NOZICK, Robert. Invariances.

16 INVARIANCES Truth and Relativism 17
care about, and act upon are relative. Does the relativist need more than To say that relativism about truth is a coherent position is not to say that
this? Would the absolutist be content if there exists at least one absolute it is the correct position. The relativisms about truth that have been formu-
truth, something firm that he can hang onto even if not do much with? lated previously and that have great interest (for instance, that truth is rela-
(And will he be content if he merely can maintain that such a statement ex- tive to one's society, or one's sex, or one's economic class) are, I think,
ists, but cannot specify which statement it is? Or will he be content if the extremely implausible. However, philosophically very interesting, though
one nonrelative truth is something like this: There exists one nonrelative socially less striking, versions of relativism can be formulated that are plau-
truth and this very statement is it.) Won't this make the absolutist's victory sible. That I have reached this position surprises me. My original intent was
hollow? Isn't the more important issue, or an equally important issue, the to argue that relativism about truth was coherent but false. To give the
question of what proportion of the truths we believe, put forth, and act nonrelativist position falsifiable content, I needed to produce formulations
upon are merely relative? of how relativism could coherently be true, and doing this led me to notice
Thus far I have imagined the relativist granting that some statement is interesting ways that relativism could well turn out to be true!
nonrelative, namely, the statement of the relativist position itself (along To proceed further, we need to understand more exactly what relativism
with its consequences). Call this the relaxed relativist position. But what involves. A property is relative only when there is an unobvious nonexplicit
about unrelaxed relativism? Is the claim that relativism is universally and factor such that the presence or absence of the property (or the degree of
without exception true undercut by that familiar question: is this relativist its presence) varies with that factor. But this is not enough. The length of
view relative or not? It seems that surely it is, for if we suppose that the (un- an object is different under different temperature conditions, yet this fact
relaxed) relativist position R (which holds that all truth is relative) itself is alone does not show that length is relative to temperature. Causal variation
relative, then . . . Then what? Then it is only relative, and so what authority under different conditions is not the same as being relative. The different
does it have over me; why should I believe it? The relativist might reply that temperature conditions are exclusive; an object cannot have different tem-
although R is only relative, I fall within its domain. R is relative to a group I peratures at the same time. Whereas, when a property is relative to a factor,
am in, or to a property that I have, etc. And so the relativist answer to the the different states of the factor can obtain simultaneously, in different re-
question of why I should believe R, even though it is merely a relative truth, spects. To ask simply whether or not the property is present, the relevant
is that it is true relative to something about me. So I am stuck with it as state of that factor must be specified.
true. (And about that statement that R is true relative to something about The probability of a statement or of an event provides an example of
me, I ask: is that statement only a relative truth? Yes, replies the relativist, something that is relative. The probability of a statement is relative to evi-
and it too is true relative to something about me. And as to that something dence. That probability will vary with different evidence, and that probabil-
about me, I wonder: is its holding only a relative truth? Yes, the relativist re- ity is not detachable from the evidence as something that holds as a free-
plies, but . . . ) If R is true relative to property I? and if I have property P and standing fact.2And the probability of an event's having a certain property is
do not lose it through knowing that truth is relative (and knowing it is rela- relative to a reference class. Different classes into which that event falls will
tive to P and knowing that I have P), then R is true relative to this property show differing percentages of events having the property in q ~ e s t i o n To.~
that I have, so shouldn't I believe it? speak of the probability (period) of a statement or of an event, we have to
This makes it look as though relativism about truth is a coherent posi- take as given or to hold constant the evidence or the reference class. Indeed,
tion, and that the quick and standard refutation is much too quick. And I this is not enough. Rather, we must speak explicitly of the probability of a
think that relativism about truth is a coherent position; the relativist claim statement relative to given evidence or the probability of an event relative to
can be structured and stated so that it need not undercut itself. Moreover, its being of a certain type.
the nonrelativist should think it is desirable that relativism be coherently There is a further feature to be mentioned when a property P is relative
formulable. For then nonrelativism (absolutism) about truth becomes a po- to the state of a factor. (This feature is implicit in the previous characteriza-
sition with (Popperian) empirical content; it excludes something that is tion, but it should be stated explicitly.) Suppose that the factor F to which
possible. That gives the nonrelativist position more bite and thus makes it the property is relative can take on different values or degrees, F,, . . . , F,,
more interesting and that these different values are not exclusive; they can hold in different
respects. We have to add that no one of these different values is privileged,

