but in the sense in which the Evening Star is the same as th Morning Star, not in the sense in which the Evening Star is the same as the Evening Star. Hegel explained the two senses by saying that identity is always in unity with difference, in contrast with the abstract idenity asserted in silly statements like mind is mind or planet is planet. Frege explained them by distinhguishing between the sense and the reference of an expression, holding that senses belong in a third ralm of reality, neither mental nor physical. I have offered an explanation by distinguishing between formal identity and material identity, arguing in detail that the difference between them is metaphysical, not just semantical or grammatical, as Russell held in his theory of definite descriptions. It is due to the difference between objects and entities, an object being whatever happens t be perceived or thought and an entity an object that survives the application to it of the concept of existence.435
375

The relation between a headache and feeling that headache is not just analogous to the relation between the world and cognition. It is a familiar and reasonably uncontroversial example, instance, of that relation. Another but very different, unfamiliar but perhaps uncontroversial, instance might be the relation between an esoteric elementary particle in physics and the theory that is the only reason for accepting its reality. The instances that antirealists usually consider, such as the relation between material things or causality and the cognition of them, are seldom familiar and never uncontroversial. So are also the instances from logical cognition, especially generic statements, to which we have devoted much of this book. To be sure, few think that logical cognition has objects, but few also see that if it did not there would not be a world, just as if there were no feelings of headache there would be no headaches. But, as we have seen, logical antirealism entails at least the cosmological version of metaphysical antirealism. And what metaphysical antirealism says about the relation between the world and cognition is not fundamentally different from what common sense says about the relation between a headache and the feeling of that headache. Hegel called his system absolute idealism> He would have gladly agreed that it is pure idealism or pure realism, had he confronted Wittgensteins reasoning that led to the latter.

In the Tractatus Wittgenstein held that the world is my world because, with respect to the content of any cognition, primitive or advanced, perceptual or conceptual, no philosophically relevant distinction between me and the world can be made. Hegel in effect held that the world is our world because, with respect to the content of any advanced cognition, no philosophically relevant distinction also can be made between me
376

and us, and thus between us and the world. In his later works Wittgenstein denied that a private language is possible, and held that a public language presupposes agreement in judgment. He thus, in effect, agreed with Hegel that individual cognition is dependent on public cognition, at least when advanced enough to require the use of language.

Berkeleys lasting contribution was to point out that the world is dependent on the mind insofar as the world is perceived. Kants central contribution was to point out that it is dependent on the mind also insofar as it is understood, conceptualized. Contemporary antirealism has added that it is dependent on the mind insofar as it is described. Berkeley could not consistently mean by perceiving a relation that a human animal bears to an object, and of course he did not. Kant could not consistently mean by concepts inhabitants of human skulls, and he did not: he regarded concepts as what transforms empirical intuitions into objects and thus is responsible for the objective character (essence, nature) of the things to which they apply; in a reasonably clear sense, Kantian concepts, even though imposed on things by us, can be said to be in those things. And contemporary antirealists, who appeal instead to meanings or uses of words, cannot consistently appeal to human activities such as speaking and writing. Gilbert Ryle was the first to see this when he distinguished between the usage and the use of a word, making clear that that philosophers had no business appealing to usage.

Ryle wrote, Hume's question was not about the word 'cause'; it was about the use of 'cause'. It was just as much about the use of 'Ursache'. For the use of 'cause' is the same as the use of 'Ursache', though 'cause' is not the same word as 'Ursache'. Hume's question was not a question about a bit of the German language. The job done with the English
377

word 'cause' is not an English job, or a continental job.436 Ryle went on to castigate the confusion between a 'use', i.e., a way of operating with something, and a 'usage'. A usage is a custom, practice, fashion or vogue. The method of discovering linguistic usages are the methods of philologists.437 In a symposium with J. N. Findlay on the topic, he remarked, The famous saying: Don't ask for the meaning; ask for the use, might have been and I hope was a piece of advice to philosophers, and not to lexicographers or translators.

Findlay concurred: the notion of use, as it ordinarily exists and is used, presupposes the notion of meaning (in its central and paradigmatic sense), and it cannot therefore be used to elucidate the latter, and much less to replace or to do duty for it . [We] cannot fully say, in a great many cases, how an expression is used, without saying what sort of things it is intended to refer to, or to bring to mind 438 What Ryle meant by use of a word was not significantly different from what philosophers have meant by concept. But Findlays important addition that the use of a word involves the things to which it is applied suggests Hegels view of concepts as sunk in the things to which they apply, as permeating them, as Strawson put it. Hegel wrote, It is not we who frame the concepts [die Begriffe]Rather the concept is the genuine first; and things are what they are through the action of the concept, immanent in them, and revealing itself in them.439

A nonzoological view of mind demands a nonzoological view of perception and conception, as well as of language. If perceiving consists, not in a minds relation to an object, but in that objects possession of certain properties, if, as we shall shortly find Strawson ssaying, concepts permeate, soak, the things conceived, and if, as we just found
378

Findlay saying, to speak of the use of a word is to speak of the things to which it refers, then the mind itself, insofar as it perceives and employs concepts and words, cannot be just in the world, it must be the world,. This, of course, was Wittgenstein pure realism. It was also Hegels absolute idealism, which he proposed after rejecting Kants transcendental idealism as unacceptably psychological and thus subjective.

Berkeley began with the simplistic notions of a self (mind) confronted with an object (material or mental) and of consciousness as a relation between the two (perceiving, imagining), questioned the independent reality of the object, and arrived at idealism. Kant reduced the self to a unified consciousness. Hegel identified that consciousness with its objects. Wittgenstein rejected the self altogether, thereby removing also the motivation for a consciousness that the self aims at objects, and allowed only for the world as reality. Hegel was philosophically far apart from Berkeley, and Wittgenstein even more so. But it is unclear that on our topic Hegel and Wittgenstein were far apart from each other. It is unclear that absolute idealism and pure realism differ in more than literary and philosophical style or approach, striking though these differences are. Wittgensteins pure realism is a realism that has earned its keep, unlike the pre-Kantian realism (nave realism), which, as Descartes, Berkeley, and Hume saw, lay helpless before the skeptic. Hegels absolute idealism held that reality is thought, but it did not hold that vegetables, animals, and minerals are mental images, much less feelings or sensations. It held only that there is no cashable difference between what they are and how we perceive and think of them. There is a world and it is the only knowable world, but its nature and contents can be known only as what perception reveals and conception allows. Justice is thus done to both realism and to antirealism.
379

If perceiving and understanding are not relations between a subject and the object perceived or understood, be it external to or in the mind, then they can belong only to the category of monadic, nonrelational, properties. They can only be properties of that object, though not properties like its color or shape. Indeed, in the case of perception and its species, e.g., seeing and hearing, and imagination, everyday language does contain adjectives, e.g., visual, auditory, and imaginary, which apply to the object, not to the subject or to a relation between them. That there are such properties in the case of understanding or conceptualizing is held by most metaphysical systems: they are the properties of an object we call its essence. Little effort should be needed to convince one that understanding the French Revolution includes knowing what a revolution is.

I have argued that in certain epistemological and metaphysical contexts I and we refer not to human beings but to cognitive perspectives or standpoints on the world, to worldviews, cognitions of the world. They refer therefore also to the world itself. The relativity of an object to a cognitive standpoint in the case of perception, to how we perceive it, is familiar because in optics, painting, and everyday life we must view and represent the object from an angle. Much the same is true, though less obviously, of the other species of perception. To feel an object by touch one must touch it with a part of ones body, and how it feels depends partly on the characteristics of the body and especially that part of body. Conception, however, is different. The relativity of an object to a cognitive perspective in the case of conception, to how we conceptualize or understand it, requires sophisticated explanation.

380

The most original and extensive such explanation was proposed by Hegel. It was, in effect, the fourth and final of the four steps, listed in Chapter Eleven, that are needed to free antirealism of anthropocentrism. Stated briefly, it says that concepts are universals, universals are properties, are properties constitute the things in the world that exemplify them (there are no Lockean I know not whats or Bergmannian bare particulars). Whether the world is the totality of facts or the totality of things, the world is thus conceptual. But concepts are no more personal than are the properties of things. Anthropocentrism is thus avoided.

In the Science of Logic Hegel wrote that the Notion [concept, Begriff] is to be regarded, not as the subjective understanding, but as the Notion in its own absolute character.440 In the Encyclopedia of the Philosophical Sciences he explained: The notion is generally associated in our minds with abstract generality, and on that account it is often described as a general conception. We speak, accordingly, of the notions of colour, plant, animal, etc. They are supposed to be arrived at by neglecting the particular features which distinguish the different colours, plants, and animals from each other, and by retaining those common to them all. This is the aspect of the notion which is familiar to understanding; and feeling is in the right when it stigmatizes such hollow and empty notions as mere phantoms and shadows. But the universal of the notion is not a mere sum of features common to several things, confronted by a particular which enjoys an existence of its own. It is, on the contrary, self-particularizing or self-specifying. [T]he real universal should not be confused with what is merely held in common.441 Hegel went on to conclude that It is not we who frame the concepts.Rather the concept is the

381

genuine first; and things are what they are through the action of the concept, immanent in them, and revealing itself in them.442

In recent philosophy we find the influence of Hegel in the works of H. H. Price, P. F. Strawson, Wilfrid Sellars, John McDowell, Robert Brandom, and many others. Strawson was especially clear about the intimate connection between the category of concepts, which is initially psychological/epistemological, and the category of universals, which is metaphysical. That connection suggests that the world and cognition are one and the same. Hegels expressed this by saying that the Absolute is Thought the world is not just dependent on mind, it is mind. (Hegels Geist, as we have seen, must not be confused with mind as used in Anglo-American philosophy.) Strawsons language was more restrained, but his position was not much different.

2. Concepts, Properties, and Universals.

The word concept entered the English philosophical vocabulary as translation of the German Begriff in order to replace idea, which in 17th and 18th century British philosophy had explicitly stood for a subjective state, often a mental image. Ideas, so understood, are incapable of serving as objective principles of classification, and classification is the principal task of advanced cognition. Indeed, Locke had stressed the importance of objective classification, but he claimed that its principles are abstract ideas, obtained from particular ideas by leaving out the characteristics that distinguish them from one another. The existence of such ideas was vigorously rejected by Berkeley
382

and Hume, on the reasonable grounds that there are no abstract mental images. Concepts (Begriffe), in contrast, were for Kant and Hegel objective principles of classification. This is also how the word concept is undertstood in current philosophy, even if limited, as it often is, to the meanings or uses of words what a word means or how it is used depends on linguistic fact, not personal whim. By objective we need not mean anything more than not subjective, anI whimsical uses of words are prime examples of what is meant by subjective. I noted earlier that while the usual English translation of Begriff in the case of Kant is concept, in the case of Hegel it is Notion, because of his distinctive metaphysical understanding of the nature of Begriffe.

