While individual philosophers continue to produce and defend as wide a range of metaphysical views as they ever have, today two outlooks have become dominant in American and French philosophy; these are, respectively, the outlooks of materialism and of relativism. Although few American philosophers actually call themselves materialists, and I do not know of any French philosophers who actually call themselves relativists, the terms “physicalism” and “naturalism” have become virtually synonymous with materialism in analytic philosophy, while the thrust of deconstructionist views is often clearly relativist, if not down-right nihilist. I have argued for some years that both styles of thought are too simplistic to be much help in philosophical reflection.1

I have already indicated some reasons for my dissatisfaction with materialism as an ideology. The materialist philosopher believes that present-day scientific theories already contain the broad outlines of a solution to philosophical problems about the nature of minds and intentionality; I have argued that there is no reason to believe that this is the case. However, my purpose here is not to polemicize against materialism and relativism, but to see what we can learn from the failure of these large points of view. There are, happily, many philosophers who reject both relativism and semantic materialism, that is, reject the idea that the semantical is going to be reduced to the physical, or even to the concepts of the “special sciences” which do not explicitly use intentional notions; but I often sense the feeling, especially among students, that the abandonment of these totalistic programs may mean the “end of philosophy”. Indeed, Richard Rorty2 has very strongly defended the view that philosophy is, in a certain sense, at an end and we must ready ourselves for a post-philosophical era. I don't think this is at all the right moral to draw from the present situation, but before we can see what the right moral is, we need to ask in more detail just where materialism and relativism have gone wrong.

Chomskian Fears of “Interest Relativity”

I begin by contrasting Jerry Fodor's attitude toward the concepts he takes as primitive—causation and the counterfactual conditional—and my own attitude as I sketched it in the last chapter. In my opinion, counterfactual conditionals and causal statements presuppose what I call “the point of view of reason”.

What I mean by this is that, for example, when we evaluate a counterfactual conditional (unless the counterfactual conditional is what I called a “strict” counterfactual conditional)3 we do not consider all of the physically possible situations in which the antecedent is true; typically, we are aware that there are physically possible situations in which the antecedent is true and the consequent is false, and yet very often we accept the counterfactual conditional as true notwithstanding. If my wife and I sit down to have breakfast, and we discover that I forgot to turn the gas stove on, and so there is no hot water for coffee, my wife might well say, “if you hadn't been so absentminded, and had turned on the stove, we would have hot water for coffee now”. I can think of a number of bizarre physical possibilities such that if any one of them had obtained, then I could have turned on the gas stove and yet the water would not have gotten hot. But mentioning any of these possibilities to my wife at that moment is not going to be a very successful strategy. As far as Ruth Anna is concerned—and she is using the counterfactual conditional in a perfectly correct way—her counterfactual is true. And what justifies her in calling it true is her knowledge of the regularity connecting the lighting of the gas stove and the boiling of the water that is put on it, and her knowledge that that regularity fully justifies her expecting that the water will be hot, conditional on my having turned on the gas stove. Note that the bizarre physical possibilities I might think of do not falsify her counterfactual.

That is not to say that Ruth Anna's counterfactual conditional is not defeasible. If it turns out that there was an interruption of the gas pressure in the line, unknown to both of us, and that the gas stove would not have stayed lit even if I had turned it on, then her counterfactual conditional is defeated. Someone might propose that what rules out the bizarre physical situations I mentioned from being relevant is that they are improbable. But if I had turned on the gas stove, then the resulting situation would have been different from the actual situation in countless ways (various molecules would have been in different places than they actually were, whole chains of causality would have been set into motion that were not set into motion in the actual world, and so on), and that situation would have certainly had a host of improbable features, since every actual physical situation always has features which are highly improbable. It is not probability as such, but probability in relevant respects that counts, and this again brings in what I called the point of view of reason.

To make what is involved clearer, let us suppose that the person who utters the counterfactual is not Ruth Anna but someone who knows no physics, and that I who hear the counterfactual know a great deal of physics. Even in such a case, I do not count as automatically falsifying the statement the fact that I know of physical situations that the speaker cannot even imagine in which I would have turned on the gas stove and the water would not have boiled. Rather I engage in the following kind of thinking: on the one hand, I put myself in the speaker's shoes, and imagine myself uttering the same counterfactual in the same situation and with the same practical point. At the same time, in evaluating the counterfactual that was uttered (and that I imagine myself as having uttered, in my imaginative identification with the speaker), if I consider it relevant, I take into account my own scientific knowledge, as well as the speaker's reasons and intentions in uttering the counterfactual. I try to decide—of course, in many cases, the decision is quite automatic, or the question does not even arise—whether there are any possible situations that I can think of that really falsify the speaker's counterfactual. In particular, if possible situations come to mind in which the antecedent of the counterfactual would be true and the consequent false, then I try to decide whether those situations—situations that falsify the counterfactual if they are relevant—do or do not have that kind of relevance. Possible-worlds semanticists would express this by saying that I try to decide if certain possible worlds are closer to the actual world than certain others.