states a relativity, but its statement of that relativity is not itself relative-or
or superior to the others. They all are on a par, equally valid, or equally
does not appear to be.5
good, or whatever. If one of these values, say Fa, of the factor was privileged
We now are in a position to state what must obtain if truth itself is to be
or superior, then the state of P relative to F. could be picked out as defining
relative. Everyone agrees that the statement made by an utterance, e.g.,
the freestanding and nonrelative property l? Relativism must be egalitarian
"The book is on the table," will vary with or depend upon the meaning of
about the variations in the factor E
the utterance and with what items in the world the terms in the utterance
In the Special Theory of Relativity, the length of an object will be mea-
refer to. The correspondence theorist of truth holds that the truth value of
sured differently by different observers traveling with different constant ve-
the statement (made by the utterance in its context) also will vary with or
locities with respect to the object. Put differently, the measured length of
depend upon the way the world is-e.g., upon whether the book is on the
the object will vary according to its velocity with respect to an (inertial) ob-
table. So much is common ground and does not establish any relativity of
server. A privileged velocity of the object (or of the observer) could be used
truth. Truth is relative only if it varies with some additional, less obvious
to pick out one (measured) length as its true, absolute length. Absolute
factor.
space, if it existed, would provide a framework for defining an object's ab-
A set T of truths contains relative truths if the members of Tare true and
solute length, namely the length that would be measured by an observer at
there is a factor F other than the obvious ones (the meaning of the utter-
rest with it in absolute space. However, within the Special Theory of Rela-
ance, the references of some terms within it, and the way the world is), such
tivity there is no framework (such as absolute space) to pick out one iner-
that that factor F can vary, and when F is varied, the truth value (viz. truth
tial observer's position as better or more privileged than another's. The
or falsity) of the members of T varies. The different values of the factor
length of an object is relative to the motion of an observer.
can coexist, and they are on a par. So the truth value of the members of T is
A similar situation holds with respect to our example of probability. The
a function of factor F as well as of meaning, reference, and the way the
probability of a hypothesis is relative to evidence, but if one batch of evi-
world is6
dence could be picked out as privileged, for example, the total evidence that
The claim that truth is relative to gender or race or class or culture or
is available to us, then the probability of a hypothesis simpliciter might be
sexual preference says that variation in at least one of these factors would
defined as its probability relative to the total evidence. (This probability
produce variation in the truth value of some statements whose content is
would shift, of course, as we acquired new evidence.) Or if one of the refer-
not explicitly about matters of gender or race, etc? Moreover, variations in
ence classes within which an event falls could be picked out as the most in-
these factors can hold simultaneously, and all variants in these factors are
formative one with regard to the event, for example, the narrowest homo-
on a par; they are equal. If truth is relative to, for instance, anthropological
geneous reference class that includes the event, then the probability of the
culture, then different cultures can consistently coexist, and they (and their
event simpliciter could be specified as its probability relative to this particu-
standards of truth) are held to be equally good. Relativism is egalitarian."
lar reference class.' Thereby would the relativity of probability be elimi-
Some recent formal theories of propositional content and the semantics
nated. However, no such specification of a privileged body of evidence or of
of context-dependent statements may appear to leave no room even to state
a privileged reference class has been successful for this purpose. The no-
the position that truth is relative. 1 provide some technical discussion of
tions of inductive and of statistical probability remain relative.
these theories in a note? but the conclusion I reach is of general philosophi-
A property is relative, then, when its occurrence or value varies with
cal interest and needs to be stated here. Relativism does not differ from ab-
some other factor, when the different values of this other factor can coexist,
solutism merely by claiming that statements have the traditional truth val-
and when no one of this factor's values is especially privileged or superior
ues True and False but possess these traditional truth values only relatively.
to every other one.
Rather, relativism about truth holds that truth values are of a new and dif-
Once the relativization is made explicit within the statement itself, how-
ferent kind. A statement or utterance can have the truth value [true relative
ever, it seems that the resulting statement (for instance, "The probability of
to factor F,] or the truth value [false relative to factor F2].And when these
hypothesis h relative to evidence e is greater than .5") is not itself relative. It
factors are different, it can have both of these truth values at once. These
INVARIANCES Truth and Relativism 21
truth values are structurally different and more complex than the tradi- fail to notice these dots of infinite continuation.) But cannot the relativist
tional truth values, and they are intrinsically relative. No statement, the (tb- quote all these, up to and including the dots, and say that all that is true for
tal) relativist says, has the traditional truth value of true simpliciter or false the absolutist but not for himselt? Is there any way that the absolutist can
simpliciter. state his position so that the relativist cannot go ahead and bracket it as
The position that truth is relative cannot be ruled out because of the ar- only relatively true?
bitrary structural constraints of a theory of propositional meaning that was Whereas the absolutist cannot easily turn the tables on the relativist who
formulated and designed for purposes quite unconnected with issues of rel- says (R): all truths are relative. If the absolutist replies (F): (R) is false, abso-
ativism in mind, especially since the relativist position can be formulated lutely false, then the relativist again can say (R2): (F) is true for you but not
within a closely related but more expansive framework. A similar point ap- for me, and the absolutist's reply of (F2) (that (R2) also is absolutely false)
plies to other restrictive theories in the philosophy of language, including only pushes further along a familiar path.
those of sense and reference in the tradition of Gottlieb Frege. Before dis- The absolutist's statement that there is absolute truth can be encircled
missing relativism as incompatible with these theories, a person would have and enfolded by the relativist, leaving the absolutist dissatisfied with what
to investigate whether cognate theories that allow relativism are possible.1 he has said previously. However, while the relativist's statement that truth is
To say that all truths are relative is to say that each and every truth is a relative can be denied by the absolutist, it does not get enfolded by him, and
relative truth. The absolutist about truth says that not all truths are relative. so the relativist can rest content in having at least already fully conveyed his
He need not hold that all truths are nonrelative. He can grant that some position. This symmetry does not show that relativism is true, of course,
truths are relative, in that their truth value can vary with variation in some only that it is infuriating. To the absolutist, that is.
unobvious factors, etc. He holds, though, that this is not true of all true
statements. There are, he claims, some absolute truths, some true statements Who Wants Relativism?
where there is no factor F (other than . . . ) such that F can vary and the
truth value of those statements varies with El1 Moreover, as we have seen, The issues concerning relativism have generated much discussion, often
the absolutist wants to claim that the absolute truths are in some sense the quite heated. People very much want the issue of whether all truths are rela-
important and fundamental ones (perhaps because all the relative truths tive to turn out a certain way. It is worth pausing to consider why this is so.
are derivable from and explainable in terms of absolute truths, perhaps for Why is so much passion aroused by this issue, and what determines which
some other reason). way people want the issue to turn out?
The absolutist has a problem, it seems, of stating a position the relativist In classroom discussions and lectures, I have asked listeners whether they
must deny. Suppose the absolutist says (A): there is an absolute truth. The think that all truth is relative. (As you interpret the terms "relative" and
relativist, conciliatory as always, then says: That (namely, (A)) is true for "truth," is all truth relative?) I also ask them whether they want all truth to
you but not for me. The absolutist then says (A2): There is an absolute be relative.13 Almost all have found themselves in the happy position of be-
truth and it is true for everyone.'* The relativist then says: (A2) is true for lieving that things are the way they want them to be. Now, it could be that
you and your group but not for my group. The absolutist then says (A3): we adjust our desires to the realities of the situation, coming to want what
There is an absolute truth, and it is true for everyone! The relativist says: (we believe) is the case. However, it seems more likely--doesn't it?-that in
(A3) is true for you but not for me. The absolutist then says (A4): It is an this case our (philosophical) belief about relativism follows in the footsteps
absolute truth, which is true for everyone, that there is an absolute truth of our desire. This gives one pause.
that is true for everyone. The relativist says: (A4) is true for you and not for What determines which desire a person has about relativism? I shall offer
me. The absolute says: I'm not going to let you always go up a level to refer some social-psychological conjectures about the factors that shape the de-
to what I say. I am going to make a statement at an infinite number of lev- sire, but I do not mean to say that the truth about relativism is relative to
els, or make an infinite number of statements. Here it is, and here they are: these factors. (Note that the philosophy to come is completely independent
(Sl), that is, there is absolute truth; and (S2), that is, (Sl) is absolutely true; of my amateur psychological conjectures.)
and (S3), that is, (S2) is absolutely true; and. . . (And, relativist, please don't Relativism about truth, along with cognate views such as that there are
no objective facts and that all facts are "social constructions," is seen by its lowered their level of aspiration and substituted other goals more easily
proponents as enhancing freedom. Objective facts are a constraint; they achieved. Thereby they are likely to find the facts of the world less of a con-
limit what one can do. straint to achieving the goals they actually have. (As for the unavoidable in-
When we have various goals, the way to get to these goals is by perform- creasing constraints that they encounter as they age, they do not think that
ing actions that launch along paths to realizing these goals. Facts can there- these being relative would make them any more escapable.)
fore constrain the achievement of your goals. It may be a fact that the ac- However, the relativity of a truth is not the same as its alterability. Even if
tions available to you do not begin any existing or possible paths that lead it is a nonrelative truth that my pen is on my desk, that is a fact easily
to achieving your goals. Or it may be a fact that every path that might lead changed.I5Whereas if it is merely a relative truth that New York City is adja-
to your goals unavoidably involves side effects that are significant (and un- cent to the Atlantic Ocean or that capitalism outproduces socialism, these
acceptable). Therefore, it can enhance your freedom and power if you are are not facts that are changed easily. Whatever these facts might hold true
free of these constraints, if the facts that constitute them or underlie them relative to, is not easily altered.
are not objective facts. Why, then, would some people welcome relativism and assume that rela-
On the other hand, the existence of objective facts, of particular ones that tive facts are less constraining?To account for such motivation, we need not
are believed to hold, can be desired because these facts are seen as a plat- assume that every relative truth is more alterable than any absolute truth.
form upon which to stand in pursuing your goals, as a means to get to your Still, for any given statement p, p almost always will be less constraining, or
goals. If it is a fact that this path will lead from an available action to your more alterable, or more escapable, if p is a relative truth than if it is an abso-
goal, then that fact enables you to achieve your goal. If it is a fact that an ac- lute one. For p's being relative at least holds out the hope of escaping from
tion will have particular side effects and you do better to avoid these, then its clutches, by losing the property relative to which it is true, or by moving
certain other available actions may facilitate this avoidance. into a group or culture relative to which it is not true. Even if you cannot
So which is the true character of facts, constraining or enabling? Clearly, change your own traits or situation, it might be liberating merely to know
facts have both aspects. (Rationality, too, can be seen as a constraint or as that someone somewhere isn't (or may not be) subject to that truth.
an enabling means.) If generally you are achieving your important goals, The desire to escape the reach of obdurate facts and compelling argu-
achieving them well enough even if not perfectly or optimally, if there are ments can extend to the theoretical realm as well. It is this desire (and not
paths leading out from your actions that enable you to reach your level of the book's illuminating description of the texture of scientific practice) that
aspiration, then you will view facts, in general, as enabling. If there are no accounts, I think, for the enormous public impact that Thomas Kuhn's The
such paths, however, then facts will appear to you mainly as constraining. Structure of Scientijic Revolutions has had since the 1970s. That was a time
Whether people welcome or oppose relativist doctrines depends, I conjec- when many cherished theoretical positions on the political left faced very
ture, upon which of the two situations they are in.14 serious intellectual challenges and difficulties. Frederick Hayek and Milton
According to this explanation, the desire that truth be absolute arises Friedman produced powerful arguments about how markets operated and
from satisfaction with the particular truths that are believed to hold, and about how socialist systems failed to operate. These arguments were but-
the desire that truth be relative stems from dissatisfaction with particular tressed by "neoclassical economic theory" and also by evident and hard-to-
truths. The philosophical preference for absolutism or relativism, then, is a deny facts about economic growth under capitalism, economic stagnation
derivative one. under socialism. What to do? Along comes Thomas Kuhn with his talk
We therefore can understand why young people are more prone to favor about different "paradigms," even in the apparently hard science of physics.
and welcome relativism than the old. Younger people often are under the To others (I do not say that this was Kuhn's motivation), Kuhn seemed to
external constraining force of parental authority, and they may have fewer supply a ready way to avoid facing very strongly backed theories, criticisms,
means or paths available to their goals. Older people, on the other hand, ei- and data. A person (it was supposed) could rest secure within his own po-
ther are more likely to have discovered or acquired means to the realization litical paradigm, which was as good as any other competing theory, and he
of their goals or, if they were continually frustrated in their pursuit, to have could feel licensed to ignore his own view's very evident difficulties. One
24 ZNVARZANCES Truth and Relativism 25
simply said, 'Riou have your paradigm and I have mine." End of discussion. and it could be their prior tolerance that predisposes some people toward
Supposedly. relativism.
Increasing economic prosperity also has made it possible for people to be Many people who hold that all truths are relative also hold that many
more relativistic, I think. In earlier times, people either had to adjust their things, previously thought to be facts holding independently of human de-
beliefs and desires to a hard and difficult world or they failed to survive. sires, beliefs, and choices, instead are "social constructions." This correla-
They would have preferred, however, that the adjustment went in the oppo- tion, I think, reinforces our first tentative explanation for the desire that
site direction, that the world conformed to their beliefs and desires. Because truth be relative, viz. it depends upon finding many central and (socially)
prosperity fortunately enables life to be lived less close to the margin, it en- important facts thwarting, frustrating, unpleasant, and undesirable. A "so-
ables people to maintain whatever their current beliefs and desires are, cial construction" is not simply a fact unexplained by biology and other lev-
at the cost of some loss of mesh with the actual world, to be sure, but with- els lower than the social. Consider such facts as the existence of educational
out very dire consequences. By thinking that all truths or facts are relative, systems or some other mode of transmission of adult knowledge to youn-
(it appears that) a person can hold that all his beliefs are true without ger people, or the existence of armies or defensive forces to protect a society
change-relative to himself and his situation, at least. (On the other hand, from outside attack. (I mean the existence of these general institutions, not
increasing prosperity also might make some people more satisfied with the particular forms that they take.) These might well be universal facts
their situation and hence less prone to relativism.) constant across all societies that last for at least two generations.17 Such in-
The relativist sees the divisions among people as so great that truth can- stitutions or practices would be social facts but not "social constructions:'
not be the same for them; the nonrelativist (as we shall see) thinks the simi- for we would understand the point and rationality of such institutions and
larity in humans and in their environments is sufficiently great to make see their practical necessity. A social construction, on the other hand, is ar-
(the nature of) truth the same for all. Might people's different attitudes to- b i t r a r y r at any rate there is no good reason, no justifying reason, for its
ward relativism also depend upon whether they want the divisions among existing or for its being anything like the way it is. Whether or not it can
people to be large and insurmountable? This puts the absolutist in a favor- easily be changed (and some social constructions might have come to be
able light, allowing him to favor the unity of humankind. However, his op- deeply embedded), a social construction is something without authority.
ponents might see the absolutist as favoring unity on his own terms, in the Hence, the thesis of social constructionism often is selective rather than
belief that the one absolute truth is the truth as he sees it. And he might universal. Only those things are held to be social constructions that the
have the power to impose that view on others, or at least to compel their speaker wishes were different and holds to lack any authority; therefore,
public assent to it. Thus, less powerful groups (women, racial minorities, they do not have to be adapted to or obeyed or adhered to.'" Because social
homosexuals, minorities on the political left) appear to offer higher per- constructions are arbitrary, it is perfectly permissible to alter them, to go
centages (of vocal proponents) favorable to relativism, perhaps in resis- against them, to violate them.
tance to the hegemony of the powerful. Current calls for multiculturali ism^' When all facts are viewed as social constructions, then facts (in general)
come primarily from groups that view themselves as being subordinate.16 lose authority. They need not be accepted as limitations upon what one
An alternative possible explanation for why some people especially wel- might do. Some objective facts are obdurate; while we may wish they were
come relativist views might be that these people are more tolerant in gen- different, we discover that they cannot (easily) be changed. Hence they have
eral. True, an absolutist about truth also might be tolerant, for an absolutist to be taken account of, be worked around, and be granted a certain, per-
might well be a fallibilist, uncertain that he definitely has hold of the abso- haps grudging, respect. (It is such facts that Freud's reality principle bids us
lute truth that exists. And an absolutist who even is certain he knows some to recognize.) If all facts are social constructions, however, then since they
particular truth might be tolerant of those who are mistaken, while a rela- lack all authority, we would do better to devote energy toward changing un-
tivist might turn out to be intolerant of various positions, especially of desired social facts rather than adapting to them or contouring our institu-
nonrelativist ones. Nevertheless, relativism might fit more easily psycholog- tions around them.
ically, and hence more often, into an attitude that is tolerant of differences, The theory that some particular thing is a social construction may be an
illuminating theory and even a correct one. However, it cannot be that ev- raining in a place at a time-raining softly there then and not raining hard
erything is a social construction; it cannot be that all truths and facts and there then? So Aristotle stated his principle of noncontradiction carefully.
things are social construction^.^^ For to say that something, for instance Nothing is both true and false at the same time in the same place in the
gender differences, is a social construction is to say that there are social pro- same respect. It is that fully and exactly specified thing which cannot be
cesses that cause this phenomenon to exist or to take a certain form. If gen- both true and false.
der differences are a social construction, then it is a fact that certain existing It is not surprising, then, that almost all philosophers have held truth to
social processes produce those gender differences. And that fact-that those be timeless. If it is true today that it is raining, while yesterday it didn't rain,
social processes exist and produce gender divisions-is not itself a social since "It rains today" is an incompletely specified proposition, no statement
construction. (Or, if it too is a social construction, the fact that it is, is not or proposition changes its truth value from one day to the next, despite ap-
itself a social construction.) William James tells the story of a person ap- pearances. Specify the date, the time, and the place, and that completely
proaching him after a lecture and saying, "The world rests on a large turtle." specified tenseless statement will be timelessly true. It is true at every time
"And what does the turtle rest upon?" James asked. "Another turtle:' said and place that it rains in Boston, Massachusetts, on June 13, 1998. Fully
the person. "And what . . . ," began James, who then was interrupted: "Pro- specified truths, since they have their times and places built into them, are
fessor James, it's turtles all the way down.'' There cannot be social construc- timeless and placeless.
tions all the way down. Almost all philosophers, therefore, agree that a fully specified proposi-
It is worth noting that even if all truth turns out to be socially relative, it tion or statement would have a fixed and unvarying truth value. W. V.
does not follow that anything and everything holds true. If all truth is rela- Quine held that the vehicle of truth is (or could be viewed as) "eternal sen-
tive, then every truth holds relative only to some factors or other, and not tences," that is, sentences with timeless verbs and all temporal indicators ex-
beyond. (For every truth, there then will be some factor relative to which it plicitly included.20This trivial formal device seems to guarantee the time-
does not hold.) It is a different matter to claim that each and every state- lessness of truth.
ment has the status of being a relative truth. That would entail that for each I think, however, that the timelessness of truth is a contentful empirical
and every statement, there is some factor relative to which it holds true. So claim, one that might turn out to be false. In order to make it possible even
if someone makes a statement, and defends himself against criticism by to state this possibility, let our ascriptions of truth to statements speak ex-
saying that every truth is relative, one can reply that, nevertheless, his previ- plicitly with two temporal indices, one referring to the time of the events
ous statement is false. There may be no factor relative to which it holds true the statement describes, and the second referring to the time when the
(in which case, its denial will be absolutely true), or even if there is, that fac- statement is true. Thus, consider the statement stating that event E occurs
tor need not apply to the speaker. What could have been true in someone at time 4. In ascribing truth to this statement, we must state explicitly the
else's mouth in some other situation is false when he speaks. Relativist doc- time at which it is true. We shall have to say things of this form: It is true at
trine does not save him. time tkthat event E occurs at time 4. And this might be compatible with its
being true at another time tmthat event E does not occur at time t,. (And
Truth in Space and Time false at time tmthat event E occurs at time 6.) The truth of even fully speci-
fied statements, then, would be relative to time.21
When he resented the principle of noncontradiction-that a statement It is interesting that Aristotle is a notable exception to the general philo-
and its denial are not both true, that no statement is both true and false- sophical consensus that truth is timeless. For Aristotle (according to the
Aristotle had to state it sharply to avoid various apparently obvious usual interpretation) held that it is neither true nor false to say beforehand
counterexamples. Cannot it be both raining and not raining? In different of a future contingent (undetermined) event that it will happen. If a sea
places-raining in Boston, not in New York. And can't it be both raining in battle is to be fought on Tuesday between two opposing fleets, and on Mon-
a certain place and not raining in that place? Raining there on Monday, not day it is undetermined which fleet will win the battle, perhaps because vic-
there on Tuesday. And can't it be both raining in a place at a time and not tory depends upon commanders' decisions that themselves are free and un-
determined, then on Monday it would not be true to say, "The first fleet will The notion of its being determinate at a time that an event occurs at an-
win the battle tomorrow" and on Monday it would not be true to say, "The other time, of its being determinate at a time that a statement holds true ac
second fleet will win the battle tomorrow." that other time, is an important one that is worth investigating. What are
Aristotle's root intuition here, I think, is this one. A statement is true at a this notion's properties, and to what extent, and in what way, is it actually
time only if something holding at that time makes it true, fixes it as true, realized and instantiated in the world?
constitutes it as true, determines that it is true, makes it determinate that it Some may say that this is all well and good but it is quite a different mat-
is true. We might say that for something to be true at a time, there has to be ter from the topic of truth. Determinately holding is one thing, being true is
information at that time that fixes it as true. Here, though, information another. I do not want to quarrel over what may be a verbal issue. I do think
does not mean something that is known but facts that hold. For something that "determinately holds" is the interesting and important notion. If deter-
to be true at a time there must be facts that hold at that time in virtue of minateness turns out to be more rooted in time and space, its span drasti-
which that thing is true; there must be facts holding at that time that make cally smaller, than once was thought, then whatever apparent distinct con-
it true. tent the further notion of "true" may have is drained of interest because of
Because Aristotle believed that which fleet is victorious in tomorrow's the way the world is, the way we have learned it to be. It is determinateness
sea battle depends upon future decisions of the captains that are not now that stitches the universe together, or fails to do so. A timeless truth that
determined or determinate, he believed that there are no facts holding now floats free of determinateness is a nonscience fiction.
that will make one fleet, or the other, victorious later, and so he believed The standard theories of tense logic hold that the following statements
1
that it is not now true that the first fleet will win the battle, and it is not now are true; indeed, they hold these to be necessary truths.23
true that the second fleet will win the battle. It is not simply that the battle I
has not yet taken place. It is not yet true which fleet will win. 1 (A) If it is true at a time that event E occurred at an earlier time, then it
But cannot we always say, when a statement holds at a certain time, that ~ is true at that earlier time that E occurs at that time.
in virtue of that fact it also was true earlier that it will hold at that later ~
I
(B) If it is true at a time that event E occurs at that time, then it is true at
time? The correspondence theorist will assert that a statement is true when all later times that E did occur at that earlier time.
it corresponds to a fact, and not merely to a future fact. The root intuition (C) If it is true at a time that event E occurs at that time, then it is true at
behind the correspondence notion of truth extends to saying that a state- all earlier times that E will occur at that later time.
ment is true at a time if and only if it corresponds to a fact holding at that
time. Yet it might seem that we always can trivially postulate additional 1 I believe that quantum mechanics makes it plausible that each of these
statements (A)-(C) is false. (One might say that quantum mechanics makes
facts (with appropriate tenses) that will mirror, at all other times, any fact
it plausible that these statements are sometimes false, but since they are as-
that holds at any time. However, if we take an ontology of facts (or of what
serted with full generality, if they are false in some cases then they simply
underlies them) seriously-as a correspondence theorist of truth must-
are false.) If these statements are false, they certainly are not necessary
then we should not multiply facts capriciously,beyond necessity. (Consider
truths of tense or of any other logic. I do not say that quantum mechanics
how one would react to someone who said that whenever a new object demonstrates that these statements are false, merely that their falsity fits
,
comes to exist, there must always have been previously existing ghostly an-
with a plausible reading of quantum mechanics that equates determinate-
ticipators of that object.) The question is whether there really are such mir- I
ness and truth.