As a principle of classification, a concept is objective insofar as the classification is objective. A classification, of course, may be subjective, fanciful, and therefore unsuitable for cognition. If it is objective, the concept that is its principle is usually taken to stand for a property common to the things classified, even if that property were to consist merely in their being called by the same name. Common properties are universals in the straightforward and innocuous sense, as the Oxford English Dictionary puts it, of Applicable to or involving the whole of a class or genus, or all the individuals or species comprising it... Opposed to particular. In everyday life we take them to be what adjectives and common nouns stand for. In logic they are the values of predicate variables, say, , individual things being the values of individual variables, say, x. The logical form of an atomic sentence is usually represented as x.

383

Cognition necessarilly involves classification, whether explicitly, as in botany and zoology, or implicitly, as in ordinary subject-predicate statements. If we take the classification to be something that we do, we may say that it consists in applying a concept to what is classified, whether by using a general word (adjective or common noun) or just thinking of it in a certain way. The aim of cognition, of course, would be objective classification this why it is cognition rather than mere fancy. Now, the metaphysical realist holds that, if the classification is objective, the thing classified would have the property for which the concept stands independently of the application of the concept or any other mode of cognition. The realist might not allow that there are common properties, universals. Also compatible with metaphysical realism are the theory that there are no properties at all, as well as the theory that properties are particulars, tropes (see below).

Metaphysical antirealists reject the realists sharp distinction between properties and concepts. The subjective antirealist does so by regarding concepts as subjective states and also denying that there are objective properties, properties in things, i.e., that classification can be objective, and thus by implication that there is cognition. The objective antirealist takes for granted that there is cognition and that cognition does requires objective classification, but rejects the sharp distinction between properties and concepts by regarding concepts as objective, as themselves being the properties of things. Hegel was an objective antirealist.

The form or character peculiar to thought is the universal, or, in general, the abstract, Hegel wrote.443 And Man is always thinking, even in his perceptions.444 There is no
384

thought without concepts. But also there are no things without universals, because there are no things without properties: Reflection.conducts to the universal of things: which universal is itself one of the constituent factors of a notion.445 This is why the Absolute, reality, is identical with Thought. Insofar as thought must be expressible by I think, the pronoun I may be taken to stand for Thought: The thinking power, the I, is therefore infinite, because, when it thinks, it is in relation to an object which is itself [Ich, das Denken, ist demnach unendlich, darum, weil es sich im Denken zu einem Gegenstand verhlt, der es selbst ist].446

Hegels rejection of the distinction between concepts (Notions, Begriffe) and properties (universals) was the essential tenet of his version of antirealism that led him to the conclusion that the Absolute is Thought. To say that Reason or understanding is in the world, is equivalent in its import to the phrase Objective Thought.447 Hegel was aware, of course, that To speak of thought or objective thought as the heart and soul of the world, may seem to be ascribing consciousness to the things of nature, noting that we feel a certain repugnance against making thought the inward function of things.448 This repugnance has not been unfamiliar in the history of antirealism. But Hegel considered it unjustified.

We shall see shortly that H.H. Price and P. F. Strawson held a view of concepts and universals not unlike Hegels. Put briefly and perhaps crudely, Hegel held that that (1) thoughts are concepts, (2) concepts are universals, (3) universals are properties of things, (4) there is nothing else in things, therefore (5) there is nothing in reality but thought. Price and Strawson would have disagreed only with (5).
385

But first we must attend to some details of the traditional debate about universals and particulars. The general idea of a universal is that of a property, which is or can be exemplified by two or more individual things, as in the state of affairs that both a is F and b is F. So understood, there is no problem of universals, for there is no temptation to think that there cannot be such properties, that, e.g., a and b cannot both be F, any more than there is for thinking that two persons cannot have the same father. The problem arises when we suppose thet the common property is in the things that have it as a constituent of them, that, e.g., F is a constituent of both a and b, even a and b have different locations in space. For, one asks, how can F if be in two places at the same time? The question is motivated by the puzzle, Where else could F be? and by a resolute rejection of the Platonis view that F is not in a and b but rather in a special, separate, realm of being, the world of Forms. But there is little to recommend this reasoning. If I am 6 feet tall, it is nonsense to suppose that being 6 feet tall is in me, much less that it is a constituent of me. And the puzzle was felt mostly under the influence of the British empiricists, who held that all things must be actual or possible objects of perception and therefore found themselves unable to imagine an individual object as anything other than its perceived qualities, their collection. But the idea that a property must be perceivable in the way its instances are is like the idea that the 20th century must be perceivable (seen? touched? smelled?) and the idea that a property must be somewhere is like the idea that the 20th century must be somewhere (in Europe? Asia? Africa?). Both confuse conception with perception, and properties with things that have properties. On these matters, Plato and Frege were closer to the truth. The color of an apple is not in the apple in the way a worm might be. To think otherwise would be what

386

Gilbert Ryle long ago called a category mistake: like asking to be shown the university after having been shown all its buildings.

Hegels, essentially equivalent, assertions that the Absolute is Thought, that reality is spirit, and that thought is sunk in things, are often misunderstood because of failing to grasp his position on universals on which they rest. The position was radically different from Lockes, Berkeleys, or Humes. It was closer to Aristotles, but also shared an important element with Platos. When Hegel says that the Absolute develops from Idea to Nature, this need mean only that universals must be exemplified, that although, as Plato held, they are at least partly responsible for the reality of the things that exemplify them, their own reality, as Aristotle held, consists in being exemplified. Nevertheless, Hegelagreed with Plato that nature cannot adhere to exhibit the strictness of the Notion. A conceptual scheme, theory, or account has a clarity and precision that are inevitably lacking in the wild multiplicity and endless diversity of its object.449 The world of becoming was dependent on the world of ideas, according to Plato, but only as a messy imitation of it.

It is convenient here to use the terminology H.H. Price proposed in his classic work Thinking and Experience. Universalia post rem the British empiricists position are subjective states, ideas, perhaps images in the mind, possibly just words. Universalia ante rem are universals as understood by Plato, entities that exist independently of both subjective states and spatiotemporal objects. Universalia in rebus are, roughly, universals as they were understood by Aristotle, entities that are in things as their properties. Berkeley and Hume denied, of course, that there are universalia ante rem. But they
387

seemed to deny also that there are universalia in rebus, common properties, presumably on the grounds that the properties of particulars are themselves particulars. This position, known as trope theory, was developed and defended two centuries later by D. C. Williams.450

The central question, according to trope theorists, is whether, e.g., the color of this page is identical with the color of the next page, or only resembles it, whether exactly or inexactly. Thus the appeal to tropes is an attempt to avoid commitment to universals by appealing to the presence instead of a relation of resemblance. But there is no fact of the matter here. If there were, we should be able to tell, in the case of the two pages, which is the right view by just looking at them after all, we see them and we also see their color. We would expect to see whether the relation between the color of this page and the color of that page is identity, i.e., that a certain color is a property common to both, or resemblance, i.e., that they are particular properties related by resemblance. But how would these alleged relations of identity and resemblance differ? The truth is that neither the philosophers who say that properties are universals nor the philosophers who say that there they are particulars, tropes, are reporting anything to be settled by experience or inference from experience, that neither are really reporting a discovery of some facts about the world. They can claim only to be offering a more illuminating description of a familiar and uncontroversial fact about the properties of individual things. The problem of universals is whether this fact is more like a paradigmatic informative identity, say, that of the Morning Star and the Evening Star, or like an ordinary relation, say, the spatial relation between the two pages. Both science and common sense hold that it is the former. Both would choose the conjunction of the subject-predicate statements a is F
388

and b is F to describe the color of the two pages, rather than a relational statements of the form aRb. In fact, there are no relevant statements of the latter sort because there are no names of particular properties, of tropes. In aRb, where R stands for resemblance, a and b could not be names, respectively, of the color of this page and the color of the next page since there are no such names. There are the definite descripions the color of this page and the color of the next page, but The color of this page resembles the color of the next page is not a statement of the form aRb.

Universalia in rebus are what in current philosophy are just called properties. They are therefore taken to be present in everything that has properties. Hegel rejected Platos universalia ante rem, and had no patience with the empiricists universalia post rem. At first glance, therefore, he seemed to accept Aristotles universalia in rebus. Indeed, we find him writing, Animal, qua animal, does not exist: it is merely the universal nature of the individual animalsBut to be an animal is the property of the particular animal, and constitutes its definite essence. Take away from the dog its animality, and it becomes impossible to say what it is.451

This remark about animal is best understood in its original context of Aristotles Categories. It concerns substance universals, what Aristotle called secondary substances. Such universals are the genera and species of what Aristotle called primary substances, that is, individual things, which are neither present in nor said of (i.e., predicable of) anything. For centuries, animality (the Animal) served as a standard example of such a universal, but so would have been caninity (the Dog). The latter is a species of the former, which is its genus. In Aristotelian terminology, both are said of, but
389

are never present in, individual things. Hegels just quoted remark applies also to property-universals, what Aristotle called accidents, say, round shape and brown color. Their instances, strictly speaking, are not the individual things in which they are present and of which they are said (predicated), but rather certain properties of those things, say, the shape of this ball and the color of that dog, which are present in but not said of those things. A brown dog is brown, but it is not its brown color.

We may perceive a dog as well as its brown color, and thus can be said to perceive also the universals of which they are instances, namely, caninity and the color brown. If we see the dogs brown color, we may be aware also of its genus, the generic property color, though not by seeing it as a separate entity, in addition to seeing the dogs brown color. But we can think of the generic property, whether or not we perceive an instance of it. We can think of it even without employing a suitable mental image, as we do when trying to imagine it. We may be said, then, to have the relevant concept, but this need mean no more than that we are capable of thinking of the universal and recognizing it in its instances. We do not have the concept therefore in the way we might have a mental image we would have the concept, but not the image, even when asleep. This is a sufficient reason for denying that concepts are mental images. Another reason is that we need concepts even to produce and then recognize the images we want.

390

rem as follows: When the universal is made a mere form and co-ordinated with particular, it sinks into a particular itself452 and [T]he opposition [of the universal] to the particular is so rigorously maintained, that it is at the same time also reduced to a particular again.453 Universals, for Hegel, are not denizens of a Platonic realm. They are not separate from the world. They are not entities that individual things merely imitate, as Plato supposedly believed. They are in things, but not as parts of things. They are not spatiotemporal parts of the world but they are properties of its spatiotemporal parts. They are concrete, not abstract, universals.