If there is a criticism to be made of the language used by possible-worlds semanticists, the language of worlds being “closer” or “farther” away from the actual world, it is that that language conceals what needs to be brought out, that what is actually being judged is not the distance of objects from one another in a hyper-space but the relevance of hypothetical situations, and the relevance of situations to a judgment is an essentially normative matter. What using the language of “closeness” does is to make a normative judgment, a judgment as to whether it is reasonable to regard something as relevant, sound like a description of a “value-neutral fact”.

I have explained my views about counterfactuals and also my views about causation in several publications.4 Yet a wellknown philosopher asked me recently if I am a “causal nihilist”. What is going on here?

It seems to me that this is what happens: when I say that the truth of a judgment of the form A caused B depends upon the context and the interests of the people making the judgment (for example, what the speakers want to know in a particular context),5 then there are those who immediately leap to the conclusion that I must be saying that such judgments are entirely subjective, or, perhaps, that they are entirely arbitrary. This is not simply conjecture on my part. Noam Chomsky reacted in the following way to precisely my claim of the interest-relativity of statements of the form A explains B: “Therefore [Putnam] is offering as a substantive metaphysical thesis that correctness in linguistics (and psychology) is what best explains the currently available data about the behavior of the speaker given some current interests; what is correct today will be false tomorrow, and what is correct depends on our current interests and purposes.”6 Now, that what is correct today will be false tomorrow is no part of anything I have ever maintained; yet to Chomsky it seems obviously to follow from views like the ones I defended in the previous chapter.

If I am not very much mistaken, Chomsky's main hidden premises are two: (1) that we are free to choose our interests at will; and (2) that interests are not themselves subject to normative criticism. Or perhaps the premise is that what is normative is itself arbitrary and subjective? (This does not seem likely to me in the case of Chomsky, although it is, I think, behind the thinking of some of my other critics.)

Consider the medical example that I used earlier. In one context I may say that John's heart attack was caused by his failure to obey his doctor's orders; he insisted on eating highcholesterol foods and he refused to exercise. Yet in a different context I may say that John's heart attack was caused by high blood pressure. Assuming that both the high blood pressure and the failure to obey the doctor's orders were important contributory causes, and that I was aware of this in both situations, I certainly would not regard my statement in the second context as contradicting my statement in the first context. If someone said to me “but yesterday you said the cause was his failure to obey doctor's orders”, I might well reply “yes, but then the question was what John could have done to avoid the heart attack, and today the question is what physiologically predisposed him to the heart attack”. A caused B and A’ caused B may sound incompatible, but they are not, in such a case.

I want to repeat again the two points that seem to me important: (1) We cannot simply choose what interests we have. The language we speak reflects who and what we are, and in particular reflects the kind of interests we have. Since we know the kinds of interests that people do have, we are able to hear what sound like contradictory statements and to understand them in a way which is not contradictory. In case of confusion, we can easily reword these statements in other ways; for instance, we can say “even given his high blood pressure, John would not have had the heart attack if he had obeyed his doctor's orders” and “even given his bad diet and lack of exercise, John would not have had the heart attack if he hadn't had high blood pressure”. (2) It does sometimes happen that it is debatable whether a given interest is or is not relevant. If a Marxist says that the cause of John's heart attack was the capitalist system, we are likely to scoff—unless, of course, the Marxist makes a case that wins us over. Whether an interest is or is not relevant is something that can itself be argued. To say that a notion is interest relative is not to say that all interests are equally reasonable.

But what makes some interests more reasonable than others? The answer is that reasonableness depends on different things in different contexts. There is no general answer. The real dividing line here is between philosophers who, either consciously or unconsciously, assume that normative notions are subjective, and hence anything tainted by them must also be subjective, and philosophers who do not start with that assumption. Make the assumption that normative notions are non-cognitive; then, of course, any account of explanation, causation, and of the counterfactual conditional which involves normative elements will be heard as making all of these forms of discourse non-cognitive.