roring facts that actually do hold at each of the other times. I doubt that
What quantum mechanics does show is that the following corresponding
purely philosophical arguments can settle such an ontological question. In-
statements about determinateness do not (always) hold true.
deed, it is difficult even to formulate a criterion that will distinguish real
facts or changes from their linguistic simulacra.22Those who attempt to 1 (1) If it is determinate at a time that event E occurred at an earlier time,
mandate the timelessness of truth by stipulation or fiat or verbal legerde- 1 then it is determinate at that earlier time that E occurs at that time.
main may prevent us from discovering the interesting empirical nature that I j (2) If it is determinate at a time that event E occurs at that time, then it
truth actually has.
I .ti
is determinate at all later times that e did occur at that earlier time.
30 INVARIANCES Truth and Relativism
(3) If it is determinate at a time that event e occurs at that time, then it is Merging streams. If light of +45-degree polarization is fed into a horizon-
determinate at all earlier times that e will occur at that later time. tallvertical detector (a forward polarizer) and then both streams are fed
into a reversed polarizer, reuniting the streams, the light emerging from this
I begin by listing some phenomena of quantum mechanics. (The re- second polarizer will have +45-degree polarization (as measured by a later
mainder of this section is somewhat technical.) The first phenomenon is +45-degree detector). Is it that the initial +45-degree light that was fed
well known. into the horizontal/vertical detector did not have a measurement made
upon it, since it was immediately fed into a reversed polarizer that merged
The two-slit experiment. Photons are fired from a source toward a thin the two streams, or that it did have a measurement made but this measure-
barrier containing two slits, which may be opened or closed. Behind the ment was erased? In either case, the light coming out of the second hori-
barrier is a sensitive screen that registers what impinges on it. When both zontal/vertical (reversed) polarizer now has the original +45-degree polar-
slits are opened, with no detector active at either slit, the far screen shows ization. The two horizontal/vertical polarizers together did not do what
interference phenomena characteristic of waves, even when the photons are either alone would do, namely, produce a stream of half +&degree polar-
released very slowly. (When either slit alone is open, the far screen shows a ization and half -45-degree polarization.
pattern of hits characteristic of particles, but this is not incompatible with a
wave phenomenon.) When both slits are open, with an active measuring The erased measurement. Within the Schrodinger wave formalism, a mea-
device at one slit, the far screen shows a (noninterference) pattern of hits surement of a system by a device establishes a correlation between the states
characteristic of particle^.^^ Moreover, an active detector at one slit, even of the system and the device. It is possible, however, for this correlation to
when it does not register a passing particle, is sufficient to nullify the inter- be erased. (Since the Schrodinger equation is time-reversal invariant, it is
ference pattern. physically possible for the process to be run backward, with the measured
particle and the measuring particle or system interacting again to reach the
The delayed-choice experiment. The decision whether and where to acti- noncorrelated state, even though people may not themselves be able to ma-
vate a measuring instrument to detect along which of two widely separated nipulate this precise a reversal.) If that happens, the information in that
tracks photons have come can be made long after the photons have been measurement is lost irretrievably.
emitted and have entered along one or another (or both) tracks. If the mea- Things are less clear if a measurement (to yield a particular result) in-
suring device is not activated, the interference pattern is exhibited; if it is volves a collapse of the wave packet. Is it possible for this to proceed back-
activated, the noninterference pattern characteristic of particles that have ward through the collapse to the previous superposition? (It is another
traveled definitely along one or another of the tracks is exhibited. If a de- question whether one can make that happen.) If so, a definite measurement
tector is present at one of the slits in the two-slit experiment, the light will involving a collapse can be erased.26In the view of Martin Scully, "it should
behave like particles and not exhibit interference patterns on the screen. If be possible to gain information about the state of a quantum phenomenon,
no detector is present, the light will behave like a wave and exhibit interfer- thereby destroying its wavelike properties, and then restore those properties
ence patterns on the screen. Delayed-choice involves delaying the choice of by 'erasing' the information." A group led by Raymond Chiao performed
whether or not there is a detector operating until after the light has passed experiments to reinstitute interference patterns in a way that seems to indi-
the plane containing the two slits. John Wheeler imagined this situation cate that "a collapsed wave function can be put back together again."27
could be realized astronomicallywith light from quasars, with a heavy grav- After collapse of the wave function, there is no fact holding afterwards
itational object such as a galaxy bending the light around it into two curved that makes it determinate what the wave function was before the collapse.
paths that meet at a far distance (and cross if they are not intercepted.) An With the two-slit experiment, a mark on (that part of the wave function of)
observer at that distant place, e.g., earth, can decide, long after the light has an electron that passes through the righthand slit will eliminate the inter-
passed the galaxy, to set up a detector where the two paths come together ference pattern. But if that mark and all traces of it are erased (perhaps the
again (so that interference will be shown) or to set up a detector later along electron has emitted a photon that is reflected back by a mirror so that it
a path, after they cross, recording the action of particles.25 gets reabsorbed into the electron), an interference pattern reemerges. If
there was a collapse whose traces have not been erased, and that collapse rion of meaning, "that a statement about the past now is meaningful only if
eliminated half of the wave function, it then is impossible to tell what the we presently are able, in principle, to veriftr it?" However, the position put
whole wave function had been beforehand. (One can know, however, that forward here is not a verificationist one. I do mean to say, though, that it is
the previous wave function was such that it gave a nonvanishing probability true now that a system was in a past state only if it now is true that some
to the situation that was got to after the c ~ l l a p s e . ) ~ ~ differential effect of that past state continues.30However, it need not be the
All actualities can have effects. (Put more pointedly, what cannot have an case now that the particular differential effect can be discovered or verified.
effect is not an actuality.) Something X has property P only when that can When Aristotle held that there were no facts holding on Monday that deter-
affect the properties of something (else),only when it is possible that there mined or fixed the victor of the sea battle on Tuesday, his point was not that
is something (else) Y and a property Q such that X's having P affects Ys on Monday it could not be verified who would win the coming battle. If
having Q. Xs having P does not require actually affecting something else; it there were facts holding on Monday that determined the victor on Tuesday,
just has to be possible that there exists a thing that X's having P affects. even if these facts could not be detected or verified on Monday, and even if
In some cases, quantum theory excludes this latter possibility. For some no one was aware of the kind of connection there was between such facts
properties P i n some situations, when no measurement is made of X's hav- and a victory the next day, nevertheless it would be true on Monday that
ing Pat time tl, when no measurement is made of whether or not X has P, it one particular fleet, say the Athenian fleet, was going to win on Tuesday.
is not possible (according to quantum theory) that there is something that The issue isn't what can be verified or what can be found out on Monday.
X's having P at time t, affectsz9Such an affecting would itself be a later The issue concerns what actually holds on Monday.31
measurement by a device at time t2;it would not be an effect of X's being P Let us separate two issues. The first concerns the concept of its being
at the earlier time tl. Quantum mechanics, on the usual interpretation, tells fixed and determined at a time that some event occurs at another time. We
us that a definite property exists only insofar as it can possibly affect some- can investigate the properties of this concept, its interrelations with other
thing. And let us add that a measured entity continues to have a property concepts, and what general and what particular statements using the con-
only as long as it is possible that there is something (some other "measur- cept actually hold true in our universe. If our universe were a deterministic
ing" device) whose properties it (does or) can affect. one then the following would hold: if it is fixed and determined at time tj
Recall now the case of the erased measurement. A measurement of sys- that event E occurs at time tj then it is fixed and determined at all earlier
tem S by device D entangled the states of S and D, but now the process has times that event E occurs at time tj. If, instead, our universe is not a deter-
been reversed, the states of S and D no longer are entangled, and the mea- ministic one, then there is (or at least can be) at least one event E and a time
surement information is irretrievably lost. (If we say there was a collapse of tj such that it is fixed and determined at time tj that event E occurs at time tj
the wave packet, we should have to say that the collapse has been reversed.) yet not fixed and determined at some particular earlier time that event E
The usual appeal to macroscopic systems or to irreversible effects in discus- occurs at time tj. And we need not restrict attention to earlier events. It
sions of measurement is designed to fix the stability of the measuring de- might be fixed and determined at time tj that event E occurs at time ti yet
vice's (or of what it affects) being in a certain state so that the information not be fixed and determined at some later time that event E occurs at time
that this is so never becomes (in principle) nonrecoverable. When the only 5. What the actual properties are of the "fixed and determined notion is an
measurement of X's having P (a certain position, say) is completely erased, empirical question.
then X's having had P can no longer affect any other thing's having any These two general statements, corresponding to (2) and (3) above, are
property. Thus, it no longer is determinate that X had I? It no longer is true shown to be false by the stochastic character of quantum mechanics. On
that Xhas P or that it had P! The effects of X's having P were, in turn, erased our assumption that truth is identical with determinateness, the stochastic
so as not to be able to affect anything further; so X's having had P does not, character of quantum mechanics is sufficient to show the falsity of the
at that later time, hold. (purportedly necessary) principles of tense logic, (B) and (C) above. The
This may seem reminiscent of verificationist views about past events. argument against (1) and (A) above was more complicated, and required
"Do you mean to say," asked incredulous objectors to the verifiability crite- something in addition to the general stochastic nature of quantum me-
chanics. It needed to invoke the delayed-choice experiment in a realist way. facts themselves are relative, then a relativist might concede that truth is al-
No current realist interpretation of quantum mechanics interprets the de- ways and everywhere correspondence to the facts, but hold that since facts
layed-choice experiment; Wheeler himself used the delayed-choice situa- are relative, so also is truth.) The "measurement problem" in quantum me-
tion as a reductio of realism. Our argument against (A) and against (1) chanics is that of understanding how measurements can yield definite re-
therefore depends upon some realist interpretation of delayed-choice that sults when the system being measured is in a superposition of states, since it
remains to be developed. There is no guarantee that this can be done. How- follows from the laws of quantum mechanics that a measuring device in-
ever, we can say, at least, that the price of a realist interpretation of delayed- teracting with such a system also will move into a superposition and show
choice is the denial of (1) above, and also (on the assumption that truth is interference properties. The time-dependent Schrodinger equation, when
identical to determinateness) the denial of (A) above. applied to the interaction of a measuring device with a system in superposi-
The notion of whether some event's occurring at a time is fixed and de- tion, does not yield the measuring device going into one particular state
terminate at another time clearly is an important one. A separate issue is (rather than into a superposition). John von Neumann held that there were
how this important notion is connected with the notion of truth. Shall it be two kinds of changes in a quantum mechanical system, the continuous
identified with that notion? This would capture Aristotle's intuition that on change (in accordance with the Schrodinger equation) that occurs when
Monday the truth about the victor of the battle on Tuesday is indetermi- the system evolves in the absence of a measurement, and a discontinuous
nate. It is not merely that the victor cannot be known or verified on Mon- change that occurs when a measurement is made (the "collapse of the wave
day but that, since no facts holding on Monday determine who will be the packet"), which yields a particular definite measured value.
victor, there is no truth holding on Monday that specifies Tuesday's victor. As one route out of the apparently intractable measurement problem,
The contrasting view that truth must be timeless denies the following the literature of quantum mechanics offers some interpretations that ex-
proposition. For a truth to hold at a time, there must be a basis for it to plicitly make the quantum-mechanical facts relative. (I follow the second-
hold then, that is, there must be facts holding then that fix that truth as also ary literature here.) Simon Kochen interprets the measurement situation as
holding. But what is the usefulness of the timeless notion of truth if it floats follows. When two systems in superpositions interact, one being a measur-
free of such a factual basis? How would we discover that truth is not time- ing device, each one relative to the other has one or another definite value;
less-surely an interesting discovery, if true-if we insist on defining or each relative to the other is in a definite state.32However, the total combina-
contouring truth so that its timelessness is guaranteed? tion, relative to the entire outside environment, still remains in a superposi-
It is worth exploring a view of truth at a time that identifies it with being tion and exhibits interference features. So a system has a definite property
fixed and determined or determinate at that time. (Readers who insist on relative to a measuring apparatus that it interacts with, but not relative to
the timelessness of truth can read on, interpreting the discussion not as the outside world that "witnesses" the two systems. On this interpretation,
concerning truth's holding at a time but as concerning truth's being fixed there is no collapse of the wave packet, no discontinuous event that consti-
and determinate at a time. Later, perhaps they will relent and see the point tutes an exception to the Schrodinger equation. Furthermore, the smear of
and the plausibility of the identification.) the indefiniteness of a superposition is not ever present, at least not so rela-
This view of truth as mired in time may seem incompatible with a corre- tive to the measuring apparatus. While both are in a superposition in rela-
spondence theory of truth, according to which a statement is true if it cor- tion to the outside world, one is not in a superposition relative to the
responds to the facts. But are the facts, the fully complete facts (which cor-
respond to fully specified statements), themselves timeless? If not, then a Quantum mechanics is our most fundamental theory of the microlevel,
correspondence theory will hold that a statement is true at a time when and so if, according to it, all definite facts (involving conjugate variables) are
only when it corresponds to facts which hold at that time. definite only relative to states of measuring systems, then we are well on our
Quantum mechanics has led some theorists to maintain that facts them- way to holding that all facts (or at any rate, all the physical facts that are re-
selves can be said to be relative. Although the relativity they speak of is not ducible to quantum mechanics) are relative. (The Schrodinger equation it-
a relativity to time, it is instructive to look briefly at these claims. (If the self would not thereby be held to be relative.) The Kochen interpretation
certainly has not been established as the definitive interpretation of quan- it, since the usual formulation of quantum mechanics offers a consistent
tum mechanics; still, it is a possible interpretation, one that avoids the dif- description according to which it now is true that it was in a superposition
ficulties of the von Neumann interpretation with its collapse of the wave back then. We can say, though, that any realist view that could be developed
packet, and it illustrates how almost all (definite) facts could turn out to be that keeps Wheeler's lessons from delayed-choice would also amount to a
relative.34 refutation of (A).) And the falsity of (C) is shown, one might plausibly ar-
Perhaps facts can be relative, but can it be a relative matter whether some gue, by the randomness of what the wave packet collapses into.
entity is real or not? Some say yes. According to the laws of quantum fields These results should not be very surprising to proponents of the stan-
in curved space-time, dard view of quantum mechanics. This, the Copenhagen interpretation
Accelerated observers just above a black hole's horizon must see the vacuum (named after the Danish physicist Niels Bohr's city of residence), holds that
fluctuations there not as virtual pairs of particles but rather as an atmo- a system is not in a definite state in the absence of an actual measurement
sphere of real particles. This startling discovery revealed that the concept of a showing it to be in that state. (And also, let us add, using the Einstein-
real particle is relative, not absolute; that is, it depends on one's reference Podolsky-Rosen criterion, when it is not true that a measurement definitely,
frame. Observers in freely falling frames who plunge through the hole's ho- with probability 1, would show it to be in that state.) If actually registering
rizon see no real particles outside the horizon, only virtual ones. Observers at a time (or definitely being such as would register at that time) is neces-
in accelerated frames who, by their acceleration, remain always above the sary for the system's being in a definite state at that time, then it is not a
horizon see a plethora of real particles.35 large step to saying the following. A necessary condition for its being true at
another time tz that a system is in a definite state at this time tl is that the
Our brief reference here to quantum mechanics and to cosmology indicates system's being in the state at time tl can possibly register at time t2.
how these very basic theories might lend themselves to the conclusion that With erased measurement, it no longer is true that the particle was in
facts are relative. Facts needing a temporal index may not be such a wild that measured state then. Speaking more carefully, there is a time tl and
idea, after all. there is a later time t2such that it is true at tl that the particle is in that state
According to the view we are exploring, the truth about the past is not at tl, but it is not true at t2that the particle is in that state at t,. What about
fixed. This lesson of the erased measurement is reinforced by the delayed- the statement that a measurement is (was) made, which does not state the
choice experiments. Some people have concluded that delayed-choice im- result of that measurement? It is true then that a measurement is made
plies that the observation we now choose to make causes long-past events then, and it can be true now that a measurement was made then. (The fact
to have occurred (and others have objected to delayed-choice because they that a measurement was made, that there was a correlation then between a
think it implies this). Delayed-choice does have the implication that truths state of a particle and a state of the measuring device, has left a trace now.
about the past are not fixed in the way that people thought, but we must be What has not left any trace is the particular result of that measurement.)
careful in stating precisely what this involves. In this sense, what holds true at a time can be incomplete; it can have
The erased-measurement situation shows that the (previous) second holes. It holds true now that some measurement was made then, but noth-
purported truth of tense logic is false. Something can be true at a time, yet ing holds true now about what measurement resulted then, and so nothing
it not be true at a later time that it was true at that earlier time. The erasure holds true now about what the state of the particle was then.37It could, for
of the effects of the measurement erases the truth that the measurement re- example, be true at time t2that p-or-q held at time tl, although it is not true
vealed. The delayed-choice experiment can be interpreted to show that the at time t2that p held at time tl, and it is not true at time t2 that q held at time
first purported truth of tense logic is false. It can be true now that a certain t1.38Since the set of truths that hold at a time is incomplete, there is a need
event occurred at an earlier time, although it was not true at that earlier to study what logic appropriately applies to them.
time that the event occurred then. It was true at that earlier time that the With delayed-choice, it is true at time tl that the particle is in a superpo-
particle was in a superposition then, but it now is true that the particle fol- sition at tl, yet it is not true at t2that the particle is in a superposition at t1.39
lowed a particular path then and was not in a superposition then.36(The With erased measurement, things can be opposite. It is true then (at the
delayed-choice experiment admits this interpretation but does not require time of the earlier measurement) that the particle is in a particular state
then, but it is true now that the particle is (was) in a superposition then. time; it is Einstein's Special Theory of Relativity that motivates making
Delayed-choice can collapse a previous superposition; erased measurement truth relative to a place. (Indeed, because relativity theory treats space and
can reinstate it. time in a unified way, as space-time, it would be unnatural not to extend
Truth is tentative. Before the discoveries of quantum mechanics, the our theory to include spatial location.) Because the velocity of light is finite,
question whether what once was true always will be what once was true, au- an event E can occur at place p at time t, yet no effects of E occur at other
tomatically received an affirmative answer. Now the answer turns out to be places at that same time tor at distant places at some particular (sufficiently