When Hegel says that universals particularize themselves, that they self-particularize, this need mean no more than that there is nothing in individual things that might connect them to their properties, that there are no Lockean I know not what substrata or Bergmannian bare particulars. When Hegel says that the Absolute develops from the Logical Idea to Nature, this need mean no more than that concepts have no reality apart from individual things. When he says that the Absolute develops from Nature to Spirit, this need mean no more than that individual things have no reality apart from concepts.

3. Conclusion.

391

of classification. This is why sometimes we use the clumsy word conceptualization instead of understanding. Insofar as the principles of classification are objective, they are indistinguishable from the classes of the things classified. Concepts thus are not words or mental images, but rather recognitional capacities. The account of classification Price preferred was the philosophy of universals, essentially Aristotles view of universals as properties of things, universalia in rebus. The properties of a thing are the natural ground of its objective classification. We may go a small distance beyond Price and say that as principles of classification concepts are not clearly distinguishable from the classes of individual things that they yield, and therefore from their corresponding common properties. For the distinction between class and property is murky. It may be worth recalling that Goodman and Quine argued at one time for nominalism regarding both properties and classes, on the grounds that both are abstract entities: We do not believe in abstract entities. No one supposes that abstract entities -- classes, relations, properties, etc. -- exist in space-time; but we mean more than this. We renounce them altogether.454 Later, at least Quine significantly modified this rather extreme stand.

Price was not an antirealist, but he might have agreed that we cannot actually distinguish between what the world really is and how we understand it. It is commonplace to speak of the child's world, the soldier's world, the scientist's world, and so forth. Such descriptions do not refer to different portions of the world. They refer to the world itself, but as understood or conceptualized in a certain particular way: the childs, the adults, or the scientists way. The phrase conceptual scheme is often applied to such

392

understanding. It suggests the possibility of there being many conceptual schemes incompatible with each other. This possibility was explored by Strawson.

According to Strawson, [Y]ou can have no cognitive contact with, hence no knowledge of, Reality which does not involve the forming a belief, making a judgment, deploying concepts.455 This is why there is no arbiter of the propriety of the application of any concept which is external to and independent of our actual conceptual practice. It follows that there can be no genuine contradiction between different conceptual practices. We cannot exit them so that we can see what they are about and then judge which is the right one. They are just different. I noted earlier that, once we have freed ourselves from the dogma that existence is expressed solely by the particular quantifier, even the attributions of existence in singular statements can seen as the outcome of applying a concept, that of existence.

Incompatibility between conceptual schemes, Strawson held, need not be formal inconsistency, the clashes between them need not call for choice between contradictory propositions. For example, he argued, there is no genuine contradiction between science and common sense regarding the reality of colors. The seeming difference between what they say is due to the different concepts they employ. And we cannot separate the concepts from their subject matter and judge independently which are faithful to it.456 For the subject matter is permeated by the concepts. To question this would be an invitation to step outside the conceptual scheme which we actually have and then to justify it from some extraneous point of vantage. But there is nowhere to step; there is no such extraneous point of vantage.457 There is no metaphysically absolute standpoint
393

from which we can judge between the two standpoints.458 The picture of a concept-free access to facts, to reality, he wrote, is confused and ultimately self-contradictory.459 Speaking of Wittgensteins remarks on aspect-seeing in his example of a picture seen equally legitimately as a duck or a rabbit, Strawson wrote, the visual experience is irradiated by, or infused with, the concept; or it becomes soaked with the concept.460 Hegel would have agreed: thought is only true in proportion as it sinks itself in the facts.461

Strawson was the author of an important book on Kant, but like Fichte, Hegel, and Goodman he rejected Kants things-in-themselves. The rejection was neither necessary nor was the interpretation of Kant supporting it adequate. As I have repeatedly observed, a view like Kants serves as the background against which we grasp that we can have no cognition of the world except as our world even though we have not created that world. Goodmans initially incredible claim that we make the world and Hegels that Reality is Thought can be seen as exaggerations of what actually follows from two tautologies, perhaps two versions of the same tautology, and a proposition that is not a tautology but is self-evident.

The first tautology is that we are the perceivers, conceivers, and describers of the world we perceive, understand, and describe. The second tautology is that the only world or reality we perceive, understand, and describe is the world perceived, understood, and described by us. The self-evident proposition is that, in philosophical contexts like Cartesian doubt and the realism/antirealism issue, we cannot coherently regard ourselves as a part, mental (an ego, a colony of egos) or
394

material (a brain, a collection of brains), of that perceived, understood, and described world. Together with the two tautologies, this proposition allows us to say that metaphysics can be both antirealist and dehumanized.

This conclusion is compatible with Kants transcendental idealism. It does not imply that there are no Kantian things-in-themselves, a reality we do not, perhaps cannot, perceive, understand, and describe. And by distinguishing between perception and conception I have in effect appealed to Kants distinction between intuitions and concepts. But I have not appealed to his distinction between empirical and pure concepts, nor have I appealed to his, or Hegels, distinction, explained earlier, between concepts and ideas. Whether these further distinctions are needed is not among the concerns of this book. Its central concern is anthropocentrism in philosophy, and in this Part it is the way antirealism can be freed of anthropocentrism. The two tautologies and the self-evident proposition show that way.

395

us from what they consider the illusions produced by incompatible but powerful and misleading analogies or pictures. In chapter 7 I remarked that metaphysicians do not discover entities hidden from the rest of us, that they do they do not have the sort of training and instruments needed for making such discoveries. What they can do is to draw attention to, and emphasize, similarities and differences between fundamental kinds of items in the world that in everyday life, and even in science are unnoticed, precisely because they are fundamental. This book is intended, accordingly, not as proposal of a substantive view or report of a discovery but as a proposal of a way of understanding the relationship between us and the world. In discussing Wittgensteins semirealism, based on his distinction between saying and showing, I avoided judgments about its truth, stressing instead its importance as an alternative to the stark positions of metaphysical realism and antirealism. Much the same can be said about my attitude toward Moores view of goodness as an indefinable, simple, nonnatural property, and my defense of epistemology as logic as an alternative to naturalistic and subjective epistemology. It would be not only simplistic but misguided to have asked which is the true metaphysics, ethics, or epistemology.

The goal of philosophy is to achieve understanding of its subject matter, not to find further information about it. And understanding involves seeing similarities and differences, employing analogies, relying on metaphors, providing illuminating pictures, rejecting misleading pictures.. It would be a misunderstanding to think that saying what some thing is like is at most second best, and that we should aim at saying what it is. To say what something is itself is, in effect, to say what it is like, namely, that it is like the paradigm for the application of the relevant predicate. Ordinarily the likeness in question
396

is so close that its presence and our reliance on it may escape our attention. To say that a certain animal is a dog is to say, at least by implication, that it is like the paradigms for the application of dog. But we are seldom conscious of the implication because the likeness is ordinarily so close that the animal before us can itself be taken to be such a paradigm, even though we did not learn the conventions regarding the use of dog by reference to it.

Appeal to a self-evident proposition is often decisive: Well, doctor, say what you will, but it still hurts silences the all-knowing physician. Reductio ad absurdum, which in effect appeals to the principle of noncontradiction, is often the only available form of argument in mathematics. Appeal to a self-evident proposition is also sometimes sufficient in philosophy: We cannot regard ourselves as inhabitants of the world when considering, asserting, denying, or doubting the reality of that world, I hope, is an example.

Appeal to a tautology is often the best way to to break the sway of a misleading picture or analogy and to dissolve an illusion. You cant both spend and save it exposes an all too frequent illusion in financial planning. When made in philosophy, it can be conducive to understanding by exposing a misleading philosophical picture and thus dispelling the illusion that the picture generates. Especially relevant to this book is the tautology that the only world or reality we perceive, understand, and describe is the world perceived, understood, and described by us. It can dispel, I hope, the illusion generated by the philosophical picture of ourselves as ghosts in machines.

397

The metaphysical picture of the world we have drawn shares with Hegels the absence of the usual subject-object dualism. Is it then idealist? If it is, the idealism certainly is not Berkeleys. Is it Hegels idealism? It depicts no Absolute that develops from Logical Idea to Nature and from Nature to Spirit, nor does it employ any dialectical method. Is our metaphysical picture then Sartres? They have in common the explicit rejection not only of subject-object dualism but also of the act-object dualism that considers consciousness as an entity. But our metaphysical picture bypasses Sartres existentialist concerns, e.g., with freedom, bad faith, sadism and masochism, which surely are irremediably anthropocentric. Is our metaphysical picture Wittgensteins, in which subject-object and act-object dualisms are also absent? If Wittgenstein had a metaphysical picture at all, it was so sketchy that the question is unanswerable. Had he had one, what our picture and his would share is the rejection of the philosophical self and thus of subjectivism in all its forms. In this respect both would be descendants of Hegels metaphysical picture. Wittgensteins assertion that The self of solipsism shrinks to a point without extension, and there remains the reality co-ordinated with it462 and Hegels assertion that the thoughts, far from being merely ours, must at the same time be the real essence of the things463 need not convey different metaphysical pictures, despite the great differences in terminology, style, and historical context.

398

consciousness it merely does not include it, much as a group portrait of a family might not include the photographer.

Our metaphysical picture is also compatible with that of a phenomenology that remains faithful to Husserls motto We must go back to the things themselves. Consciousness is not in our picture because it is not among the things themselves, an additional item in the world competing with them for a place in reality. It has no inhabitants or qualitative nature. There are only its objects. In itself it is nothing. Nonetheless, consciousness is the revelation of things and the regulator of which are admissible into the world. It is a revelation on the level of perception (sensibility, outer or inner; intuition, sensory of intellectual). It is a regulator on the level of conception (understanding, thought, reason, language). Perception opens the entrance to the world, but conception is the gate-keeper. So understood, consciousness is what we have called cognition, sometimes even knowledge, the outcome of the successful employment of cognition. We can responsibly regard as objects in the world only those revealed and then allowed by consciousness. To this extent our metaphysical picture is compatible also with that of pure idealism: the world as mind. It does justice to what makes idealism plausible. But though our picture contains only what consciousness reveals and accepts, it does not exclude Kantian thingsin themselves. It merely does not include them, much as the group portrait of a family might not include the grandparents.