Fodor agrees with me that these forms of discourse have cognitive value, but for very different reasons. Fodor responds to those who would deny our right to use counterfactuals and ceteris paribus clauses by saying that, after all, we use them in geology (and other “special sciences”).7 Now, what is the significance of the fact that we use them in geology as opposed to the fact that we use them every day of our lives, in the kitchen, as it were? The answer is obvious: geology is a “science”. And genuine sciences, Fodor assumes, tell us what we have to assume to be there mind independently. Yet sciences like geology do not pretend to confine themselves to the “primary qualities” of realist metaphysics. Texts in biology are good examples of how we use language in certain kinds of explanations; they are not, and do not pretend to be, an inventory of “the furniture of the universe”, and only a bad case of scientism makes a philosopher mistake them for such an inventory.

Relativism

Richard Rorty has been in recent years one of the most important interpreters of continental philosophy to an American audience. Like the continental philosophers that he interprets, Rorty rejects the label “relativist”. But almost all of his readers have classified him as some sort of a relativist, and it is not hard to see why, particularly in his Philosophy and the Mirror of Nature. Although Rorty later repented of this formulation,8 in that work he identified truth, at least truth in what he called “normal” discourse, with the agreement of one's cultural peers (“objectivity is agreement”). I shall say a little later how Rorty escapes, or thinks he escapes, the charge of relativism. But it is natural on first meeting this formulation to take it in a relativistic spirit. So taken it says that truth in a language—any language—is determined by what the majority of the speakers of that language would say.

At this point one has to take account of Rorty's distinction between normal and hermeneutic discourse. In Philosophy and the Mirror of Nature, the idea was that much discourse is governed by standards on which the speakers of a language are in agreement. In that book (as in his recent Contingency, Irony, Solidarity) those standards were compared to an algorithm, that is, to a decision procedure of the kind computers carry out. This is not the only point at which we will see Rorty—in spite of his well-advertised break with analytic philosophy—employing devices which are very similar to the devices employed by analytic philosophers. The appeal to the notion of an algorithm in explaining how it is that certain things are true and certain things are false in the language of a community, even if meant as a metaphor, is a good example of this.

Rorty's picture is that under normal circumstances English speakers do not come into disagreement about a question like “Are there enough chairs for everybody in the dining room tonight?”. The statement that there are enough chairs is, if true, a truth of “normal discourse”, and its truth is certified by procedures on which the members of the community are in agreement. If agreement cannot be reached on a question because members of the community hold allegiance to paradigms which are incommensurable, the discourse is “hermeneutic”. The best one can do is try to understand the other in such a dispute and to keep the conversation going, according to Rorty. Statements uttered in hermeneutic discourse are true only in an honorific sense; each side in a dispute calls its statements true, but this is just rhetoric designed to persuade the other side to change its allegiance.

Although Philosophy and the Mirror of Nature contains brilliant criticisms of the kinds of metaphysics that Rorty rejects, Rorty's positive views are spelled out very elliptically and incompletely. In particular, it is not clear what the notion of agreement of one's cultural peers comes to, apart from the metaphor of an algorithm. If I say to my wife “our kitchen needs painting”, the only cultural peer who is aware that I think our kitchen needs painting in this case is my wife (assuming I don't discuss the matter with anyone else). In a sense, my cultural peers agree: that is, all of my cultural peers who actually know that I made the judgment agree it is true. Does that mean the judgment is true? Let us take a more extreme case. Let us suppose that I live alone and I think my kitchen needs painting, and I don't discuss this judgment with anyone. In that case all of my cultural peers who were aware of my judgment (namely me) agree that it is true. Does that mean that it is true, on Rorty's theory?

Most readers of Rorty take him to be saying that a judgment in normal discourse is true just in case one's cultural peers would agree if they were present, or if they were informed of the relevant circumstances. But the appeal to counterfactuals is something that Rorty himself has rejected, at least in a paper written after Philosophy and the Mirror of Nature.9 According to Rorty, to appeal to what people who were not actually present would say if they were present is to appeal to “ghost observers”. I don't know whether this represents a change of mind on Rorty's part, or whether he already rejected the counterfactual interpretation of his formula when he wrote Philosophy and the Mirror of Nature. In the latter case, I don't know how to interpret the formula at all. It seems likely that the metaphor of an algorithm seduced Rorty into assuming that a verification procedure is something which would give a result if applied, as a matter of objective “computational” fact, independent of who employs the algorithm. If so, he was unconsciously borrowing a picture from a philosophy which is diametrically opposed to his own.