negative, or at least "not necessarily." "Once a truth, always a truth" does close) later times. It takes time for the effects of Es occurrence at p, for the
not necessarily hold. Truth does not necessarily stay fixed. news of its occurrence there, to arrive elsewhere. We therefore cannot ask
For it to be true at time tz that some entity or event E has a property P a t simply whether it is true at t that E has occurred at place p. We also have to
time tl, it must be possible (at t2) for Es having P at tl to affect some Y's ask where that is true. It is true at place p that it is true at t that E occurs at p
having some property Q at t2.If all effects, and all possibilities of effects, are (at time t), but it need not be true at other places r (at time t) that E occurs
erased, it no longer is true at those later times that E had P at tl. But what if at p at t. The news has not reached these other places yet; it has not regis-
exactly those same effects of Es having P at tl also could have been pro- tered at those other places yet; it is not yet fixed and determinate there.
duced by some other thing 2's being R at 6 (without Es being P at tl)?The (Here we must suppose that E at p is not causally determined by some ear-
strongest form of the view under development-the form that I tentatively lier event E' that also registers at the other place r in question, where rat t is
put forward here-holds that (the possibility of) the later effects of Es hav- in the future light cone of E', or else rat t could carry the information that
ing P must not only exist but also allow the unique recoverability of the fact makes it determinate that E occurs at p.)
that E had P at t1.40 It is an empirical question whether truth is timeless, and also whether it
We have discovered as a (possible) consequence of quantum mechanics a is spa~eless.~~There is no necessity about it. The universe could be such that
relativity of truth, and of facts, to time, to the existence of effects at a time.41 truth is neither. And according to current physical theory, it seems that
I do not claim that quantum mechanics entails this, merely that it is a plau- truth holds at a place-time.
sible interpretation of quantum mechanics, a plausible hypothesis about The forward light cone of an event E is a spreading (at the speed of light)
what else holds if quantum mechanics does.42 spatial volume whose events can be causally affected by E (which is at the
The Copenhagen or Aristotelian motivations about determinacy lead to point of the cone). The backward light cone of an event E is the spatially
a theory that makes truth relative to a time. A statement is true at a time converging spatial volume (coming to a point at E) of events that can caus-
only if its truth is fixed and determinate (by facts that hold) then. The pos- ally affect E.
sible relativization of truth to time, depending upon the results of physics, As a first approximation, we might say that the only truths that hold at a
is a surprising result but a digestible one. However, the view (that a state- place at the current time are of two sorts. First, those truths concerning
ment is true at a time if and only if it holds true in virtue of a fact that holds events in the backward light cone of (an event at) that place at the current
at that time) can be extended even further. Since conceptual change is eas- time. In that case, the statement that event E happened at place p at time t
ier to accept one small step at a time, prudence dictates my not immediately can be true at some later place X at a later time t' yet not be true at another
pushing this line of thought to its furthest limit. A consequence that seems later spatially even more distant place Y at that same later time t', because p
like a reductio ad absurdum of a view when it comes in one big leap will be at t is in the backward light cone of X at t' but not of Y at t'. The second
treated seriously if the way to it has been paved by already assimilated steps. sort of truth that holds of a place at the current time concerns those future
Nevertheless, I press on. events that are determined by what holds at that place at the current time.
It seems natural, even if somewhat less intuitive, to extend the theory to It is true here-now that some later event will occur there-later only if fac-
include spatial location. A statement is true at a place only if its truth is tors that hold here-now are sufficient to determine that event there-later.
fixed and determined (by facts that hold) there. It was quantum mechanics It is true here-now that event E occurs there-then only if there-then (or
(or temporal indeterminacy) that motivated viewing truth as relative to a some uniquely recoverable sufficient cause of E at there-then) is in the
INVARIANCES Truth and Relativism 41
backward light cone of here-now (and so could affect here-now) or if there- hold at each and every place-time? Or does contemporary physics tell us
then is in the forward light cone of here-now and is causally determined by that our universe contains very few place-times (Leibnitzian monads) that
factors holding (now) at here-now, or if Eat there-then has some uniquely mirror whatever holds elsewhere? There can be interfering factors between
recoverable sufficient cause C, where C is in the backward light cone of place-times, when the causal laws linking them hold only all things being
here-now and is uniquely recoverable from this spot. That is one restriction. equal; in that case, whether the truths that hold here fix what holds else-
We are not yet quite done with the complications. It is not exactly "caus- where depends upon whether interfering factors (and which ones) actually
ally affect" that is in question. For quantum mechanics is a nonlocal theory. did operate. But apart from this, it seems that what is true locally does not
Events at one place-time can fix, that is determine according to scientific determine the structure of the global space-time.47If the physics we dis-
law (but not a causal law, so they do not causally determine), an event out- cover greatly fragments and localizes the facts that hold, some philosophers
side the first event's forward (or backward) light cone.44An event of mea- may recoil from seeing truth itself as similarly localized. Instead, they might
suring the spin of a particle here-now as +%fixes the spin of a correlated define truth period as what is fixed by the (logical) union of the facts that
particle there-now as -Y, It is true here-now that the correlated particle hold at each of the place-times at those place-times. But that would make
has a spin of -% there-now. So truth is not restricted within the light cone the timelessness (and spacelessness) of truth a derivative and definitional
after all. What is true here-now are not only statements about events light- thing that corresponds to nothing real beyond a disjunction of the scattered
cone-linked (with chains of unique recoverability) to here-now. Non- facts, which nowhere are actually gathered and registered. The apparently
locality can extend the tentacles of truth more widelyj5 What is true here- weighty reasons for holding that truth must be timeless (as well as the intu-
now are statements about events there-then that are fixed by events and itive counterexamples to opposing views) may come to be seen as depen-
facts holding here-now. Generally this will be within here-now's (or its re- dent upon, and derived from, a conceptual framework that itself is under-
coverable cause's) past and future light cone, sometimes more narrowly be- cut by a physics that roots its facts in local space and time.48
cause of erased measurements, sometimes more widely because of nonlocal Where st is a place-time or spatiotemporal region, "It is true at st that p"
fixing of facts. behaves like "It is determined at st that p." Because Aristotle believed the
To take the step to indexing truth to a time (because of the delayed- outcome of a future sea battle might be undetermined by current facts, he
choice and erased-measurement experiments) is one thing. To also index believed that it is not now true that such-and-such a fleet will win the sea
truth to a place, because of considerations from Special Relativity, is an- battle."g
other matter that raises its own special problems. Are there truths at a point An appealing and standardly imposed condition of adequacy on a theory
instant? Truths in a room?46Does which truths hold there depend upon of truth is:
which person is there and what he knows, upon which books are there and
(C) It is true that p if and only if p.50
what sentences it contains? Perhaps there is some way to define the infor-
mation in an object, and so in a space-time volume, and thereby to define A theory that relativizes truth to place-times cannot easily endorse this con-
what holds truth there-then, as fixed by the information that holds there- dition, in its full strength, but it can put forth some consequences of it. One
then and by what holds elsewhere-elsewhen that scientific laws are able to consequence, stated in terms of something being determinate at a place-
(uniquely) connect with the information that obtains there-then. time, would be:
The theory that relativizes truth to place-times is not verificationist. It
does not depend upon what we know or upon what anyone knows. It does (D) It is determinate at place-time st that [it is determinate at place-time
depend upon determination, upon what facts holding at some place-times st that p] if and only if it is determinate at place-time st that p.
fix about other place-times. As a statement about truth, this corresponds to:
And such determinate fixing might turn out to be sparser than we had
assumed, with the consequence that truth is much more localized than we (T) It is true at place-time st that [it is true at place-time st that p] if and
previously thought. Do physical laws permeate the universe so that they only if it is true at place-time st that p.
However, even though we have restricted this statement to one and the Now what about statement (S) itself? If it is true and it applies to itself, then
same space-time, it will be too strong to assert the implication from right to there is some time when it is not true. Provided that time is not now, the
left if there can be enough information at st to make it determinate there time when I am asserting it, there will be no paradox. There will be a mo-
that p yet not enough to make it determinate there that it is determinate ment, call it 6, when it is true that there is a statement, call it M, that is true
there that p.5' If such a situation can obtain, then only the following weaker at all times. But that moment is not now, that moment need not last for
condition can be put forth: long, and at all other times M will be false. The proponent of an omni-
temporal truth does not want there to be just one time when a statement
(Dl) If it is determinate at place-time st that [it is determinate at st that is omnitemporally true but for that to hold (of that particular statement)
p], then it is determinate at st that p.
at all times. This, however, does not follow merely from an argument that
Putting this conditional (but not biconditional) proposition in terms of applies (S) to itself.53
truth, we have: The theory of truth presented here is not a theory about space-time that
itself is outside space-time. Truths outside space-time are a remnant of Pla-
(TI) If it is true at place-time st that [it is true at st that p], then it is true tonism, and, whether or not that structure holds for necessary truths, it is
at st that p. inappropriate for empirical truths. Our theory places truth within space-
Condition (T'), unlike condition (C) above, does not allow the term "true" time.
to be eliminated from the sentential context "It is true that p."52 We might say that truth is weakly spatiotemporal when there is some
But what of statements (Dl) and (TI) themselves; are these also rela- (fully specified) statement that holds true in some place-time but does not
tivized to space-time?And is the theory of truth as relativized to space-time hold true in every place-time.54Truth is weakly spatiotemporal when there
itself also relativized to space-time? Are quantum mechanics, which sug- is at least one truth that doesn't register everywhere and everywhen. This is
gested the relativization of truth to time, and relativity theory, which sug- compatible with there being a particular and quite small spatiotemporal re-
gested its relativization to space, themselves also relativized to space-time? gion where every truth registers. (I add "quite small" because if we let the
More generally, mustn't there be some topmost level stating a truth with- region be large enough-all of space and time-then each truth will regis-
out a temporal index, so that at this topmost level there are unrelativized ter somewhere within that region.) Let us call such a region in which every
truths? Mustn't the very theory we are stating here be stated as a truth that truth registers an omniscient region. All truths that hold anywhere are fixed
is not relativized to time? and determinate there. We might say that truth is strongly spatiotemporal
To say that truth is relativized to time is not to say that there is no truth when there is no omniscient region, no point or small place where all truths
that holds at every time, that is, no truth for which it is true at every time register and hold.
that it is true at every time. Perhaps statements (Dl) and (TI) and quantum Quantum mechanics has led us to maintain that truth is relative to a
mechanics and relativity theory all are such. However, such omnitemporal time. And the considerations that led to this conclusion, when consistently
truth would not be a feature of the notion of truth but instead a feature (if pursued, lead to the further view that truth is relative to a time and place.
it turned out to be so) of the universe as it concerned these particular Truth is relative to spatiotemporal position. Spatiotemporal position is a
truths. And there is no necessity that somewhere up the line there must be surprising and unexpected factor in the context of truth, and all spatio-
some such omnitemporal truth. temporal positions are equally good. So the present view counts as relativ-
Suppose there isn't any omnitemporal truth, at any level up the line of ism about truth. It might aptly be described as the Copenhagen Interpreta-
generality or at any metalevel either. We can state this position as the fol- tion of Truth.
lowing statement (which only might be true; whether it is true depends This Aristotelian-Copenhagen interpretation of truth is not put forward
upon what the universe actually is like): as a necessary truth about truth. I do not claim that truth must be rooted in
space-time. What character truth actually has is an empirical question. But
(S) For every statement p, there is a time ti such that at time ti p is not notice how strong are the conditions that must be satisfied for truth to be
true. space-timeless. For any truth that holds at a place-time st, it must also be
A
C
44 INVARIANCES L Truth and Relativism 45
determinate at all other place-times that this truth holds at st. Let us say purposes. However, that would leave our concern with truth instrumental-
that, in this case, this truth registers at every other place-time. Moreover, for at least originally.
truth to be space-timeless, the fact that this truth registers at each and every Was William James right, then, in saying the truth is what works? We
other place-time must also register at each and every place-time, and so o n might see James as depicting the value of truth, not its nature. Rather than
up the line. And all this must hold, all the way up the line, for every truth hold that truth simply is 'serviceability,' we can construe truth as that prop-
that registers at any place-time. This exhibits the structure that in the case erty, whatever it is, that underlies and explains serviceability. If one property
of shared knowledge among different persons has been termed "common underlies the serviceability of various statements about different subject
knowledge," so let us term such infinitely extended registering, common matters, that property will have to be very general and abstractly stated. The
registering. Truth transcends being rooted in space and time only when various theories of truth-correspondence, coherence, etc.-then would be
there is common registering, throughout all of space and time, of each and explanatory hypotheses, conjectures about the nature of the property that
every local fact. Only then does Truth subsist in its own Platonic realm. underlies and explains serviceability. (And if it turns out that the serviceabil-
ity of different kinds of statement is explained by distinct proper tie^?)^^
The Truth Property In general, it is better to act upon truths than upon falsehoods. We more
often achieve the goals of our action when the beliefs they are predicated
This position that truth is relative to place and time is a far cry from social
upon are true. Not every time, however. We can imagine cases in which a
relativisms wherein truth is relative to a culture or to a sex or gender or to a
belief is false, yet acting upon it happens to dovetail with what is true in
social position. The correspondence theory of truth says that a statement is
the world so as to produce that action's success, that is, the achievement of
true if it corresponds to the facts. Adding references to the places and times
the goal that was aimed at. And sometimes an action that is based upon the
where these facts are fmed and determined does not appear to make any
truth will fail. The best way that a goal might be reached, or even the only
reference to anything social. So (according to the correspondence theory)
possible way, need not be guaranteed to reach it. Nevertheless, as a statisti-
must not what is true for one person be true for all? How can truth vary
cal matter, we more often achieve our goals when we act upon the truth.
from person to person, group to group, or society to society?Won't a state-
Now, consider those statements or propositions that we frequently are
ment that is fully specified (and contains no indexical expressions that refer
successful in acting upon, and also those statements that we usually are un-
to a society or group) be true for every society or group, if it is true for any
successful (or are less frequently successful) in acting upon. What do the
one? From its being true period, doesn't it follow that it is true for everyone?
statements in the first group have in common (that statements in the sec-
How, then, can we even make sense of relativism about truth?
ond group lack)? We can view theories of truth as presenting the property
In The Nature of Rationality, I wrote:
these statements in the first group share, the property that explains why
It seems reasonable to think that our original interest in truth was instru- (more frequently than otherwise) we are successful in acting upon these
mentally based. Truths served us better than falsehoods and better than no statements. Some would hold it is the property of corresponding to the
beliefs at all in coping with the world's dangers and opportunities. Perfectly facts that explains the greater statistical likelihood of success in action, oth-
accurate truth was not necessary, only a belief that was true enough to give ers that it is coherence, and so on through the range of theories of
(more) desirable results when acted upon. What was wanted were "ser- Facts, however, are not merely components of hypotheses about the beliefs
viceable truths," and to be serviceable, a belief need not have been precisely upon which you act to successfully achieve your goals. Facts also are en-
true. Truth, then, would be rather like what Rawls has called a primary good, countered more directly when they thwart your achieving your goals (as
something that is useful for a very wide range of purposes, almost all, and you act upon false beliefs). You sometimes discover facts by bumping up
hence that will be desired and bring benefit (almost) no matter what our against them.
particular purposes might be. So we might desire true beliefs and come to be We need to refine this view of truth. Truth is not the one property that
concerned with truth because true beliefs are useful for a very wide range of the statements in the first group have in common, for there is not just one
such property. All the statements will have the property of being expressible ticular dovetailing goal. The success will be a pinpoint success; the false be-
in under 327 words, for in~tance.~' The truth property not only is common lief is not a good basis for achieving other goals in different situations in
to the statements in the first group, it explains why the statements in the which it also could be acted upon. Its success is not robust.62
first group are (as a statistical matter) more often successfully acted upon.58 It may be wondered whether success in action is even an indication of
Although its instrumentality is what identifies the truth property, we can the belief's truth. It is not just that the belief acted upon may be only ap-
(come to) care about truth (and what has that truth property) intrinsically. proximately true, or approximately true in a certain domain for certain
Not every possible explanation of why some statements can be success- purposes, but enough so to yield success. Navigation in accordance with a
fully acted upon would count as a theory of truth, though. If God (existed geocentric theory leads to successful goal achievement, yet that theory is
and) were constantly intervening to sometimes produce success in action, not true, even approximately. Rather, its consequences are close to the truth,
sometimes not, then what the statements in that first group would share is concerning the particular goals of navigation. So the geocentric theory is in
"being temporarily favored by God as a basis of action:' but this property the set of beliefs that we act upon successfully, yet it is quite false. What is
would not count as a truth property. In such a case success in action would needed, therefore, is a two-stage theory. The success of acting upon some
not depend upon truth at beliefs is explained in terms of their having the truth property. The success
There will be some appropriate general condition that a property must of acting upon other beliefs, which do not have the truth property, is ex-
satisfy for it to qualify as a truth property. (Perhaps that condition is Tar- plained in terms of some of their consequences' closely matching the conse-
ski's condition that S is true if and only if p, where "p" is replaced by a sen- quences of beliefs that themselves do have the truth property. (A full and
tence of the language and "S" is a name of that sentence; or, in the case of knowledgeable explanation would show why the consequences do closely
our theory that truth is rooted in space and time, some weakening of that match.) In both cases, what explains success in action is, ultimately, the
condition, such as (T') above, but this time treated as a condition that must truth property.'j3
hold true of truth, not as one that must in each case be provable via a truth There are, however, cases in which seeing the probability of danger or
definition.) We may understand the statement that truth is socially relative failure as lower than it is-a false belief-helps to sustain someone's action
as claiming that there are alternative and different truth properties (each of or courage, and so raises the probability of success above what it would be
them satisfying that appropriate general condition) which explain different if the person's belief were completely accurate. (Alternatively, overestimat-
people's, or groups', or cultures' success in action. The people not only are ing a danger can lead a person to be more cautious and hence to avoid that
acting upon different beliefs; their success in action is explained by different danger.) Belief affects motivation, which in turn affects the likelihood of
truth properties, according to which different particular propositions are success. And sometimes a belief about the probability of success, or about
true.60 recovery from some disease, initially false, can be made true by whole-
The success in action I speak of is a success in achieving the particular hearted adherence to it. Normally, we think truth is independent of what