399

which exist464 Fichte thought the moral was that we must rely on faith, rather than knowledge. The views of consciousness we find in Hegel, Wittgenstein, and Sartre, however, suggest that we should rather say: All that I know is the world itself. The things in it are: - they are the only things which exist. But we might be wiser if we refused to choose between the metaphysical picture Wittgensteins pure realism conveys and that of Fichtes pure idealism, and welcome both. Our philosophical understanding would then be richer. Of course, the welcome need not be unreserved. Pure realism does not entail physicalism, and therefore it is free of the latters rashness and implausibility. Pure idealism does not entail doctrines about an absolute or a dialectical development, and it need not be expressed in the jargon of German Idealism or of Husserlian phenomenology. As to which is preferable, suffice it to note the advantage of the close proximity of pure realism to both common sense and science. Pure idealism and Husserlian phenomenology lack that advantage.

A metaphysical picture, of course, is not a metaphysical system. Our project has been to free philosophy first epistemology and ethics, then metaphysics of anthropocentrism, not to construct ethical, epistemological, and metaphysical systems. But even if we distrust the claims to philosophical knowledge that such systems make, we have left room for them. Our metaphysical picture allows for the investigation of numerous ordinary, traditional, philosophical topics, to which, as I pointed out in the Introduction, the realism/antirealism debate is irrelevant because they are affected neither by the presence nor by the absence of anthropocentrism. The dehumanization of philosophy proposed in this book is compatible with any standard theory about such a topic.

400

For example, much of the history of metaphysics has consisted of arguments for the existence of Platonic Forms and of God. The arguments are not as bad as many take them to be. The usual objections to Platos Forms are cases of ignoratio elenchi: that the Forms play no role as efficient causes and thus are irrelevant to our knowledge of the world, or that we are not acquainted with universals that are not exemplified. If the Forms were proposed by Plato as causes, obviously they would have been formal, not efficient causes, in Aristotles terminology. And if by acquaintance is meant sense-perception, the Forms were explicitly proposed as examples of entities that cannot be perceived through the senses. Much of the history of metaphysics has also consisted of arguments for the existence of God. The objections to these arguments have been less than compelling. The standard complaint about the ontological argument rests on an assumption less plausible than any premise of the argument, namely, that the concept of existence does not stand for a property (perfection) but is properly expressed by the particular (existential) quantifier. The cosmological argument from the need for a first cause has unimpeachable motivation that even scientists share. The cosmological argument from design, however, is on shakier grounds because of the anthropocentrism latent in its use of the concept of design.

Our picture leaves room for much else in the history of metaphysics that is less unfashionable today than Platonic Forms and the existence of God. A prime example is the project of ontology: the listing and description of the most general kinds (categories) of entities and the relations among those kinds. Other examples are the questions whether the properties of individual things are universals or particulars (tropes), and whether individual things are merely bundles of their properties.
401

There is also the novel metaphysical topic that emerged from our rejection not only of the metaphysical self but also of consciousness as ordinarily understood. I suggested that the familiar distinctions among kinds of consciousness, such as vision, audition, memory, imagination, and conception, should be understood as distinctions among monadic properties of the putative objects of the consciousness, perhaps already expressed by the adjectives visual, auditory, and imaginary. Related to this topic is the question of how the thesis of antirealism should be understood in the case of cognition that is not advanced, e.g., that of neonates and adults who never acquired a language.

There is also the question whether what the metaphysical pictures drawn earlier, realist and idealist, do not include nonetheless ought to be in some fashion included: consciousness in the case of the realist picture, Kantian things-in-themselves in the case of the idealist picture. No science can answer such a question. If it can be answered at all, only metaphysics can do it. If it cannot be answered, only metaphysics can explain why it cannot.

402

403

NOTES

In Paul Arthur Schilpp, The Philosophy of Bertrand Russell (Open Court: 1989), 19-20.

Benedict de Spinoza, Ethics (New York: Hafner; 1963) Part One, Proposition VII, Note.

Ethics, Part Five, Proposition XVII.

4

Hegel, Faith and Knowledge, trans. Walter Cerf and H.S. Harris (Albany: State University of new York Press, 1977), 89.
5

Anthropocentrism may be considered discrimination in favor of humans, which is why defenders of animal rights call it speciesism. Its narrowest form is egoism, discrimination in favor of just oneself, what Kant called the dear self, commonly considered a paradigm of immorality. It has logical descendants, which can be ordered according to their distance from it. Nepotism discrimination in favor of ones family seems to come first. Next, perhaps, is discrimination in favor of ones friends, part of Polemarchuss definition of justice in Platos Republic. Then might come chauvinism, discrimination in favor of ones country. Even further from egoism is racism: discrimination in favor of ones race. Sexism, discrimination in favor of ones gender, also belongs among the logical descendants of egoism, but as a position it is logically unstable. It conflicts with other forms of discrimination that the sexist is likely to accept. Some members of ones family and some of ones friends are likely to belong to the opposite sex, and so do roughly half of ones compatriots and of the members of ones race. Classism (Thrasymachuss advantage of the stronger, Marxs dictatorship of the proletariat, the early Bostons better people) would also be a logically unstable position, for similar reasons, though it has been part of important theories, namely, Platonism and Marxism, which

rest on independent grounds.

6

Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA: MIT Press, 1991), 35

Bertrand Russell, Religion and Science (Oxford: Oxford University Press, 1997), 24.

Nelson Goodman, Ways of Worldmaking (Indianapolis: Hackett, 1978), 6. For a detailed and incisive discussion of common sense, see Noah lemos, Common Sense

(Cambridge: Cambridge University Press, 2004).

10

Immanuel Kant, Critique of Pure Reason, trans. Paul Guyer and Allen W. Wood (Cambridge:

Cambridge University Press, 1998), B37-4

11

E.g., at Critique of Pure Reason, B xxvii-xxviii.

12

Critique of Pure Reason, B 42-44. For the general distinction between transcendental an empirical idealism, see A 369-70. Heidegger remarked that Kants transcendental conception was possible only on the basis of the subjectivity of mans essence. (On the Essence of Truth, in Basic Writings, 118.)

13

Ibid. B 69. Ludwig Wittgenstein, Tractatus Logico-Philosophicus, trans. D. F. Pears and B. F. McGuinness

14

(London: Routledge, 1972), 5.63. In references to the text, I shall use Wittgensteins numerical designations of sentences. All italics, upper-case letters, and parentheses in the quotations will be

Wittgensteins.

15

Martin Heidegger, Being and Time (New York: HarperOne; Revised edition, 1962), 27.

16

Martin Heidegger, On the Essence of Truth, section 9, Note, in D. F. Krell, Martin Heidegger:

Basic Writings (New York: Harper & Row, 1977), 138.

17

Jean-Paul Sartre, Being and Nothingness, trans. Hazel E. Barnes (New York: Philosophical Library,

1956)

18

See, for example, J. L. Austin, How to Do Things with Words (New York: Oxford University

Press,1976)
19

Tractatus, 5.6.
20

Martin Heidegger, Letter on Humanism, in Basic Writings, 193.

21

Martin Heidegger, The Origin of the Work of Art, in Poetry, Language, Thought (New York:

Harper&Row, 1971), 74, trans. Albert Hofstadter.

22

Bertrand Russell, A History of Western Philosophy (New York: Simon and Schuster, 1972), 707-708

23

Steven Pinker, The Stuff of Thought (New York: Viking, 2007),

24

Hilary Putnam, Ethics without Ontology (Cambridge: Harvard University Press, 2004), 21-22.

25

Ways of Worldmaking, especially chapters I and V. Hilary Putnam and James Conant, Words and Life (Cambridge: Cambridge University Press, 1995),

26

308.
27

The phrase state of affairs is often used as a substitute for the word fact. A distinction between

actual and possible states of affairs is also often made, true statements supposed to correspond to the former and false statements to the latter. Fact then, if used at all, would be applied only to actual states of affairs. For the sake of terminological simplicity, I will use only fact and ignore the question whether false statements correspond to anything.
28

Tractatus, 5.61. Tractatus, 4.441.

29

30

Tractatus, 4.0312

31

See my Metaphysical Realism and Logical Antirealism, in Richard Gale, ed., Guide to

Metaphysics (Blackwell, 2002).

32

Gustav Bergmann, New Foundations of Ontology (Madison, University of Wisconsin Press, 1992,

edited by William Heald), 173. Page references in the text will use the abbreviation NF. Healds introduction is obligatory reading for all interested in Bergmanns philosophy. He has also included an invaluable glossary.
33

Tractatus, 4.1272.
34

In Thinking and Experience (London: Hutchinson's University Library, 1953) H. H. Price described concepts as recognitional capacities. See also my Skepticism about the External World (New York:

Oxford University Press, 1998) and Being Qua Being (Bloomington: Indiana University Press, 1979).

35

Semirealism should not be confused with what has been called quasi-realism regarding ethical

statements . See Simon Blackburn, Essays in Quasi-Realism (Oxford: Oxford University Press, 1993).
36

MS 302, Diktat fr Schlick 1931-33. Quoted by David Stern, Sociology of Science, Rule

Following and Forms of Life, in M. Heidelberger and F. Stadler, eds., History of Philosophy and Science, 347.

37

Roderick M. Chisholm, A Realistic Theory of Categories (Oxford: Cambridge University Press,

1996), 99-105. Chisholm speculated that the mind might be microscopic part of the brain.

38

The distinction between a weak and a strong sense of know was made by Norman Malcolm in

Knowledge and Belief, included in Knowledge and Certainty (Englewood Cliffs, N. J.: Prentice-Hall, 1963), 58-7

39

Compare G. E. Moores explanation, mentioned in chapter 2, of what he meant by calling the

fundamental propositions of ethics self-evident. See Principia Ethica (Cambridge: Cambridge University Press, 1903), Preface.

40

J. L. Austin, Sense and Sensibilia (Oxford: Clarendon Press, 1962), p. 115. Original italics.

41

Bertrand Russell used the suggestive phrase egocentric particulars, instead of indexicals. He

wrote: One of the aims of both science and common sense is to replace the shifting subjectivity of egocentric particulars by neutral public terms (Human Knowledge (New York: Simon & Schuster, 1962, 85).

42

Following certain remarks by Wittgenstein in The Blue and the Brown Book (Oxford: Oxford

University Press, 1958, 66-67), G. E. M. Anscombe has argued that I is not a referring expression. In using it, she says, getting hold of the wrong object is excluded, and that makes us think that getting hold of the right object is guaranteed. But the reason is that there is no getting hold of an object at all (The First Person, in Samuel Guttenplan, ed., Mind and Language, Oxford: Oxford University Press, 1975, 45-65). Her point is not Humes. It has nothing to do with what she could or could not find.
43

See, for example, Hegels Logic, Part Five. Bertrand Russell, The Philosophy of Logical Atomism (Chicago and La Salle: Open Court, 1985), 37.