goals of the action. A biologist who espoused Mendelian genetics in Stalin- we think about it. But thinking a placebo is effective helps to make it so.
ist Russia might be less successful than his Lysenkoist colleagues in avoiding However, such phenomena are relatively rare. We would have a very differ-
labor camps in the Gulag. Does this mean that a biologist in the Soviet Un- ent notion of truth if, for a vast segment of our beliefs, the act of believing
ion is more likely to succeed acting upon a falsehood? No, a biologist who affected the probability of the belief's holding true."
wishes to achieve the goal of avoiding imprisonment in a labor camp will We have presumed thus far that each and every truth is wholly true. Yet,
be more likely to succeed if he acts upon the truth about what will elicit rather than just flat-out success or total failure in acting upon various be-
such persecution, and a biologist who wishes to raise plants with certain liefs, there can reliably be different degrees of success in action upon dif-
characteristics will be more likely to succeed in this goal if he acts upon the ferent beliefs. And these differences can be counted upon; beliefs have dif-
(Mendelian) truths about genetics6' The times when acting upon a false fering degrees of accuracy. (A general belief may differ in its degree of
belief actually will help achieve a goal involve that belief's dovetailing with accuracy across different situations; its degrees of accuracy may fall along a
a very particular situation, and also the action's being aimed at a very par- gradient.) If the truth property that underlies the differing degrees of suc-
cess in action itself can come in degrees (if, for instance, there can be differ- fit these intrinsic natures, we would be succeeding in describing the world
ent degrees of correspondence to the facts), then we might aptly speak of as it is when undescribed. However, the world does not come already
degrees of truth. Different purposes in action then might require (or make carved up and awaiting these, or any, particular descriptions.
do with) differing degrees of truth. Should we think of the world as it is undescribed as a homogeneous
blob, infinite in all spatial and temporal directions? Any description is an
Is a Theory of Truth Possible? arbitrary division, an arbitrary imposition upon the world's homogeneity.
(But homogeneous according to which descriptive terms?) Such a world,
Various objections have been raised in the literature to the very possibility however, would not lend itself to the differentiating descriptions that we do
of an illuminating theory of truth. It is held that we can use different vocab- offer. Our terms would gain no grip on, or in, that undifferentiated unity.
ularies to divide up the world, different terms to describe the world. Just as Even if the world (as it is undescribed) does not uniquely determine its
a jigsaw puzzle carves up a picture into different shapes, so too a language is correct description, it can be described in a range of ways but not in all pos-
able to carve up the world into various configurations. We use terms like sible ways. And only a range of worlds, not all possible ones, can be de-
"chair," "house," "nation-state," and "river," but some other culture and lan- scribed by the terms and concepts that we do use. The world has the poten-
guage, it is said, might divide things quite differently, for instance having a tiality to be described by some terms and not by others.66Hence, when we
term "zanzar" that refers to anything which is either a chair or a river or do successfully describe it, even though our description is not the unique
half of a house. possible one, this does show us something about the world "as it is undes-
However, even if terms embody arbitrary divisions of the world, it does cribed," namely, that particular undescribed world lends itself to this par-
not follow that the statements or propositions made with these arbitrary ticular description. The description thus describes the (undescribed) world
terms are not true. Define an "ourth" as anything that is either a chair or a in its potentiality.
river or half of a house or is identical with Grover Cleveland. That is, ac- It also is said that truths are statements or sentences, and so are depen-
cording to us, an arbitrary term. Be that as it may, it is true that all zanzars dent upon the existence of a language; therefore, truths did not exist before
are ourths. And it also is true that the object in the corner of my office is a any languages did (somewhere or other in the universe). The question
zanzar, and that Grover Cleveland is an ourth but not a zanzar. The arbi- "What kind of thing is 'true' a property of?" is a vexed one. Each existing
trariness of the constituent terms does not make the truth of these state- answer faces its own formidable problems. (Some hold that the vehicle of
ments an arbitrary matter. We may not be much interested in statements truth is an abstract entity that is not language-based.This entity is termed a
composed of these arbitrary terms, but that lack of interest and that arbi- proposition; there are problems about its identity conditions.) But even if
trariness do not affect whether these statements are true. Statements with the vehicle of truth is a sentence or a statement, this does not show that
gerrymandered terms can be true, and statements with quite natural terms what makes the sentence or statement true, what it corresponds to, also de-
can be false, as is the statement that there are no giraffes in the United pends upon language, and so did not exist before language did.67Some
States right now. would add that there weren't any facts either before the existence of lan-
It also is maintained that the world itself does not have any natural divi- guage, because facts just are languagelike entities that were hypostatized to
sions intrinsic to it. Hence, our terms do not "cut nature at its joints."65It is match language's true sentences. (P. F. Strawson has written that facts and
not clear what or whose theory of truth requires that nature be so cut. Per- true sentences were made for each other, but, as we shall see, we needn't
haps the idea is that, with no intrinsic character to things and no natural suppose that each true sentence has its own separate fact.)
boundaries awaiting description, truths do not correspond to anything ex- It has been held that it is empty and unilluminating and circular to say
isting independently of our descriptions. Thus, some say that we cannot de- that people are successful in achieving their goals (when they act upon a
scribe the world in its character as undescribed. If the world did come di- belief) because their belief is true or because it corresponds to the facts,
vided into objects having intrinsic natures, then apparently these natures since the only reason we have for thinking that the belief is true is that the
would hold apart from our descriptions, and so if our terms did match or goals of the action are achieved. Similarly, it has been held that it is empty
and unilluminating and circular to say that scientific hypotheses and theo- sis is not trivial, for there are alternative explanations. It might be that the
ries yield correct predictions because they are true or because they corre- success occurred by accident this one time. This explanation will be in-
spond to the facts, since the only reason we have for thinking the hypothe- creasingly implausible, though, the greater the number of successful actions
ses and theories are true is that they yield correct predicti~ns.~~ or correct predictions based upon that particular belief or theory. Recall the
Let us linger for a moment over the issue of circularity. Suppose a person earlier objections to the claim that success in action is an indication of
has a heart attack of a certain sort, and this is explained as being caused by a truth. There are ways, the objection held, that false beliefs can lead to
particular kind of structural defect in the heart. Yet the only reason for successful action or prediction. (For instance, a false belief might inspire a
thinking that person did have that structural defect is that he died of that person's confidence, which leads to successful action.) The existence of al-
kind of heart attack. That appears to be an objectionable circle. (Why did ternative hypotheses to explain successful action or correct prediction gives
the bridge collapse? It had a structural defect. Why do you think it had that content to the claim that in a given case (or in most cases) it is the truth of
structural defect? It collapsed.) the belief that is the explanation. If we can check whether these alternative
It would be a circle if we said that event A caused event B, and event B hypotheses do hold, and if we can eliminate them, that will leave the truth
caused event A. Here, the causes go in a circle. It also would be a circle if we of the belief or theory as the most reasonable explanation extant.
said that event A is our reason for believing event B occurred, and that One of the alternative explanations in the case of scientific prediction,
event B is our reason for believing that event A occurred. Here, the reasons though, is not easy to eliminate. One can successfully navigate according to
go in a circle. However, it is not a uniform circle to say that event A caused the geocentric theory, which is false, because its predictions in this particu-
event B, and that our reason for thinking event A occurred is that event B lar domain match or approximate the predictions of the true heliocentric
occurred. Here "causes" goes in one direction, and "is a reason for think- theory. One can successfully make predictions according to Newtonian the-
ing" goes in the other direction. It is not evident why this kind of nonuni- ory, which is false, because its predictions in certain domains match or ap-
form circle is objectionable. proximate the predictions of the true theory of relativity. It always seems to
There might be existing well-confirmed hypotheses that all heart attacks be a possibility that our correct predictions occur, not because the theory
of this type are caused by that specific kind of structural defect, and that all we are using is true, but because in this domain it matches some other true
bridge collapses of this type are caused by that type of structural defect. Ev- theory that has not yet been formulated or dreamt of. This does not mean
ery one of the large number of cases that we have examined in the past fits that we now should abandon our current theory. And it certainly does not
this generalization. Every such heart attack was found, upon independent mean that we should abandon the notion of truth. For this alternative ex-
physical examination, to occur along with that structural defect. We have planation invokes the notion of truth, not for our current theory but for
good reason to believe that all are, and so we have good reason to believe some alternative unspecified one.
that this particular one is, even though we have not (yet) performed an in- We have considered whether there is any bar, in principle, to stating a
dependent examination to discover the existence of the defect in this case. correct theory of truth. It also may be wondered whether we ever do state
It is not circular to explain the heart attack by the existence of the defect, any (completely accurate) truths. Our language could have some false
even though our only reason for believing that the defect did exist in this metaphysical assumptions built into it (assumptions about substances and
case is that the person did have the heart attack. attributes, or whatever). Such assumptions may be only roughly accurate
In the case of the link between belief and truth, however, we do not have but accurate enough to have been built into our cognitive apparatus by the
this kind of direct statistical evidence. We do not establish background gen- processes of evolutionary selection. There may be assumptions that are so
eralizations through an independent ascertainment of the truth of the be- deep that we do not yet notice them, yet everything we say might be in-
lief (or hypothesis) and then discover that its truth is highly correlated with fected with their metaphysical inaccuracy. (Even our current mathemati-
successful action (or successful scientific prediction). cal formalisms, and hence also the mathematical formulae of our physics,
Instead, we have an explanatory hypothesis: what explains success in ac- might embody some such assumptions.) Perhaps, then, we have yet to state
tion (or prediction) is that the belief (or hypothesis) is true. This hypothe- X
one completely accurate truth.69
Euclidean geometry might be a useful way of viewing physical space, those consequence^.^^ This position seems to embody empirical claims: that
evolutionarily instilled but not strictly accurate. Could even the notion of adopting certain concepts will lead to certain social results, that continuing
truth itself be one that is evolutionarily instilled but not strictly applicable? to use other concepts will impede the reaching of certain social results. Pre-
Paul and Patricia Churchland have argued that our ordinary psychologi- sumably, Rorty believes it is true (or a fact) that his proposed conceptual re-
cal notions of belief and desire are components of a theory (folk psychol- forms will lead to those results. Sooner or later, it seems, a proponent of
ogy) that might be defective and nonreparable; these notions might be des- Rorty's position must claim and affirm that certain things will in fact lead
tined to become defunct when a fully accurate psychological theory comes to other things, that this indeed will occur in the world, given the way
along70Is the notion of truth part of (what we might call) folk epistemology, the world is. And if these statements can correspond to the way the world
which conceivably is a defective theory that might also be replaced, along is, why not other statements as well? (Or would Rorty say that these state-
with its component notion of truth? If truth is that property of some beliefs ments do not correspond to the way the world is, but that our accepting
that explains why acting upon those beliefs, in contrast to others, leads to them has certain beneficial consequences? But then, is it a fact that . . . ,
the achievement of the goals of the action, then the theory of truth finds its etc.?)
place as part of the theory of mind. Within the structure of belief-desire Recent postmodernist theorists have excoriated the notions of objective
psychology, it marks truth as the property of acted-upon beliefs that is con- truth, objectivity, and rationality, holding these to be empty and valueless
ducive to the realization of desires. So if future neuroscience were to under- notions, mere masks for privilege and power. The arguments that have been
mine the notions of belief and desire, could the associated notion of truth provided for such strong and startling claims can be countered or shown to
lag far behind? It seems plausible, however, that whatever such a neurosci- be confused or inconclusive, yet this rebuttal does not dispose of the post-
ence will substitute for beliefs, for instance, neural networks of a certain modernist position. Let me explain.
sort or in a certain state, a concept that plays a somewhat similar role to In The Examined Life, I offered the following interpretation of Zen en-
that of truth then will apply to this substitute to somehow connect the neu- lightenment within Buddhist philosophical the01-y.'~The self is a theoretical
ral networks or events to the way the world is. posit, and once its existence is accepted, everything falls into place around
Does the denial of fked philosophical points, of fixed and guaranteed in- it; everything is seen through the filter of that one theoretical piece.
tellectual categories and truths, open the way to chaos? Some postmodern- Through meditative practice (and theoretical reasoning), one comes to re-
ist writers seem to think that if there are no self-evidently compelling axi- alize that the self does not exist (the "no-self doctrine"). Eliminate (the be-
oms or methods that are knowable a priori, from which all truths (or at lief in the existence of) the self, and everything else then falls into a quite
least the vast majority of the ones that we care about) can be established, different pattern and gestalt. (In the familiar drawing of gestalt psychology,
then all truths are relative or somehow nonobjective. If "foundationalism" if you follow the direction to see that part of the drawing not as a women's
fails, then objective truth totters-or so we are told. "No judgment is chin but as a nose, the whole picture looks very different. It is a picture of
or could be objective in the classic sense of justifiable on totally context- something else entirely.) Since the posit of the self organizes every other
transcendent and subject-independent grounds," Barbara Herrnstein Smith piece, it is difficult to argue against it. The other pieces, already in their or-
writes. In this allegedly classical sense, "the only reasons that count as ganized place, create a space that requires the self, which only it can fill. It
'good' ones are those that are certifiably deduced by pure reason from therefore is difficult to make a local argument against the self. An alterna-
universally valid, transcendentally necessary principle^."^^ Charles Sanders tive global picture is needed.
Peirce, who formulated the first explicitly nonfoundationalist view-he Consider a jigsaw puzzle containing ten pieces that do fit together in a
called it "pragmaticismn-certainly did not believe that foundationalism's pattern, yet if you put those pieces together that way, then you cannot fit to-
failure entails relativism about truth. gether the rest of the pieces to complete the puzzle. Those ten pieces do in-
Richard Rorty has claimed that the notions of truth, correspondence to terlock, but their total interlocking shape fits nowhere; you have to disperse
the facts, rationality, and objectivity all stand in the way of (what he holds those pieces and start again in order to successfully complete the puzzle. (It
to be) desirable social consequences; so these should be replaced by other would be fiendish to manufacture a puzzle of this type unannounced.)
concepts (such as solidarity) whose adoption and use will better lead to Might some central philosophical notions, such as objective truth, be im-
pervious to a local counterargument? Other conceptual pieces, once in their could have made a structurally similar argument about costs against-the na-
place-a place they acquired because of this central piece--can be used to scent secularism that challenged it, yet this would not have sufficed to dem-
protect the central piece, or to reinstate it if it gets temporarily displaced. onstrate its superiority.)
These other organized pieces leave a hole that only that one particular cen- Even someone who rejects postmodernist theories can welcome their ex-
tral philosophical piece can fill. Yet they have been organized that way by istence, for they force us to look anew at some fundamental notions-a
the central piece with its particular shape. No wonder the other pieces leave quintessential philosophical task. The challenges and errors of the post-
that shaped a hole. Quine presents an image of some statement not easily modernists have the beneficial effects that John Stuart Mill describes in dis-
being given up because it is central to many other things.74But this central cussing liberty of discussion. Although I have spoken of someone's con-
purported truth might be held in place by other things that it placed into structing and assessing a global postmodernist theory, this seems to be the
their positions. If these other things were positioned differently, they might last thing existing postmodernist theorists want to do. They would claim, in
hold something else in place. Or nothing. justification, that there can be no such global theory or "totalizing narra-
Postmodernism questions central pieces of long-standing philosophical tive," yet an induction from particular past failures (such as Christian es-
views. The apparently compelling local arguments against postmodernism chatology or Marxism) does not establish this claim. In any case, it is so
are not decisive, I suggest, because the local pieces they rest upon have their much easier-and apparently so much more fun-to sabotage the exist-
place and connections to the central pieces because of those very central ing theory. However, the irresponsibility and the intellectual confusion of
pieces that postmodernism puts under question. much postmodernist writing do not entail that the theories and concepts it
A global argument against postmodernism will pit (a cleaned-up version attempts to undermine are secure.
of) the current position against an alternative global postmodernist view
and argue that the former is superior. But is this comparison to be made Is Truth Socially Relative?
and judged by the standards and criteria of the current view or by those of
the postmodernist view that purports to replace the current view? This is- We now can return to our initial question of whether truth is socially rela-
sue is not insurmountable, for the standards of comparison themselves may tive. It is obvious that the very same statements are not serviceable for ev-
turn out not to be controversial, in that the two views may share them. eryone. The statements that serve an engineer may not serve an artist; the
Also, there is no guarantee that a view will count itself superior to another, statements that serve a person in the frozen tundra may not serve someone
even according to its own standards. However, we currently are at an earlier in the tropics. However, the same property underlies the serviceability of
stage of the discussion. The alternative global postmodernist view, the one these different statements, and that one property is what gets demarcated
that rejects currently central philosophical pieces and places many other by a theory of truth-so says the nonrelativist.
pieces in a new pattern, has not yet been built. So we don't know yet what it Yet the general characterization of truth does not require that it be
is that is to be compared. one and the same property that underlies serviceability for every society,
Arguments that say that such an alternative global postmodernist view group, economic class, and sex. The general characterization leaves it open
cannot be constructed also might rest upon local, even if very central, pieces whether two groups can differ in their truth property. It might be dis-
of the current view, and hence be inconclusive. Still, such arguments can tinct properties that make statements serviceable to members of different
show how very great would be the new view's intellectual cost, by showing groups. For each group, truth is what underlies and explains serviceability,
how much (of what currently is central) that new view would have to give but what does this for the different groups is different. What truth is, is fixed
up. (Proponents of the new view might welcome these costs, holding that by its (particular kind of) explanatory role in explaining servi~eability?~ but
the greater the havoc wreaked upon the old view, the better.) However, the situation of groups can be so different that the explanation of ser-
pointing out the unavoidable costs of the new view will not tell us what viceability in their two cases must differ. What truth is, truth's nature, is dif-
might be its great and compensating benefits. We cannot know these until ferent for the different groups. This constitutes a social relativism about
the new view has been constructed and operates for a while to show its own truth.
unexpected consequences. (After all, the once-reigning religious worldview The notion of serviceability is explicitly relational. Something is ser-
INVARIANCES
viceable for some person or for some group. The underlying property that
-'r Truth and Relativism
overlap but are not identical.) The two group's different needs will be dif-
57
explains this relational serviceability need not itself be relational, however. ferentially encompassed by the two differing truth properties, and even for
I
For instance, being digestible by human beings is a relational notion (and