44

45

David Kaplan has distinguished between pure indexicals, such as I, which involve no genuine

demonstration and do not admit attachment of a noun, and demonstratives such as this, which do, e.g., in this man. See his Demonstratives, in J. Almog et al, eds, Themes from Kaplan, Oxford: Oxford University Press, 1977), 565-614.

46

Gottlob Frege, Thought, in M. Beaney, M., The Frege Reader (Oxford: Blackwell), 325-326.

47

Beaney, 36

48

F. Strawson, On Referring (Mind, 1950) and Introduction to Logical Theory (London: Methuen,

1952), 175-79. In the former, more influential work, Strawson did not use the term presupposition, and wrote instead of some sense of imply that is not equivalent to entails or logically implies.
49

Does he care about it? He neither cares nor doesnt care; he is dead (Introduction to Logical Theory, 18).

50

Introduction to Logical Theory, 175.

51

Principia Mathematica, 1.1. Bertrand Russell, The Problems of Philosophy (Oxford: Oxford University Press, 1953), 71-7
52

G. E. Moore, A Defense of Common Sense, in Philosophical Papers (London: Allem & Unwin, 1959), 32-60. I discuss the argument in Skepticism about the External World.

53

Being and Nothingness, 22 Italics in original.

54

For a succinct statement of Kants view, see Critique of Pure Reason A 598/B 626.

55

Ibid, A 225/B 272 .

56

Ibid., A 493/B 521.

57

I develop this view in detail in Skepticism about the External World.

58

Bertrand Russell, Introduction to Mathematical Philosophy (New York: Simon and Schuster, 1971),

164.
59

Immanuel Kant, Logic, trans. Robert S. Hartman and Wolfgang Schwartz (New York: Dover, 1974), 144.

60

G. E. Moore, Principia Ethica, Revised Edition (Cambridge: University Press, 1993), 36.

61

Principia Ethica, 193.

62

Brian Hutchinson, G.E. Moores Ethical Theory: Resistance and Reconciliation (New York: Cambridge University Press, 2001).

63

Principia Ethica, 198.

64

Principia Ethica, 76-77.

65

Thomas Aquinas, Summa Theologica, Part Two, Question 94, Article

66

Principia Ethica, 197.

67

Critique of Pure Reason, A 797/825 A 801/B 829.

68

Critique of Pure Reason, A 841/42 B 869-70.

69

Critique of Pure Reason, B 18.

70

W. V. O. Quine, Two Dogmas of Empiricism, Philosophical Review 60 (1951): 20-43.

71

Immanuel Kant, Groundwork of the Metaphysics of Morals, trans. H. J. Paton (New York: Harper

and Row, 1964), 57.

72

Principia Ethica, 135-36, 143-47, 236-38, 245-47.

73

Principia Ethica, 133. The propriety of Moores use of beauty as an example here is astutely discussed byRobert Audi in Intrinsic Valuee and Reasons for Action, in Terry Horgan and Mark Timmons, Metaethics after Moore (Oxford: Clarendon Press, 2006), pp. 80-81.
74

Letters to Russell, Keynes and Moore, ed. G.H. von Wright (Oxford: Blackwell, 1974) 9.

75

See the next chapter for more details about Wittgensteins view.

76

Tom Reagan, Bloomsburys Prophet (Philadelphia: Temple University Press, 1986), 144.

77

Bertrand Russell, A Free Mans Worship, in Mysticism and Logic (London: Allen & Unwin, 1917),

50.
78

Bertrand Russell, Scientific Method in Philosophy, in Mysticism and Logic, 109.

79

Ludwig Wittgenstein, Notebooks, 1914-1916, trans. G. E. M. Anscombe (New Yor: Harper & Row,

1961), 74.
80

In Skepticism about the External World I argue for an analogous conclusion in epistemology regarding the existence of bodies, of an external world.

81

Tractatus, 6.13, 6.421. Critique of Pure Reason, A 802/B 830.

82

83

G. E. Moore, Some Main Problems of Philosophy (London: Allen & Unwin, 1953), 277.

84

Ibid., 267.

85

Ibid., 298.

86

Ibid., 268.

87

Principia Ethica, 198,

88

Principia Ethica, 34.

89

G. E. Moore, Ethics (Oxford: Oxford University Pres, 1912), 73.

90

Principia Ethica, 4.

91

Cf. Philip Kitcher, The Evolution of Human Altruism, Journal of Philosophy, XC (1993), 497-516.

92

Skepticism in Ethics (Bloomington and London: Indiana University Press, 1989).

93

David M. Armstrong, Universals and Scientific Realism ( Cambridge: Cambridge University Press,

1978).

94

Evan Fales, Causality and Universals (London: Routledge, 1990).

95

G. E. Moore, Identity, Proceedings of the Aristotelian Society, n.s., v. I, 1900-1901.

96

Causality and Universals, 221-224

97 Principia, 202. 98

Frege had used the phrase logical objects for the objects of arithmetic in the context of his project of reducing arithmetic to logic, a project continued later by Russell and Whitehead. Bertrand Russell used it in his posthumously published Theory of Knowledge: The 1913 Manuscript (London and New York: Routledge, 1984), 97, claiming that we must be acquainted with logical objects in order to understand logical terms such as particulars, universals, relations, dual complexes, predicates. Wittgenstein read Russells manuscript and criticized it severely. Several years later he asserted in Tractatus Logico-Philosophicus, There are no logical objects (4.441).

99

Paul Engelmann, Letters from Ludwig Wittgenstein, 143-144. Cf. David G. Stern, Wittgenstein on

Mind and Language (New York: Oxford University Press, 1995), 8, 70-74.
100

Cf. Skepticism in Ethics (Bloomington: Indiana University Press, 1989).

101

In Alfred North Whitehead and Bertrand Russell, Principia Mathematica to *56 (Cambridge: University Press, 1962), 9.13, an individual is defined as anything that is neither a proposition nor a function.

102

Notebooks, 74.

103

Ludwig Wittgenstein, Culture and Value, edited by G. H. von Wright and translated by Peter Winch (Chicago: University of Chicago Press, 1980), 3e.

104

Ludwig Wittgenstein, Philosophical Occasions 1912-1951 (Indianapolis, IN: Hackett, 1993), 40,

44.

105

Notebooks, 83e. Wittgenstein adds: the work of art is the object seen sub specie aeternitatis.

106

Letters to Russell, Keynes and Moore, 7

107

Metaphysics and Nonsense, 66.

108

Cf. Hilary Putnam, The Many Faces of Realism (LaSalle: Open Court, 1987), 1 Putnam identifies his notion of a lived-in world with Husserls notion of Lebenswelt.
109

Universals and Scientific Realism, 25. Franz Brentano, Psychology from an Empirical Standpoint (London: Routledge, 1973), trans. by

110

A.C. Rancurello, D.B. Terrell, and L. McAlister. 2nd ed., intr. by Peter Simons, 1995.

111

Thought, in Beaney.

112

Jerry Fodor, The Language of Thought (Cambridge: Harvard, 1975).

113

Gustav Bergmann, , Meaning and Existence (Madison, University of Wisconsin Press, 1959). See

also New Foundations of Ontology (Madison: University of Wisconsin Press, 1992), ed. by William Heald.

114

Works, Novum Organum, Aphorisms Concerning the Interpretation of Nature and the Kingdom of

Man, LIX.

115

Leviathan, Part I, Chapter IV.

116

John Locke, An Essay Concerning Human Understanding Vol. 1 (London: Dent, 1960), 126. But in his Treatise of Human Nature, Hume disagreed: no truth appears to me more evident, than that beasts are endowd with thought and reason as well as men.

117

Science of Logic, trans. A. V. Miller (London: George Allen, 1969), 31.

118

Thinking and Experience, 36-38.

119

Noam Chomsky, Rules and Representations, 248. Chomsky appeals to innate structures and speaks of a possible universal grammar of faces.
120

Ludwig Wittgenstein, Philosophical Investigations, trans. G.E..M. Anscombe, (Oxford: Blackwell, 1991), ## 329-330.

121

J. R. Bambrough, Universals and Family Resemblances, Proceedings of the Aristotelian Society,

LXI (1960-61): 207-223.

122

Michael Dummett, Truth and Other Enigmas (Cambridge: Harvard University Press, 1978), 438.
123

Ibid. 458.

124

Donald Davidson, Rational Animals, in Subjective, Intersubjective, Objective (Oxford: Oxford

University Press, 2001), 100.

125

W.V. Quine, From Stimulus to Science (Cambridge, Mass.: Harvard University Press, 1995), 88-

89.

126

From Stimulus to Science, 93.

127

Nelson Goodman and Catherine Z. Elgin, Reconceptions in Philosophy (Indianapolis and London:

Hackett, 1988), 154. But elsewhere, including this same work (see vii, 9, 11, 155, and section VIII, 3), Goodman points out that what he calls symbol systems need not be linguistic, that they could be, e.g., notational or pictorial.

128

Keith S. Donnellan, Belief and the Identity of Reference, Midwest Studies in Philosophy XIV

(1989), 283. Kripkes puzzle was that a Frenchman who had never been out of France might have said, presumably on the basis of authority, Londres est jolie, and continued to believe this, though when later he actually visited London he might have said equally sincerely London is not pretty, if he happened not to know that he was speaking of the same city. See Saul Kripke, A Puzzle about Belief, in Avishai Margalit, ed., Meaning and Use (Dordrecht: Reidel, 1979).

129

Michael Dummett, The Logical Basis of Metaphysics (London: Duckworth, 1991), 214-215.

130

The Language of Thought.

131

Tim van Gelder, What Might Cognition Be, If Not Computation? The Journal of Philosophy,

Volume XII, number 7 (July 1995), 345.

132

van Gelder, 373.

133

Wilhelm von Humboldt, On Language, (Cambridge: Cambridge University Press, 1988), 54-55.
134

Ferdinand de Saussure, Course in General Linguistics (Chicago and LaSalle, Ill.: Open Court, 1986), 111 [155].

135

William James, The Principles of Psychology (New York: Henry Holt, 1890, 46

136

Philosophical Investigations, #34

137

Helen Keller, My Story, ed. John Albert Macy, 234. Mark Twain is reported to have said that

the two most interesting characters of the nineteenth century were Napoleon and Helen Keller (ibid., 225).
138

Kalvis M. Jansons, A personal view of dyslexis and of thought without language, in L. Weiskrantz, Thought without Language, 502-03.
139

J. B. Watson and W. McDougal, The Battle of Behavior (New York: Norton, 1929), 33.
140

Leonard Bloomfield, Language (Chicago: University of Chicago Press, 1933), 28.