I the statements marking the groups' common needs, the two different truth
what is digestible by animals that digest cellulose is not digestible by us), yet properties provide different (but overlapping) explanations of service-
having certain chemical structures (the property of substances that under- ability.
lies their being digestible by the human digestive apparatus) is a monadic, Some philosophers (leaving issues about groups aside) have held that
not a relational, notion. But even if what always underlies serviceability is a different properties underlie and explain the serviceability of beliefs in such
monadic, nonrelational property, the truth property would be relative if different areas as mathematics, ethics, interpersonal relations, and beliefs
different groups had different nonrelational truth properties. about objects as different as elementary particles and ordinary middle-
We should distinguish three ways in which truth can be relative. First, the sized perceptible physical objects. Could there be a different explanatory
, property for each of these areas, without any one overarching property that
notion of truth can be relative by containing a (perhaps inexplicit) factor
that can vary. Second, the truth property can be relative in that it varies includes each as a subproperty?" (Yet the respective truth properties of
1
from group to group (or it is explicitly relational and refers to groups that statements from different areas must mesh to some extent, since we some-
differ). Particular truths can turn out to be relative too, even if the notion I times successfully reason by combining such truths; for instance, mathe-
of truth itself is not relative and neither is the truth property. Einstein iden- matics is applied to formulate theories of the social domain.) Similarly, two
tified some truths about spatial and temporal intervals as relative (to the ve- groups' truth properties may each be conjunctions that overlap (intension-
locity of an observer) although the notion of truth itself (and the truth ally as well as extensionally) in one conjunct yet differ in their other con-
property) was not (held to be) a relative notion. So the third way in which junct~.One property might aid members in the first group in their inter-
truth can be relative is that every particular truth turns out to be relati~e.'~ personal relations (and hence be a conjunct of their truth property), while
The truth property is socially relative, I said, when what explains ser- I another property might aid members of the second, differently situated so-
viceability for two social groups are two distinct properties. Even if we cial group in their interpersonal relations.
doubt that this is a fully coherent notion, it is one to explore, for it offers so- The correspondence theorist might object that we do not know of two
cial relativism its best chance of being true. If that chance turns out to be plausible proposals about truth properties; there is only one, his, viz. that
small, even on the most favorable conceptual suppositions, that will be in- correspondence to the facts is the truth property. However, the point that
structive. the truth property may not be unique does not require proposing a com-
Surely, though, there will be considerable overlap in the explanatory pletely different theory of truth. For correspondence theorists themselves
properties. The members of each group need to avoid bumping into ob- disagree about how to formulate the correspondence theory of truth and
jects, place food in their mouths, tell day from night, and find their way 1 what exactly the correspondence truth property is. Consider then two plau-
back home. Won't it be the very same property of statements, call it P, that sible proposals about the details of the correspondence theory of truth that
enables them to do this? Won't their supposedly distinct truth properties yield two interestingly different specifications of the truth property, both of
each have to include Pas a subpart or conjunct, and hence won't they over- which fall under the general rubric of "correspondence." Must one of these
lap? However, while the distinct truth properties must apply to some of the proposals be the right one and the other wrong? The difference between the
same statements that P applies to, thereby overlapping in what they apply two (correspondence) truth properties might match the differences in the
to, these truth properties needn't apply in the same way. Their extensions situations of the two groups. Could not each theory correctly state the truth
(what the properties apply to) will overlap but their intensions (their defin- property of a diflerent group? Even a confirmed correspondence theorist,
ing characteristics) need not. The overlap of the extensions of the two truth therefore, might countenance relativism about truth.
properties may be enough to explain why the two group's beliefs are ser- Let us summarize. The general characterization of truth that we offered is
viceable yet, nevertheless, the two truth properties are different. (There are I
relational. Truth is what explains a person's or group's success in action; it is
some things that one applies to that the other one does not; their extensions what is serviceable for the person or group. The truth property, which ex-
plains the relation of serviceability for a person or group, may be a non- Here I have spoken of what property it is rational for each group to use.
relational property that statements can have apart from their relation to a Earlier, however, the truth property was identified by its explanatory role. It
person or Social relativism about truth is the claim that diferent is that property of beliefs that explains the (statistical) success in acting
nonrelational properties P, and P2 are the truth properties for different upon them. It need not be the property that the people are acting on, in the
groups; PI explains the relation "serviceable for group Gl" while P2explains sense of adverting to or consciously taking into account or computing in
"serviceable for G2." Nonrelativism holds that the same nonrelational truth accordance with. Won't that underlying explanatory property be the differ-
property underlies these two relations (serviceable for group G1 and ser- entiated property? Here is where the analogy to different domains is imper-
viceable for group G2). fect. The specification of the one common differentiated truth property
What specific arguments might be offered for the claim that truth is so- (that explains everyone's success in action) would have to refer explicitly to
cially relative? Suppose that the different domains (of ethics, of impersonal people's differentiated social roles and situations. The one common truth
facts, and of people's motivations) do have different truth properties under- property, therefore, would have to be a conjunction, something like the fol-
lying them. So far, no social relativism. But different groups of people (as lowing: PI if you are an entrepreneur, and Pz if you are a hired employee,
delineated by economic class or race or gender or sexual preference or and P3if you are a slave, and P4if you are a priest in an animistic sect, and
whatever) are differently situated in a society; they have different power, P5if you are speaking to someone more powerful than yourself and unable
status, and influence. Because of the groups' different situations, statements to deceive him about your ultimate aims, etc. This does make truth appear
in the different domains are differentially important to their members. Per- nonrelative in its form, because the same truth property applies to every-
haps the powerful people do not have to be very concerned about under- one, but at a cost. All the relativization that a social relativist could desire
standing other people's motivations because they do not need, very often, has been explicitly incorporated into the (conjunctive, socially differenti-
to gain the spontaneous willing cooperation of others. They can shape peo- ated) truth property itself.
ple's behavior by issuing commands or payments, without paying close at- Success in realizing the goals of an action depends upon the goals aimed
tention to the nuances of those people's motivations and concerns. Since at, the action done, and the nature or state of the world. The action a per-
(beliefs from) different domains are important to different groups, the un- son performs depends, in turn, upon his goals, upon his beliefs (about the
derlying property that explains their successful goal-achievement will vary. nature of the world), upon the range of alternative actions that might be
Since the different groups will have different truth properties, truth is rela- done, and upon the decision rule he uses to select a particular action (given
tive to the group. So runs one argument for the relativity of truth. those goals, beliefs, and possible actions). Hence a person's success in realiz-
To this it might be objected that everyone should use a differentiated ing his actions' goals depends upon all of these factors.
truth property, and the very same one, of this form: truth property PI in do- If truth is that property of beliefs that explains the successful achieve-
main Dl, property Pz in domain 4 , . . . ,property P, in domain D,. Rather ment of goals, then (in principle) truth might vary with, and be relative to,
than each person's using the one truth property across the board that serves the nature of the world (as the correspondence theory holds) and also to
him best in the domain that is most important to him (given his social po- the particular kind of goals the person pursues. (For it has not been ex-
sition), every person should use the same more differentiated property. It cluded that different properties of belief would explain success in achieving
will serve everyone well across all the different domains. However, there are different kinds of goals.) Moreover, and more surprising, truth might vary
cognitive and computational costs (and costs in effort) in using a differenti- with, and be relative to, the decision rule the person uses to choose among
ated truth property. It may not be worthwhile for members of particular actions. For it also has not been excluded that it would be different proper-
groups to incur those costs. Cost-benefit analysis might show that it is more ties of belief that explain success in achieving goals under different decision
rational for each person to use a rule of thumb utilizing one truth prop- rules. One property of belief might fit the decision rule of maximizing ex-
erty-which one that is depends upon which group the person is in. In do- pected utility, a second property might fit the decision rule of satisficing,
ing this, the person will not often go far astray. To be sure, an unlimited be- while a third property of belief might fit the decision rule of minima~ing.'~
ing for whom there are no such costs might use a perfectly differentiated Might the truth property also be relative to how rational people are, or to
truth property. But does this show that relativism is false or that it is true? the way in which they are rational? Would some completely irrational deci-
sion rule give rise to a different truth property? Should we demarcate the or sexual preference. I want to say something even stronger. None of them
range of rational decision rules by saying these are the ones that give rise to area8*
some particular truth property, such as corresponden~e?~~ I do worry that I Yet I do have to note the possibility that similar situations might not con-
may have followed out the implications of a roughly accurate view of verge to the very same truth property. Similar situations might lead merely
truth-truth is what explains success in acting upon beliefs-to the point to similar truth properties. In that case, truth properties would have a linear
where its lack of perfect fit begins to show. Is the possible relativity of truth relation to situations, and this would constitute a social relativism about
to a decision rule a new and interesting discovery, or is it a reductio of tak- truth, albeit a mild one. But if some social relativism does result with a
ing the starting point too literally? well-behaved linear relation of truth property to situation, the situation is
A relativity of truth also might bring in its wake a relativity of logic. Valid more striking if the relation of truth property to situation is nonlinear. Very
rules of logical inference are identified as those that preserve truth. Differ- similar situations then would have very different truth properties. Relativ-
ent truth properties might be preserved by different rules. If truth is relative ism then indeed would be, in the technical sense, chaotic. Across the actual
and can vary from group to group, then there is the possibility that the range of human situations, it is an empirical question whether the relation
rules of logical inference also will vary and so be socially relative. (A relativ- of truth property to situation is converging, linear, or nonlinearly chaotic.
ity of logical rules is not entailed by the relativity of truth, however. Even if My own bet, or guess, is for c~nvergence.~~
truth differed from group to group, these different truths could be pre- I seem to be left with putting forth nothing more than a belief or a guess
served by the very same rules.) that truth is not socially relative. Nonetheless, since I am now saying that
The absolutist is claiming, on our construal, that people are similar the commonalities among people and their situation are sufficiently great
enough in their nature, their environments, their human situation, and that truth is not relative to culture or social class or gender, etc., how can I
their mode of decision that the very same truth property underlies the suc- also have said, near the beginning of this chapter, that it is highly plausible
cessful actions of them all. The absolutist believes in the unity of human- that truth is relative? Phrasing the issue in terms of the unity of human na-
kind. The relativist, on the other hand, on our construal, claims that the ture, and of humankind, leads to a further widening of purview in consid-
commonalities among people and their situations are too weak for all peo- ering the relativist position. The relativist about truth claims that the truth
ple to share a common truth-predicate. The differences in the nature of property can vary among different groups. Different specifications of the
people, or their social positions, or their environments, or their decision groups yield different relativist claims.
rules, are so great, he says, that different properties of belief are needed to There are four versions of the claim of Actual Relativism on Earth.
explain why these different people tend to be successful in acting.81 Among past, present, or future beings on Earth, some two have different
The relativist, therefore, is making a very strong claim, and faces the dif- truth properties. There is the wider claim of Actual Relativism in the Uni-
ficult explanatory task of coming up with several different specific proper- verse. Among (past, present, and future) actually existing beings anywhere
ties that do explain the successful actions of different (or differently situ- in the universe who have beliefs, some two have different truth properties.
ated) persons. The absolutist, on the other hand, needs to formulate only There is the claim of Possible Relativism on Earth. This comes in two ver-
one theory, albeit one of wider scope. I myself find the social-relativist sions. First, some actual beings elsewhere in the universe, if they were on
claim about truth-that truth is relative to culture, or social class, or gen- Earth, would differ from the Earth's human inhabitants in the truth prop-
der, or sexual preference-highly implausible. The commonalities among erties that explained their successful actions here. Second, some possi-
people and the unities of human nature and the human situation are suf- ble beings could exist on Earth with a different truth property here from
ficiently great, I think, to make people's truth property identical. The social- that of actually existing people. The Earth's environment is not powerful
relativist claim that we are denying is a coherent claim, however. When we enough to constrain every organism here that is capable of belief to the
deny this relativist claim, we exclude a coherent possibility and thereby say same truth property. There is the still wider claim of Possible Relativism in
something with empirical content. The truth property is not socially rela- the Universe.There are possible beings, capable of existing in the actual uni-
tive. Not all particular truths are relative to culture or social class or gender verse (and capable of having beliefs), who would have different truth prop-
erties. The actual universe is not so uniformly ecologically constraining ev- groups with differing truth properties to enable them to resolve the partic-
erywhere (including in interstellar space and in black holes) as to restrict all ular questions they are concerned with.
possible beings to the very same truth property, wherever these beings If (as I believe) actual human beings have enough in common to give
might be." Finally, there is the claim of Relativism in Possible Universes. them the same truth property, we can ask what the specific basis of our
Among the different possible universes (one of which is the actual one) and common truth property is. What specifically is it that people share that suf-
the different possible beings capable of existing in them (who are capable fices to make truth nonrelative among them? What is it about the combina-
of having beliefs), some two beings (either in the same universe or in differ- tion of human nature and the human physical and social environment that
ent universes) differ from each other in their truth properties. So at least makes the same property (of beliefs) explain people's successful actions?
one also differs from us. (And might we or beings very similar to us have a And how contingent is this basis, and therefore our truth property; how
different truth property in the very different environment of some other sensitive to variation in this basis is our particular truth property?85
possible universe?) If the answer to the question "Is truth relative?" turns out to be "yes"
Now, I am as absolutist about truth as the next person-or so I had when we take the very widest view across possible beings and possible envi-
thought until I considered these different relativist positions. Which ones ronments, then the interesting question becomes "How relative is truth?"
does the absolutist mean to contest? The first two relativisms about actual How wide is the group that shares the same truth property with us, and in
beings on earth, surely, the absolutist will deny, but after that things begin how wide a range of environments does that sharing continue?86
to be cloudy. By the time we reach science-fiction cases about other beings Scientific knowledge changes people's possibilities of action; it provides
in this universe, not to mention other possible universes, who would con- more efficient means for achieving goals, and it opens new alternatives.
fidently deny or care about denying those relativist positions? There is abso- In every human culture, acting upon the accepted results of scientific in-
lutism enough (for now) if the truth property does not vary among (actual) quiry will lead to more successful achievement of the goals of particular
human beings existing thus far and in the foreseeable future; future en- actions undertaken. (This does not mean that there cannot be serious
counters with extraterrestrials might press other questions upon us. Never- and significant unintended consequences of these actions.) We might see
theless, if one of the latter positions on the list did hold, that would show the spread of science then, its results and its consequent practice, as a
that truth is, in an interesting theoretical sense, relative. Relativism would force driving us, if we do not already share it, to the same truth property.
be a philosophical truth, though not a practical concern. So perhaps our Far from being relative truths, scientific results tend to make everyone's
(philosophical) question should be, not whether truth is relative, but in truth property the same across cultures.87In this sense, science unifies hu-
what way truth is relative. How relative is it? manity.
To say that truth is relative is an incomplete specification. To answer the Our investigation of the concepts of absolute and relative truth has en-
question of whether all truth is relative, we must ask relative between whom? countered, not a dichotomy but a gradation. The factor that a truth is rela-
It is not relative (I believe) between myself and any other contemporary hu- tive to may be wider or narrower. Is the truth relative to larger or smaller re-
man being. (It is not relative to any factor in which we differ.) It may well gions of space-time?Is it relative to characteristics shared by all groups of
be relative between human beings and other existing or possible beings in rational creatures, or only to characteristics had by some groups, and how
the universe. Even when there is such relativity, still, for any particular wide a range of groups is this? The narrower the extent of the factor that a
question at issue between two groups, we can ask whether that question can truth is relative to, the more relative the truth. The wider the extent, the less
be resolved on the basis of statements that the (different) truth properties relative. If the truth is relative to characteristics shared by almost every pos-
of the two groups lead them both to accept. Just as scientific observa- sible region and almost every possible group of individuals, then, although
tion need not be theory-unladen, but merely neutral enough or acceptable it is not wholly absolute, it comes close; it is minimally relative. At the limit,
enough for the proponents of whatever theories actually are under consid- one tiny step more, it is absolute. It is an empirical question how extensive
eration to be able to resolve the issues among those theories, so too state- the characteristic is that a particular fact or truth varies with, and so it is an
ments need not be (absolutely) nonrelative, merely acceptable enough to empirical question how relative a particular fact or truth is. On the dimen-
sion that runs from the (wholly) absolute to the (completely) relative, many I accept that as true, for (3) has property P,, the property that underlies the
truths fall in between. serviceability of my group's statements. However, someone else from an-
other group need not accept (3) as true, for statement (3) itself may not
have the property that underlies serviceability for the members of his group,
Does Relativism Undercut Itself? and hence may not have the property that for them constitutes truth.
We have been led not only to hold that relativism about truth is coherent To the standard question of whether the relativist consistently can state
but to think that truth (probably) is relative to spatiotemporal region and, his relativist doctrine, the answer seems to be yes. However-the relativist
when the purview goes beyond actual human beings, that it also is relative will be quick to note-it does not follow that his doctrine, consistently
to a group in an environment. stated, will be accepted by every group, or even be true for every group.
Relativism about all truth faces a question of its own coherence that That depends upon what their truth property is, and whether it applies to
more particular relativisms do not, for the statement of a particular relativ- statement (2). Thus we have the possibility that one group affirms relativ-
ism (e.g., ethical relativism) need not itself be a statement within that par- ism as true and another denies it, where each group is not merely uttering
ticular domain (e.g., ethics) that is claimed to be relative. It may be a state- the truth as they see it but asserting what is true according to their truth
ment about the domain but not be within it, and so not even appear to property.
apply to itself. Does the absolutist contradict the relativist? He does deny that truth var-
How can relativism about truth, of the sort we have been considering, be ies with some factor (of type) E and this the relativist affirms. Someone
consistently stated? When I am in group G1 and I say may wonder, however, whether these two are talking about the same thing
when they both speak of "truth," since they have different truth properties.
(1) P2explains serviceability for the members of another group G2, Notice, first, that it is not given that their truth properties are different. The
relativist is claiming that there are different truth properties, but the one he
if this statement is true, what truth property am I using, the one incorpo- himself is using need be no different from the one a particular absolutist
rating P,, which underlies the serviceability of my group's statements, or the uses.88In that case they are talking about the same notion of truth, and the
one incorporating P2,which underlies the serviceability of theirs? claim of one of them, in terms of that notion, will be false. But even if they
If I, a member of group G1, am putting forth statement (I), then if it is are using different truth properties, still, as they speak, they both can be us-
true (relative to me), it must have property P,. But that doesn't mean that ing the same general notion of truth, namely, "that property, whatever it is,
G2's statements that are spoken of also must have property P,, or that I that underlies people's successfully acting on certain of their beliefs." (And
need to accept P2as the property adequately underlying the serviceability of both of their general notions also can include the requirement that the
my beliefs. truth property satisfy (some appropriate weakening of) the Tarski condi-
When I put forward statement (I), or put forward the more general tion that S is true if and only if p.)
statement The (weak) absolutist denies that all truths are relative, but he may grant
For different groups, different properties underlie and explain the that some truths are. What reason is there to grant that there is such a cate-
(2)
serviceability of their beliefs, gory of relative truths? Once it is known what factor F a particular truth p is
relative to, cannot we formulate the nonrelative truth "p holds relative to E"
I am claiming that both statements (1) and (2) are true, and hence that and doesn't this (as C. I. Lewis suggested) obviate the need for any relative
both do have my truth property, which, whether or not I know this, is PI. I truths at all? Cannot, and should not, the relativity always go explicitly
need not be claiming that the members of every other group, with the dif- within the truth, which then itself is nonrelative; why need it lurk outside
fering truth properties underlying serviceability for them, must recognize the truth to cloud it?
(1) and (2) as true. Consider now the statement There is this reason to continue to speak of relative truths. We may not
know how to state a truth as absolute because we do not yet know a
(3) Statement (1) has property P,. ground-floor absolute framework relative to which it holds, one that explic-
itly mentions all the factors with which truth could vary. Nevertheless, But an accurate representation is not the same as a true and "more
someone might maintain that the notion of relative truth is secondary be- accurate" does not necessarily mean better. The greater accuracy might
cause it is eliminable by reference to the existence of an absolute framework, make the representation too confusing or too difficult to work with, as does
whether or not we know what that framework is, as follows: a map with too many details. What is wanted is something accurate enough
for the purposes. (Is "true" just being accurate enough for most normal
Statement S is relatively true if and only if there exists a framework F, such purposes, and as purposes shift over time, or vary from group to group, can
that F is a ground-floor framework, and "S is true relative to F" is absolutely truth also then shift or vary?) Belief is an intermediate mechanism statisti-
true. cally connected to inclusive fitness.* Yet when combined with certain goals,
The notion of relative truth would be eliminable because all statements us- such as a desire to commit suicide immediately, accurate beliefs might ill-
ing it would be replaceable in context by statements using only the notion serve inclusive fitness. Other organisms might have different intermediate
of absolute truth. However, not only are we not yet in a position to specify mechanisms that do not involve separate beliefs at all, but rather some enti-
what that absolute framework is; we are not yet in a position to say that ties that (we would say) join beliefs and desires in certain combinations
there is such an absolute framework. Must there, after all, be a ground floor? only, and so leave these organisms less vulnerable to accurate beliefs in the
Could not there be an infinite number of deeper and deeper frameworks? service of harmful desires.
(Compare the question of whether there must be a deepest physical theory, If truth is tied to beliefs and representations, and to their function in
or whether there could not be an infinite number of levels of depth.) guiding action, then we can consider alternative functional substitutes that
Is it necessary to go all the way to a ground-floor framework, though? also guide behavior but do not involve beliefs or representations or some
Once the factor F relative to which p is true is known, cannot we state, not analog of the notion of truth. Our own mode of cognition and of action
just that p is relatively true, but that "p is true relative to F' is itself true? Yes, might be treated in a wider comparative theory of different modes of cog-
that last statement will itself be true, but the question is: will it be a relative nition, behavioral guidance, and reproductive survival. Such a theory also
truth, holding only relative to still further factors, or will it be an absolute might make truth appear somewhat parochial, not (as relativism does) by
truth, for whose own truth no further factors need be stated or invoked or saying that truth varies from person to person, or group to group, or place
depended upon? The person who means to replace all (purportedly) rela- to place, but rather by saying that truth appears only in certain contexts,
tive truths by nonrelative ones that make explicit the relative factor F needs those that are belief- or representation-guided; in other contexts, truth does
the statement that mentions the factor F to be true without itself depending not even get a grip. Such a view, too, would make truth more limited than
upon some further different factor. That is a very large task. In the mean- the absolutist believes, not by making it relative but by making it local. This
time, I want to be able to say that it is true that I am now typing at the com- view of truth's locality would gain force if combined with the claim that the
puter. There should be some notion of truth that enables me to do this other modes of guidance are equally conducive to survival and reproduc-
without being committed to any particular ground-floor absolute frame- tion, and also equally "good," however that is to be spelled out. (It would
work of factors, beyond which truth cannot vary, and even without being beg the question, though, to say that since these latter modes do not involve
committed to the existence of some such ground floor or other. The notion truth and so do not involve the special kind of knowledge that something is
of relative truth fits the bill. the case, they cannot be equally good.) Truth may be local, not just if it is
Relativism about truth seems to make it parochial and limited. ("That is relative but if it gets a grip only within certain action-guiding structures,
just your truth.'' Even when the "you" is quite widespread, the truth's limits the ones that involve beliefs or representations.
still are discernible.) Someone might argue for correspondence as the uni-
versal truth property as follows: the truth property is what explains success T h e Correspondence Theory
in acting upon beliefs; if a belief is a representation of the world that guides
our actions, then surely any being acting upon a belief will do better acting There is a form for defining truth common to many specific definition^.^' S
upon an accurate representation-and that constitutes correspondence. is true if and only if there is a p such that S says (only) that p, and p. Apart
&
from technical difficulties, this form leaves much to be desired. The first the universe were different, if some other universe were the actual one. It
part, that there is something that S says, that what S says is (that) p, seems would not characterize what truth is in every possible world. Perhaps a
to be something to be illuminated by a theory of language. How it is that longer list could give us, in extension, a delineation of truth in each possible
language can state something about the world is an intricate question, but world. But what the philosopher wants, it seems, is a general characteriza-
it is not the question of t r ~ t h . 9SO,
~ the piece of the definition of truth (of tion of truth in any possible world, an illuminating statement of what all
the above form) that seems especially to concern truth is the part that fol- truths have in common in virtue of being truths.
lows "there is a p such that S says (only) that p." Here one might expect Suppose the physicists' hopes of a final unified theory (of everything)
something like "and p is a fact:' but instead we simply get "and p," the mere were realized. All facts are explained by, and follow from, one fundamental
repetition of p. So much the better, hold proponents of the redundancy and law L,. Then we can say that S is true if and only if S follows from L,. (L,
disquotational theories of truth. Yet we want something more. If the above also is one of the things that follow from L,.) This is a general characteriza-
form indeed could serve as a definition of truth, then what we want is not tion of truth, not simply a list, and it would give us a deep characterization
(simply) a definition of truth. We do not want a completion of of truth in the actual world. However, since there could be other universes
built around other fundamental laws or sets of them, this characterization
S is true if and only if , would not delimit truth for those other universes. We might say: it would
where the right side includes "and p," when S is a term of the metalanguage be a theory of what is true but not a theory of truth.
that refers to the declarative sentence substituted for p. Rather, we want to A theory of truth is a general characterization of truths in any possible
understand what it means to say that p holds, for any declarative p. We seem world. If we had a deep theory of what is (actually) true, we might be less
to want a completion of interested in a characterization of truth for any possible world. That latter
task might strike us as like wondering how many angels can dance on the
p if and only if , head of a pin. Those things-angels, other universes--don't exist, and the
a completion that provides something more illuminating than the trivial "p subjunctive questions (if angels did exist, how many could . . . ; if another
if and only if p." And here we might expect that p will be analyzed into universe were actual, what would truth be there?) might seem less than
components (such as objects, properties, relations), and that we will be told pressing, and their answers less than illuminating.
what relations or combinations these components must stand in for p, in Consider, as an analogy, the question of what constitutes a good wine.