141

B.F. Skinner, Verbal behavior (New York: Appleton-Century-Crofts, 1957), 449.

142

Edward Sapir, Language (New York: Harcourt, Brace&World, 1921), 19.

143

Edward Sapir, Language, Culture and Personality (ed. David G. Mandelbaum, Berkeley and Los Angeles: University of California Press, 1963), 16 (Reprinted from The Status of Linguistics as a Science, Language 5, 209.) See also A. H. Bloom, The Linguistic Shaping of Thought: A Study in the Impact of language on Thinking in China and the West; Laurence Erbaum, 1981; Richard B. Brandt,
144

Language, 1949, 8.
145

Language, Culture and Personality, 157.

146

New York Times, October 11, 2010.

147

Benjamin Lee Whorf, Selected Writings, 226.

148

Benjamin Lee Whorf, Language, Thought and Reality: Selected writings of Benjamin Lee Whorf, ed. John B. Carroll, (New York: John Wiley and Sons, 1956), 240. He also added: Our Indian languages show that with a suitable grammar we may have intelligent sentences that cannot be broken into subjects and predicates...When we come to Nootka, the sentence without subject or predicate is the only type...Nootka has no parts of speech; the simplest utterance is a sentence, treating of some event or event-complex. ( 242; reprinted from Language 5 (1929): 207-214). Zellig Harris describes Whorf's view as :occupational imperialism for linguistics," in "Distributed Structure, Fodor and Katz, 38, n. 1

149

Hilary Putnam, Realism and Reason. Philosophical Papers. Volume 3 (Cambridge: Cambridge

University Press, 1983), xvi.

150

For example, Brbel Inhelder and Jean Piaget, The Growth of Logical Thinking (New York:

Basic Books, 1958), trans. Anne Parsons and Stanley Milgram.

151

Wilhelm von Humboldt, On Language, 215.

152

See, for example, his Rules and Representations, chapters 3 and 5, and Language and Nature.

153

In Rules and Representations, he wrote: What many linguists call universal grammar may be regarded as a theory of innate mechanisms, an underlying biological matrix that provides a framework within which the growth of language proceeds. (187). As I remarked earlier, on this issue Chomsky is much subtler than the typical contemporary materialist. For more recent examples, see his Language and Nature, Mind, 104, 1-60; Powers and Prospects (London: Pluto, 1996); Comments: Galen Strawson, Mental Reality, Philosophy and Phenomenological Research, vol. LVIII, no. 2 (June 1998). Indeed, he often uses the phrase "mind/brain," but when questioned about what the slash in it means, he admits that this is a problem and disavows any inclination to accept reductionism or eliminative materialism. See Noam Chomsky, Language and Thought (Wakefield, Rhode Island & London: Moyer Bell, 1993), 79-87; the question was asked by James H. Schwartz, ibid., 71-77.
154

Rules and Representations, 244.

155

Noam Chomsky, Rules and Representation, 165.

156

Nelson Goodman, Problems and Projects (New York: Bobbs-Merrill : 1972), 74.

157

Cf. Rules and Representations, 251.

158

Language and Problems of Knowledge, 171.

159

Ibid., 191.
160

Steven Pinker, The Language Instinct, 57-58.

161

Steven Pinker, The Language Instinct, 124. Karen Wynn's piece appeared in Nature 358 (1992), 749-50.
162

Elizabeth M. Brannon and Herbert S. Terrace, "Ordering of the Numerosities 1 to 9 by Monkeys," Science, Vol. 282 (23 October 1998), 746-749; for a judicious discussion in the same issue, see Susan Carey, "Knowledge of Number: Its Evolution and Ontogeny," ibid., 641163

Thought Without Language, ed. L. Weiskrantz (Oxford: Clarendon Press, 1988), 194. Baillargons works are cited at the end of Leslies.

164

Gabriel Horn, Thought without Language in Birds, in L. Weiskrantz, Thought without

Language (Oxford: Clarendon Press, 1988), 299.

165

David Premak, "Minds with and without Language," in L. Weiskrantz, ed., Thought and Language, 64.

166

The Language Instinct, 77.

167

Albert Casullo, A Priori Justification (New York: Oxford University Press, 2004).

168

Laurence BonJour, In Defense of Pure Reason (Cambridge: Cambridge University Press, 1998), pp.

102, 106, 5.2)

169

For a detailed discussion of the general question of the a priori, see The Concept of Knowledge

(Evanston: Northwestern University Press, 1970).

170

Of Mind and Other Matters, 20.

171

See Concept and Object, Function and Object, Thought, and Negative Thoughts, in Michael

Beaney, ed., The Frege Reader (Oxford: Blackwell, 1997).

172

See An Ontological Inventory, Journal of the British Society of Phenomenology and

Existentialism, 1974. Also Logic and Reality (Madison: University of Wisconsin Press, 1960) and New Foundations of Ontology (Madison: University of Wisconsin Press, 1994), ed. W. Heald.

173

Hilary Putnam, The Dewey Lectures, Journal of Philosophy, 1994, 513.

174

Theory of Knowledge: The 1913 Manuscript, 99.

175

The Philosophy of Logical Atomism, 71-

176

In logic nothing is accidental: if a thing can occur in a state of affairs, the possibility of the state of

affairs must be written into the thing itself. (2.012). ....We have said that some things are arbitrary in the symbols that we use and that some things are not. In logic it is only the latter that express: but that means that logic is not a field in which we express what we wish with the help of signs, but rather one in which the nature of the absolutely necessary signs speaks for itself [zeichen sich aus] (6.124).
177

6.124 The propositions of logicpresuppose that names have meaning and elementary propositions sense; and that is their connexion with the world 3.203 A name means an object. The object is its meaning.

178

4.022 A proposition shows [zeigt sich] its sense.

A proposition shows [zeigt] how things stand if it is true. And it says that they do so stand

179

4.21 The simplest kind of proposition, an elementary proposition, asserts the existence of a state of

affairs.

180

4.25 If an elementary proposition is true, the state of affairs exists: if an elementary proposition is

false, the state of affairs does not exist.

181

Cora Diamond, The Realistic Spirit: Wittgenstein, Philosophy and the Mind (Cambridge, MA: MIT

Press, 1991).
182

Warren Goldfarb, Metaphysics and Nonsense, Journal of Philosophical Research XXII (1997).
183

David Pears, The False Prison (Oxford: Oxford University Press, 1987).
184

M.S. Hacker, Insight and Illusion (Oxford: Clarendon, 1972), 20-4.

185

The Realistic Spirit, e.g., 181, 194ff..

186

Warren Goldfarb, Metaphysics and Nonsense, Journal of Philosophical Research XXII (1997), 66. See also, in the same issue, Cora Diamond, Realism and Resolution: Reply to Warren Goldfarb and Sabina Lovibond, 75-86.

187

Philosophical Investigations, #304.

188

The existence [Bestehen] and non-existence [Nichtbestehen} of states of affairs [Sachverhalten] is

reality [Wirklichkeit] (2.06).

189

See the perceptive discussion in Max Black, A Companion to Wittgensteins Tractatus (Cambridge:

Cambridge University Press, 1964), 13.

190

We call the existence of states of affairs a positive fact, and their non-existence a negative

fact (2. 06).

191

Critique of Pure Reason, A 369-70. Cf. T. Geach, Saying and Showing in Frege and Wittgenstein, in Essays on Wittgenstein in Honor

192

of G.H. von Wright, ed. Jakko Hintikka, Acta Philosophica Fennica 28 (Amsterdam: North Holland, 1976).

193

Gottlob Frege, The Basic Laws of Arithmetic, ii, 147.

194

Critique of Pure Reason, B ix.

195

Ways of Worldmaking, x.

196

Aristotle, Metaphysics, 1005 b 7-35.

197

Principia Mathematica, Introduction.

198

Gustav Bergmann, New Foundations of Ontology (Madison, University of Wisconsin Press, 1992, edited by William Heald), 173. Page references in the text will use the abbreviation NF. Healds introduction is obligatory reading for all interested in Bergmanns philosophy. He has also included an invaluable glossary.
199

Beaney, M., The Frege Reader (Oxford: Blackwell, 1997), 337.

200

Ibid., 351.
201

Frege, G., 1968, 537.

202

Frege, G., 1968, 34

203

Peter Geach and Max Black, Translations from the Philosophical Works of Gottlob Frege (Oxford: Blackwell, 1970) 83.
204

Function and Object and Concept and Object, both in Beaney.

205

Beaney, 358.
206

Ibid., 358.
207

Ibid., 363.
208

The Philosophy of Logical Atomism (Chicago and La Salle: Open Court, 1996), 101 -103. In Marsh, 235-36.
209

Marsh, 289.
210

R. C. Marsh, ed., Logic and Knowledge (London: Allen & Unwin, 1956), 215-16.
211

Marsh, 288; italics in original.

2 212

Marsh, 214).
2 213

Rudolf Carnap, Introduction to Symbolic Logic and its Applications (New York: Dover, 1958), 107).
2 214

Principia Mathematica, 4
2 215

Gustav Bergmann, Logic and Reality (Madison: University of Wisconsin Press, 1964), 70.
2 216

But see E. Peterson, Real Logic in Philosophy, The Monist 60, 2, 1986, and Logic Knowledge, The Monist 72, 1, 1989.
2 217

Jean-Paul Sartre, Being and Nothingness, 21-45.

2 218

Marsh, 235-36.
2 219

Michael Dummett, The Justification of Deduction, in Truth and Other Enigmas, 311.
2 220

Ibid., 44

221

Gustav Bergmann, New Foundations of Ontology (Madison, University of Wisconsin Press, 1992,

edited by William Heald), 173. Page references in the text will use the abbreviation NF. Healds introduction is obligatory reading for all interested in Bergmanns philosophy. He has also included an invaluable glossary.

2 222

Gustav Bergmann, Generality and Existence, in Logic and Reality (Madison: University of Wisconsin Press, 1964), 69. Originally published in Theoria, 28 (1962), 1-26. Page references in the text will use the abbreviation LR.

223

New Foundations of Ontology, 167.

224

Philosophical Review, 69 (1960), 18-40, also included in Logic and Reality.

2 225

Logic and Reality, 47.

2 226

Logic and Reality, 49-51.

2 227

Logic and Reality, 50.

228

Letters to Russell Keynes and Moore, 73.

229

Tractatus, 6.52 Tractatus 6.521. Tractatus 6.432.

230

231

2 232

Bergmann used these terms in The Glory and the Misery of Ludwig Wittgenstein, Rivista di Filosofia, 52, 1961, 587-406, Italian translation, included in Logic and Reality.
2 233

Alfred North Whitehead and Bertrand Russell, Principia Mathematica to *56 (Cambridge: University Press, 1962), xvii.

2 234

Russell: the Journal of the Bertrand Russell Archives 10, 2, 107-09.

2 235

Cf. Bertrand Russell, The Philosophy of Logical Atomism (Chicago and La Salle: Open Court, 1996), 103.

236

Letters to Russell, Keynes and Moore, 71. In Principia Mathematica (Part I, Summary), an

elementary proposition is defined as one that contains no reference, explicit or implicit, to any totality.

237

The classic discussion of the two interpretations of quantification is Ruth Barkan Marcuss, in

Interpreting Quantification, Inquiry 5 (1962): 252-59.

2 238

Cf. Max Black, A Companion to Wittgensteins Tractatus, 281-8

2 239

G.E.M. Anscombe, Introduction to Wittgensteins Tractatus (London: Hutchinson University Library, 1959), 145.

240

Robert Fogelin, Wittgenstein (London: Routledge, 1987, second edition), 65.

241

Letters to Russell, Keynes and Moore, 73.

2 242

Ludwig Wittgenstein, Philosophical Remarks (Chicago: University of Chicago Press, 1975, ed. Rush Rhees, tr. Raymond Hargreaves and Roger White), 115. For the origin of the text, see the Editors

Note.

243

Ibid., 136.

244

Ludwig Wittgenstein, Philosophical Grammar (Berkeley and Los Angeles: University of California

Press, 1974, 268), composed after 1929 but published posthumously (for the origin of this text, see the editors Note in Editing).

245

Philosophical Grammar, 259.

246

Philosophical Grammar, 265.

2 247

Ludwig Wittgenstein, Remarks on the Foundations of Mathematics (Cambridge: MIT Press, 1994), 300.

248

Logic and Reality, 71. Philosophical Grammar was published 22 years after Generality and

Existence.

249

Logic and Reality, 68, 70.

250

New Foundations, 61.

251

New Foundations, 43.

252

New Foundations, 64.

253

New Foundations, 167-68.

254

New Foundations, 208.

255

New Foundations, 219.

256

New Foundations, 204.

257

New Foundations, 234.

258

New Foundations, 235.

259

New Foundations, 202.

260

New Foundations, 65.

261

New Foundations, 168.

262

New Foundations, 23

263

New Foundations, 216-17.

264

New Foundations, 235.

265

New Foundations, 235.

266

New Foundations, 236.

267

See his Freges Hidden Nominalism, Philosophical Review, 67 (1958). Included in Meaning and Existence, (Madison, University of Wisconsin Press, 1959).

268

New Foundations, 225.

269

New Foundations, 65.

270

New Foundations,

59.

271

New Foundations,

e.g., 59, 21

272

Antoine Arnauld and Pierre Nicole, Logic or the Art of Thinking, translated and edited by Jill Vance

Buroker (Cambridge: Cambridge University Press, 1996), 116.

2 273

Logic or the Art of Thinking, 118.

2 274

These and other generic statements are discussed in M. Krifka, F.J. Pelletier, G. Carlson, A. ter Meulen, G. Chierch & G. Link, Genericity: An Introduction, in Gregory N. Carlson and Francis Jeffry Pelletier, eds., The Generic Book (Chicago: University of Chicago Press, 1995), 1-124.
2 275

F. Strawson, Introduction to Logical Theory (London: Methuen, 1952), 195.

276

Nichomachean Ethics, Book I, 3. W.D. Ross translation. Steven Pinker, Language Learnability and Language Development (Cambridge: Harvard University

277

Press, 1984), 97.

278

Frederick Schauer, Profiles, Probabilities and Stereotypes (Boston: Belknap/Harvard, 2003), 9.

279

Logic or the Art of Thinking, 263. Logic or the Art of Thinking, 114-15.

280

2 281

I discuss it in Metaphysical Realism and Logical Antirealism, Richard Gale, ed., Blackwell Guide to Metaphysics (Oxford: Blackwell, 2002), 282-382, and "Bergmann and Wittgenstein on Generality," Metaphysica, Vol. 7, No. 1, April 2006, 123-145.
282

Husserl, Logical Investigations, I, 25

2 283

Aristotle, De Interpretatione 7, in J. L. Ackrill, Aristotles Categories and De Interpretatione (Oxford, Oxford University Press, 1963), 129.
2 284

Critique of Pure Reason, B 3-4.

2 285

John Dewey, Logic: The Theory of Inquiry (New York: Henry Holt, 1938), 256.

286

Willard Van Norman Quine, Word and Object (Cambridge: MIT Press, 1960), 134.

287

See, for example, J. Higginbotham and R. May, Questions, Quantifiers, and Crossing, Linguistic

Review, 1, 41-79; J. Barwise and R. Cooper, Generalized Quantifiers and Natural Language, Linguistics and Philosophy, 4, 159-219; Stephen L. Read, Pluralitive Logic, in Robert Audi, ed., The Cambridge Dictionary of Philosophy.
288

Nicholas Asher and Jacques Morreau, What Some Generic Sentences Mean, in The Generic Book.

289

Regarding The French are good soldiers, Arnauld wrote: [it] means that the French who are

soldiers are usually good soldiers (Logic or the Art of Thinking, 116).
2 290

M. Krifka, F.J. Pelletier, G. Carlson, A. ter Meulen, G. Chierchia & G. Link, Genericity: An Introduction, in The Generic Book, 25.
291

Gregory N. Carlson, Reference to Kinds in English (New York & London: Garland, 1980), 38.

2 292

Reference to Kinds in English, 30, 36.

293

C.S. Peirce, Vague, in Dictionary of Philosophy and Psychology, ed. J.M. Baldwin (New York:

MacMillan, 1902), 748.

2 294

Gregory N. Carlson, Truth Conditions of Generic Sentences: Two Contrasting Views, in The Generic Book.
295

Kant, Critique of Pure Reason, B 3-4. Norman Kemp Smith translation.

2 296

Metaphysics 1011b25.
2 297

Philosophical Investigations, ## 243-27

2 298

Ludwig Wittgenstein, The Blue and the Brown Books (Oxford: Oxford University Press, 1958), 24-5.

299

The Blue and the Brown Book, 24.

3 300

Critique of Pure Reason, A137/B176 - A138/B177.

3 301

Critique of Pure Reason, A144/B183.

3 302

Bertrand Russell, Theory of Knowledge (Routledge, eds. Elizabeth R. Eames and Kenneth Blackwell, London: 1992), 97.
3 303

Cf. Philosophical Investigations, # 215.

3 304

Thinking and Experience, 38.

3 305

Thinking and Experience, 134.

3 306

Jean-Paul Sartre, Being and Nothingness, , 21-45

3 307

Being and Nothingness, 9-11.

3 308

Thinking and Experience, 124, upper case in original.

3 309

New Foundations of Ontology, 223.

310

See William James, The Principles of Psychology (Cambridge: Mass.: Harvard

University Press, 1983), especially Chapter IX.

3 311

See Being Qua Being and Skepticism about the External World.
3 312

See, for example, Wilfrid Sellars, Empiricism and the Philosophy of Mind, edited by Robert Brandom, (Harvard University Press.; Cambridge, MA; 1997).
3 313

Critique of Pure Reason, Preface to Second Edition.

3 314

The Principles of Psychology, 21.

315

See Susan Curtiss, Genie: A Psycholinguistic Study of a Modern-Day Wild Child (Academic press, 1977) and the more general and readable work by Russ Rymer, Genie: An Abused Childs Flight from Silence (New York:HarperCollins, 1993).

316

Richard Fumerton, Realism and the Correspondence Theory of Truth (Rowman & Littlefield, 2002).

Fumerton focuses squarely on the correspondence theory of truth, calling it alethic realism, and is skeptical about the possibility of achieving an independent and interesting conception of metaphysical realism. The term alethic realism is also used by William Alston in A Realist Conception of Truth (Ithaca, N.Y.: Cornell University Press, 1996). Paul Horwich has argued for a minimalist theory of truth, but it is unclear that his view of reality is minimalist. See his Truth (Oxford: Clarendon, 1998), second edition.
317

Israel Scheffler, The Wonderful Worlds of Goodman, Synthese (v. 45, 1980), 204.

318

Wittgenstein wrote that objects are the substance of the world (021), and that Empirical reality

[Realitt] is limited by the totality of objects (5.5561). By object [Gegenstand] he did not mean the

ordinary things of common sense, yet he contrasted facts with things sans phrase when he declared that The world is the totality of facts, not of things [Dinge], and when he used object and thing interchangeably in A state of affairs (a state of things) is a combination of objects (things) (01).

319

Reinhardt Grossmann, Phenomenology and Existentialism (London: Routledge, 1984), 227.

3 320

Reinhardt Grossmann, The Existence of the World (London: Routledge, 1992), 10.
3 321

Phenomenology and Existentialism, 114.

322

David Armstrong, A World of States of Affairs (Cambridge: Cambridge University Press (March 13,

1997), 136.

323

Michael Devitt, Realism and Truth, Second Edition (Princeton: Princeton University Press, 1997),

28.
3 324

Realism and Truth, 32.

325

Hilary Putnam, Words and Life (Cambridge: Harvard University Press, 1994), 301.

3 326

Paul Horwich, Truth, second edition (Oxford: Clarendon, 1998), 104-05.

3 327

William Alson, A Realist Conception of Truth (Ithaca and London: Cornell University Press, 1996), 41.
3 328

William F. Vallicella, A Paradigm Theory of Existence: Onto-Theology Vindicated (Dordrecht, Boston, London: Kluwer Academic Publishers, 2002), 13, 166-7, 192-3.

329

Jan Westerhoff, Ontological Categories: Their Nature and Significance (Oxford: Clarendon Press,

2005), 74.

330

Ibid., 28.

331

Ibid., 71.

332

Phenomenology of Spirit, 67-79. In Tractatus Wittgenstein wrote: To perceive a complex means to perceive that its constituents are

333

related to one another in such and such a way (5.5423). Gustav Bergmann agreed, especially in New Foundations of Ontology, ed. William Heald (Madison, University of Wisconsin Press, 199

334

F. Strawson, Truth, Aristotelian Society Supplementary Volume 24 (1950), 136-137. See also his

Reply to John R. Searle in Lewis Edwin Hahn, The Philosophy of F. Strawson (Chicago and LaSalle: Open Court, 1998), 402-04.
3 335

Ibid., p 139.
3 336

Donald Davidson, The Structure and Content of Truth, Journal of Philosophy LXXXVII, 6 (June 1990), 304.