general, to hold. Someone might be content with a list, drawn from an encyclopedia of wine.
We seem to want a general characterization of t r ~ t h . 9Not
~ every general Xis a good wine if and only if X is from this vineyard in the year 1983 or X
characterization will do, however. Consider the coherence theory of truth. is from that vineyard in the year . . . etc. Suppose the disjunction covers all
There seems to be no reason why our universe should be the most coherent vineyards through the last year (after which year all the vines are destroyed
one, no reason why all actual truths should be members of the most coher- and there are no further wines made from grapes). We would have a list of
ent set of statements. It seems possible that another universe might be more good wines, and so have defined "good wine" in extension. But we would
coherent than the actual one, and also that sufficiently imaginative and not have a characterization of good wines, a specification of what they all
skilled science-fiction writers might describe such a more coherent uni- have in common. Suppose, instead, that someone does produce such a gen-
verse. Suppose it were the case, though, that the actual universe is in fact the eral specification, listing the properties of good wines. Now we know what
most coherent possible universe, according to some criterion of coherence. makes something a good wine. Yet it might be objected that those things
All the actual facts are explainable by one ultimate law, etc. Suppose, fur- would not be good wines if the atmosphere of the planet were different, or
ther, that it is an accident, a brute fact, that the actual universe is the most if the laws of chemistry were different, or if differently constituted organ-
coherent possible one; it could have been otherwise. Then the coherence isms were drinking them. What is wanted, says this objector, is a general
theory of truth would give us a characterization-and not just a list-of the characterization of "good wine" for each and every possible world. It is not
set of all (actual) truths. But it would not delineate what truth would be if so clear that we would desire this, and such a thing might have to be so gen-
r-
I
era1 and abstract (applying as it must to every possible world) as to be true. Whatever facts make p true and make q true make p-and-q true. There
unilluminating. needn't be an additional conjunctive fact, p-and-q, which makes the state-
We can distinguish the breadth of a general characterization, the range of ment that p and q true.
possible worlds in which it applies, from the depth of a characterization, The truths of molecular propositional calculus statements are built up
how much it tells us, and explains, about the actual things of that kind. And from the truth values of their component propositional parts. The truths of
there might be a tradeoff between depth and breadth. Still, we might want the predicate calculus, Tarski taught us, are built from the satisfactions of
more than an account for the actual world; we might also want an account their component nonmolecular open statements. The truth of a definite de-
for worlds very similar to the actual one. Each trait, depth and breadth, is a scription statement, for instance, "The man in the corner has property
virtue, and different purposes might lead us to seek different types of ac- Bertrand Russell held, is built from the satisfaction of its components. If
counts. some person has the property P, and that person is the man in the corner
I have said that we want to know what, in general, must hold if (any) p. (which Russell analyzes, within the scope of the existential quantifier, as: x
One answer would be provided by a final theory of the one ultimate law L,: is a man in the corner, and for all y, if y is a man in the corner then y is iden-
p if and only if p follows from L,. This does tell us when p (actually) holds. tical with x ) , there need be no additional fact that the man in the corner has
However, we might want, not simply a characterization of when p holds, the property I? That statement is made true by the previous facts, without
even one that applies to other possible worlds, but a microtheory of when p the existence of any additional fact. A true statement is made true by facts
holds, of what makes p hold. This would be a theory that looks within the (or by their absence), yet there need not be any separate fact that it (alone)
structure of p itself, and delineates how p's holding depends upon the com- states.
ponents of p. Other relations, in addition to the ones listed above, are components of
One sort of theory (though perhaps one that is not "micro" enough) the "makes-true" relation. Philosophers of science have described scientific
speaks of facts. What must hold if p, is that p corresponds to a fact. The cor- reduction: one theory T, is reduced to another theory TI when (in the sim-
respondence theorist holds that the truth property, or more modestly our plest case) all the laws of T2can be derived from the laws of TI along with
truth property, is correspondence to the facts. But then we want a theory of linkages (equivalences) of terms in Tzto terms in TI.Thus thermodynamics
what components make up a fact, and how they do this, and also a clear is reducible to statistical mechanics, when temperature is identified with
statement of what the relation of correspondence is. Such an approach the mean kinetic energy of molecules. A statement that the temperature in
raises familiar and troublesome questions: are there negative facts, what are a room is such-and-such is made true by a fact about the mean kinetic en-
facts composed of, do they exist alongside ordinary entities in the world, ergy of the molecules in the room. There is no additional or separate fact
and so about temperature that makes this statement true.97
In a criticism of J. L. Austin's presentation of a correspondence theory of Similarly, truths of constitution might enter as components of the
truth, P. F. Strawson tells us that facts are entities cooked up to be precisely makes-true relation. When one object 0 is composed of entities 01, . . . ,on
what true statements state. For every true statement, a corresponding fact in certain relations, then it is truths about o,, . . . , o, in these relations that
is reified. However, facts and true statements are not, as Strawson com- constitute and make true the statements about 0. Correct philosophical
plained, made for each other. Not every true statement requires a distinc- analyses also will constitute components of the makes-true relation. If S
tive fact as the one it states; different true statements can state and corre- knows that p is analyzed as p's being true, S's believing that p, and (for in-
spond to the same fact.95 stance) S's belief that p tracking the truth that p, then this will enter into the
The relation of a true statement to a fact or set of facts is a complicated truth definition, specifying what makes true someone's knowing some-
one. Not every separate true statement states a separate fact. Nonatomic thing. If X at time t, is the same object as Y at earlier time t,, if and only if X
truth-functional statements in the propositional calculus are made true by is a continuer of Y, X continues Y closely enough to be it at the later time,
the truth values of their component statements. If no facts make p true, and X is Y's closest continuer at time t,, then this will enter into the truth
then not-p is true.96Whatever facts make p true or make q true make p-or-q definition, specifying what makes true something's being the same object as
an earlier The truth of someone's knowing something, or of one ob- satisfaction conditions.lo1By introducing the makes-true relation, which
ject's being the same as an earlier one, is asymmetrically dependent upon adds clauses to Tarski's recursive machinery, we have reduced the number
the analyzing clauses. The analyses of the analytic philosopher then might of predicates (and so derivatively of truths) that must simply be listed.
be seen as increasing understanding not only of the notion that is analyzed Some of Tarski's atomic truths will not be absolutely subvenient. Still, the
but also of the notion of truth, in that it specifies particular components of definition of truth in world wi continues to contain a list, the list of the ab-
the makes-true relation. solutely subvenient predicates (and so derivatively of truths) of wi. This
Another component of the makes-true relation might be specific forms surd quality is eased somewhat by the fact that to obtain a definition of true
of supervenience, wherein one property or fact holds in virtue of others (period), we add that wi is the actual world. We shall have to see later
that it is completely derivative from.99Some philosophers have held that whether this reference to actuality suffices to provide the further common
ethical facts supervene upon nonethical facts, and that mental facts super- quality of truths that critics of the simple list desire.
vene upon neurophysiological and more broadly physical facts. When one How is it that Tarski himself has managed to avoid speaking of actuality?
fact is supervenient upon another, rather than two facts' being present, it He does not offer a definition of S's being true in world wi, and then add
seems that there are two true statements, but only one (underlying) fact. that w; is the actual world. He simply proceeds to offer a definition of Ss
The one fact makes both statements true.loO being true (simpliciter). He is not concerned with modality and sees no
The makes-true relation itself can be construed as a supervenience rela- need to speak explicitly of possible worlds or any functionally similar no-
tion. By utilizing it, or its clauses, a definition of truth shows how some tion. He simply states the truth conditions for atomic sentences within the
truths are supervenient upon others. The truths in a world wi that are not actual world, and he is able to do this, without mentioning actuality, simply
made true by any other truths in that world, and so are not supervenient by speaking within the actual world. The Tarski procedure can be repeated
upon any other truths in that world, we might term absolutely subvenient in in any possible world to define truth there (and we also have added clauses
that world wi. to his recursive procedure). He offers a particular instance of it, specific to
Not only can the truths vary from one possible world to another; the the actual world. But since he has not explicitly specified the application of
makes-true relation can also vary from one world to another. Strictly, we his general procedure to the actual world, since he has not explicitly demar-
should subscript a makes-true relation, indicating the world wi in which it cated the notion of actuality but merely has relied upon the fact that (as he
holds. It seems that two worlds could have the same set of first-level truths assumes) he is speaking within the actual world, he has not explained the
yet differ in their makes-true relation, and hence in the way they are con- notion of actuality and hence has not explained the notion of truth (sim-
stituted. pliciter), even if he has succeeded extensionally in marking out all the actual
A definition of truth for a world w; will divide into two general parts. truths.'02
First, it will list (or do the equivalent of listing) the subvenient truths of There is no reason to believe that we already know all the components of
world wi. This is the base clause of the definition of truth for that world. the makes-true relation, all the different ways (in the actual world) in which
Second, it will (recursively)specify all other truths of wi as made true by wi's a statement can be made true. (This relation of being made true is an asym-
subvenient truths, in accordance with w;s makes-true relation. metrical one; a statement is true because of what it is made true by.) We are
This will give us a definition of: S is true in world wi. We can advance to a not yet able to state all the relations that facts stand in to the true state-
definition of Ss being true (period) by adding that one of the worlds in ments that they make true. Hence, we cannot now put forward a substan-
which S is true is the actual world. (S is true if and only if there is a world wi tive and complete correspondence theory of truth.'03
such that S is true in wi and w, is the actual world.) A theory of truth, then, There is another reason why we cannot yet state a substantive and com-
needs to specify what it is for a world to be the actual world. I turn to this in plete correspondence theory of truth. We do not yet know what the compo-
Chapter 3. nents of facts are. Are there ultimate components (absolute atoms)? There
It has been complained against Alfred Tarski that in the base clause of his is a temptation to say that the components are things or objects, their prop-
definition of truth he simply lists the "atomic" predicates and gives their erties and their relations. But this accords too well with grammatical cate-
74 INVARIANCES
gories, and so it raises the suspicion that linguistic units are being projected
as ontological categories. Of course, another possibility is that these are ac-
curate ontological units, and humanity very early got hold of them in its
formation of corresponding linguistic categories. But how did we get so
smart so early? One might grant that humanity got hold of a division that Invariance and Objectivity
served it well enough, but that is no reason to think such categories will oc-
cur in our eventual best explanatory theory. What will our theories end up
with as the components of those facts which stand in the makes-true rela-
tion to all true statements?Will these be particles, strings, fields, space-time
Are there objective truths and facts? If so, are there also subjective facts and
manifolds, mental states, all of the above, or none of them but something as
truths, or is the adjective "objective" redundant when applied to facts? Can
yet undreamt of? We are not yet in a position to say. Relativism does not
beliefs be objective? Is objectivity in beliefs always a virtue? (In the present
follow, however, from this lack of final ontological knowledge.
climate of opinion, one has to ask whether it ever is a virtue.) Is objectiv-
Although we cannot yet specify the makes-true relation, and although we
ity just a male trait? Worse, is it (in the current phrase) merely a "white,
cannot yet specify the components of the facts marked by this relation, nev-
heterosexual, male" trait? (The complete contemporary locution is "dead,
ertheless we can hold that something about the world makes true state-
white, heterosexual males." What I really object to is the "dead" part. Leav-
ments true. This is, I grant, a weaker statement than I would wish to put
ing aside the vapidity of assuming that the only statements worth attending
forth. It is merely the doubly existentially quantified statement that there
to are very recent ones, it is not nice to pick on people when they cannot
exists something about the world, and there exists some specification of the
fight back.)
makes-true relation, so that a statement is true when that thing about the
The notions of "objective" and "subjective" are contrasting notions, at
world stands in that specified relation to the statement. Weak as it is, how-
least insofar as objects and subjects themselves contrast. Something is ob-
ever, some people seem to want to deny it.
jective when (or to the extent that) it is determined in its character by the
We do not have to say, though, that this something which makes state-
features of an object; it is subjective when it is determined in its character
ments true is facts. Perhaps our explanations need not halt at the level of
by states such as consciousness, emotions, and desires that are intrinsic to
facts. It is not clear whether facts are an additional ontological category that
being a subject.
we must countenance, or whether they too are reducible to some combina-
The objectivity of a fact or truth depends upon its own content and
tion of the ultimate ontological components.lo4
character; the objectivity of a belief depends upon the character of the pro-
To know the correct and deep theory of truth's nature requires far more
cess that gives rise to the belief. It depends upon what kind of factors the
than the mere ability to state particular truths. It requires a knowledge of
belief is based upon, and how. The objectivity or subjectivity of a belief can
the ultimate dependence relations, and of the ultimate explanatory and
crosscut the objectivity or subjectivity of the fact believed; all four combi-
ontological factors. A theory of truth, therefore, arises closer to the end of
nations are possible.' I begin with a discussion of objective facts, turning
inquiry than to its beginning. Do not be surprised that we have not reached
later to the objectivity of beliefs.
it yet.
Objective Facts
There are three strands to our ordinary notion of an objective fact or objec-
tive truth. First, an objective fact is accessible from different angles. Access
to it can be repeated by the same sense (sight, touch, etc.) at different times;
it can be repeated by different senses of the same observer, and also by
75

NOZICK, Robert. Invariances.

Uploaded by

Copyright:

Available Formats

NOZICK, Robert. Invariances.

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NOZICK, Robert. Invariances.

Uploaded by

Copyright:

Available Formats

What are the three strands to our ordinary notion of an objective fact or objective truth according to the passage?

What does the passage say about the objectivity or subjectivity of a belief?

16 INVARIANCES Truth and Relativism 17

respects. We have to add that no one of these different values is privileged,

ness and truth.

For instance, being digestible by human beings is a relational notion (and

You might also like