337

W.V.O. Quine, The Pursuit of Truth (Cambridge: Harvard University Press, 1992), 77-88.

3 338

Alan R. White, Truth (New York: Anchor, 1970), 80-81.

339

Bertrand Russell, My Philosophical Development (London: Allen and Unwin, 1959), 172-73. I

discuss the status of relations in Being Qua Being, Appendix A (Bloomington: Indiana University Press, 1979).
3 340

W. V. Quine, "Whitehead and Modern Philosophy," originally published in 1941 and reprinted in Selected Logic Papers, enlarged edition (Cambridge: Harvard University Press, 1996), 16.
3 341

Chisholm suggested that the objection was a mere stricture in his reply to my States of Affairs, in Radu J. Bogdan, ed., Roderick M. Chisholm (Dordrecht: Reidel, 1986).

3 342

Thought, 34

343

Gottlob Frege, On Sinn and Bedeutung, in The Frege Reader, 158.

344

Ibid., 156.

345

For a criticism of the definition of knowledge as justified true belief, see my Skepticism about the

External World (New York: Oxford University Press, 1998).

346

I have argued elsewhere that this is also how G. E. Moores view that good is indefinable should

be understood. See my That Simple, Indefinable, Nonnatural Property Good, The Review of Metaphysics XXXVI (1982), 51-75; and Skepticism in Ethics (Bloomington: Indiana University Press, 1989).

347

Cf. Hilary Putnam, The Face of Cognition, Journal of Philosophy 91, no. 9, 1994 Dewey Lectures,

Lecture 3, and most of his other writings since then; Nelson Goodman, Problems and Projects (New York: Bobbs-Merrill : 1972); Michael Dummett, Truth and Other Enigmas (Cambridge: Harvard University Press, 1978).
348

Immanuel Kant, Critique of Practical Reason, trans. Lewis White Beck (Indianapolis: Bobbs-

Merrill, 1956), 68.

349

Groundwork, 125.

3 350

Kant, Critique of Pure Reason, A 58 - B 7

351

Kant, Logic (New York: Dover, 1988), 55.

352

Ibid, A 651/B 679.

353

Ibid., dxvi.

354

Richard Rorty, Consequence of Pragmatism (Minneapolis: University of Minnesota Press, 1982),

163.

3 355

W. V. Quine, The Ways of Paradox (New York: Random House, 1966), 105.

356

John Searle, Intentionality (Cambridge: Cambridge University Press, 1991, 218. But Searles

conclusions are different from mine.

357

For a detailed argument, see Quentin Smith, Language and Time (New York: Oxford University

Press, 1993).

358

Science of Logic, p.583. For a recent example, see David Eagleman, Incognito (New York: Pantheon, 2011). David Hume, A Treatise of Human Nature (Oxford: Clarendon Press, 1951), . An Essay Concerning Human Understanding, Book IV, chapter ix, section 3. Ibid, 25

359

360

361

362

3 363

Critique of Pure Reason, B 131 13

364

Ibid, A 350. Ibid, B 404. Ibid, B 136-137.

365

366

3 367

Ibid, A 107.
368

406. Ibid, Part I, Book I, Chapter 2. Fichte, The Science of Knowledge (Cambridge: Cambridge University Press, 2008, edited and

369

370

translated by Peter Heath and John Lachs), 98.

371

Beyond Good and Evil, # 17

372

Hegels Logic, trans. William Wallace (Oxford: Clarendon Pressa, 1975), # 41, 4

373

Hegels Logic, 47, 75. Phenomenology of Spirit, 47. Hegels Logic, 24, 38. Hegels Logic, 23, 36. The Status of Sense-Data, in Philosophical Studies, 174. See also A Defence of Common Sense

374

375

376

377

in J. H. Muirhead (ed.) Contemporary British Philosophy, 2nd series (Allen and Unwin, London: 1925), 211-212, 223.

378

G. E. Moore, Lectures on Philosophy (London: Routledge, 1966), 163. For an earlier discussion of egoism see my Skepticism in Ethics (Bloomington and London: Indiana

379

University Press, 1989), and of skepticism see Skepticism about the External World (New York: Oxford University Press, 1998).

380

Phenomenology of Mind, 58. Hegels Phenomenology of Spirit, 56.

381

382

The Refutation of Idealism, 25. Ibid., 20. Jean-Paul Sartre, Interview, The Philosophy of Jean-Paul Sartre, edited by Paul Arthur Schilpp

383

384

(LaSalle, Il.: Open Court, 1981), 11.

385

Jean-Paul Sartre, TheTranscendence of the Ego (London: Routledge, 2004). Also Being and

Nothingness, 11-14.

386

Martin Heidegger, Poetry, Language, and Thought, trans. Albert Hofstadter (New York: Harper,

1976) 121.

387

Hegels Logic, 38, 6 Hegels Logic, 3, 6. Empiricism and the Philosophy of Mind, p. 176.

388

389

3 390

Cf. Donald Davidson, The Very Idea of a Conceptual Scheme, Proceedings and Addresses of the American Philosophical Association 47 (1974), 5-20.

391

Phenomenology of Spirit, # 177.

392

Terry Pinkard, Hegels Phenomenology and Logic: an overview, in Karl Ameriks, ed., The

Cambridge Companion to German Idealism (Cambridge: Cambridge University Press, 2000), 164. See also Pinkards Hegels Phenomenology: The Sociality of Reason (Cambridge: Cambridge University Press, 1996).
393

Phenomenology of Spirit, analysis of # 177. Phenomenology of Spirit, # 177. Phenomenology of Spirit, 112-113. Hegels Logic, # 86.

394

395

396

3 397

Hegels Phenomenology of Spirit, 2.

3 398

Hegels Logic, # 13.

399

Science of Logic, p. 580.

400

Hegels Phenomenology of Spirit, trans. A. V. Miller (Oxford: Oxford University Press, 1977), 43.

401

Hegel, Elements of the Philosophy of Right, trans. H.B. Nisbet (Cambridge: Cambridge University

Press, 1991), 239.

402

Moving examples can be found in Sebastian Junger, War (New Yoek: TwelveBooks, 2010), which

also includes a useful bibliography.

403

See Karl Mannheim, Ideology and Utopia, trans. L. Wirth and E. Shils (New York: Harcourt, Brace

and World, 1936).

4 404

E.g., Alan Goldman, Epistemology and Cognition (Cambridge, MA: Harvard University Press, 1986) and Steve Fuller, Social Epistemology (Bloomington: Indiana University Press, 1988).

405

Berkeley, Principles of Human Knolwedge, Part I, # 41. G. E. Moore, The Refutation of Idealism, included in Philosophical Studies (New York: Harcourt

406

Brace, 1922)

407

The Refutation of Idealism 1.

408

Meditations, III. Essay Concerning the Human Understanding, I, 1, 8. Treatise of Human Nature, Book I, Part I, Sect. I.

409

410

411

Critique of Pure Reason, B 376-377.

412

Hegels Logic, 3, 6. I shall follow the practice of most translators of writing notion with an initial capital, in order to mark that in Hegel it is largely a technical term.
413

Hegels Logic, 1, 3. Hegels Logic, 164, , 229. Hegels Logic, 24, 36. Science of Logic, p. 627. Science of Logic, p. 597.

414

415

416

417 4 418

Science of Logic, p. 755.

419

Hegels Logic, 14.

420

Science of Logic, p. 824. Science of Logic, 71.

421

422

E.g., at Science of Logic, 61.

423

Science of Logic, p. 587. Science of Logic, 824. M. J. Inwood, Hegel, London: Routledge, 1983, 157.

424

425

4 426

Science of Logic, 149.

427

Hegel's Lectures on the History of Philosophy, London New York: Routledge and Kegan Paul ;

Humanities Press, 1974), 257.

428

Science of Logic, p. 536.

429

Science of Logic, p. 555. Spinoza, Ethics, New York: Hafner, 1963, Part Two, Proposition VII, Note, 84.

430

4 431

Hegels Logic, 213, 274, 275.

432

Hegels Logic, 24, 36.

433

Hegels Logic, 213, 274, 275.

434

Phenomenology of Spirit, 120. See Being qua Being. Gilbert Ryle, Ordinary Language, Philosophical Review LXII (1953), 170. See also Gilbert Ryle,

435

436

"Use, Usage, and Meaning," Proceedings of the Aristotelian Society Supplementary Volume XXXV (1961).
437

Ordinary Language, 173-74. Gilbert Ryle; J. N. Findlay, Symposium: Use, Usage and Meaning, Proceedings of the Aristotelian

438

Society, Supplementary Volumes, Vol. 35. (1961), 233.

439

Hegels Logic, 163, z2, 228. Science of Logic, 586. Hegels Logic, 163, 227.

440

441

4 442

Hegels Logic, 163, 228.

4 443

Hegels Logic, 20, 29.

444

Hegels Logic, 24.

4 445

Hegels Logic, 24, 37.

4 446

Hegels Logic, # 31.

4 447

Hegels Logic, 24, 37.

448

Ibid. Science of Logic, p. 607.

449

4 450

D. C. Williams., The Elements of Being, Review of Metaphysics 7: 3-18, 171-192. I have discussed the view in detail in Resemblance and Identity: An Examination of the Problem of Universals (Bloomington and London: Indiana University Press, 1966).

451

Hegels Logic, 24, 37.

4 452

Hegels Logic, # 13.

4 453 454

Hegels Logic, # 80, 113. N. Goodman and W. V. Quine, "Steps Toward a Constructive Nominalism," Journal of Symbolic

Logic, 12 (1947).
455

Analysis and Metaphysics, 86.

4 456

P. F. Strawson, Scepticism and Naturalism (New York: Columbia University Press, 1985), 45.
4 457

P. F. Strawson, Analysis and Metaphysics (Oxford: Oxford University Press, 1992), 64.
4 458

Scepticism and Naturalism, 38.

459

Analysis and Metaphysics, 86.

4 460

F. Strawson, Imagination and Perception, in Freedom and Resentment and Other Essays (London: Methuen, 1974), 57.

461

Hegels Logic, 23, 36. Tractatus 5.64. Hegels Logic, 41, 67. Fichte, The Vocation of Man, trans. William Smith (Chicago: Open Court, 1910), 88-90.

462

463

464