Since Rorty is too hard to interpret, let us simply imagine a typical relativist who does use counterfactuals unselfconsciously, and who holds that what is true in a culture is determined by what the members of the culture would say (if they would fall into irresolvable disagreement, then this relativist may say, like Rorty, that the sentence in question does not belong to “normal discourse”, or he may regard it as having no truth value, not even a relative truth value). The very unselfconciousness with which such a relativist uses counterfactuals is the problem. If the truth or falsity of the statement that my kitchen needs painting depends on what my cultural peers would say, then what determines that? What determines what my cultural peers would say?

Contemporary analyses of counterfactuals suggest that two things determine it: (1) which possible situations (or which possible worlds) are close to the actual world (or, as I would prefer to say, relevant to the statement when we consider the actual situation in which it was made); and (2) what would happen in those possible situations. For a physicalist, the latter is no problem: if the possible situations are completely described in the language of physics, say by a “state function” in the sense of quantum mechanics, or of whatever theory may be a successor to quantum mechanics, what will happen in that situation (or the probability that any given thing will happen in that situation) is determined by the laws of fundamental physics (or Fodor might say that it is determined by the laws of fundamental physics plus the laws of the relevant “special sciences”).10 But this account makes the truth value of a counterfactual depend on the notion of something being a law of physics (and/or a law of the special sciences)—not a law of the accepted physics (or, respectively, special sciences) but a law of the true physics (special sciences), whatever that may be—and this is hardly acceptable to a relativist. Even if the notion of truth is itself interpreted relativistically in this formulation, the relativist has a problem: the truth value of the statement that my kitchen needs painting depends (for the relativist) on the truth value of the statement that people (in various hypothetical situations) would say that the paint in my kitchen is dingy and peeling, and that in turn depends on what the relevant laws are (physical, biological, psychological, and so on), and that in turn depends on what people would say the relevant laws are.

Relativists may, of course, deny that they need a “semantics” for counterfactuals. They may just insist that the counterfactual is clear as it stands, and that its truth needs no explanation at all. But metaphysical innocence, like other kinds, once lost is hard to regain. Once one has seen how difficult it is to give an account of the truth of a counterfactual, it is hard to see why someone who regards the truth of ordinary non-counterfactual statements as problematic, as a notion to be given up or radically modified, would regard counterfactual truth as unproblematic.

Let us suppose that our typical relativist does so regard it. At this point he or she encounters the following paradox. It is a fact about our present culture that there is no philosophical unanimity in it: we do not all accept any one philosophy, and certainly we are not all relativists. Moreover, this is likely to continue to be the case for some time. Rorty himself would very likely regard this lack of philosophical unanimity as a very good feature of our culture, and one which he would like to preserve. But if, as a matter of empirical fact, the statement “the majority of my cultural peers would not agree that relativism is correct” is true, then, according to the relativist's own criterion of truth, relativism is not true!

This inconsistency is not a logical inconsistency, because it depends on an empirical premise about the culture, but the empirical premise is one that few people would challenge. Rorty himself would say that his account of truth in Philosophy and the Mirror of Nature was not meant to apply to hermeneutic discourse but only to normal discourse. Thus the assertions of both relativism and anti-relativism are not themselves true or false in the sense in which statements in normal discourse are true or false. If I say that a philosophical utterance is true, I am, on Rorty's view, simply “paying it a compliment”. To put it in another way, the statement of what sounds like relativism is for Rorty not the announcement of a metaphysical discovery, but a bit of rhetoric: a bit of rhetoric whose purpose is to get us to change our ways, to give up talk of truth and falsity rather than to express some kind of metaphysical truth.11 Relativism a la Rorty is rhetoric.

Relativism and Solipsism

Typical relativists do think, paradoxically, that they have made some sort of metaphysical discovery. What are they to do if relativism involves a contradiction (or if a contradiction can be derived from relativism using logic that they do not challenge and uncontroversial12 empirical fact)? The move of “deconstruction” is from relativism to nihilism. Instead of offering a formula which is supposed to tell us what truth is, deconstructionists announce that the notion of truth is incoherent, part of a “metaphysics of presence”.13 Alan Montefiore reported to me that he once heard Derrida say “The concept of truth is inconsistent, but absolutely indispensable!” But what does it mean to say that a concept we find indispensable in daily life is “inconsistent”? Some uses of the word “true” may be inconsistent (as the familiar semantical paradoxes illustrate), but what does it mean to say that every use of the word “true” is inconsistent (or that every use of the word “leads us back to” such suspect notions as “presence”, “full speech”, and so forth)?

The only evidence that French philosophers offer for this astounding claim is that certain accounts of truth are, if not inconsistent, at least no longer metaphysically satisfactory. (According to Derrida, even the notion of a signifier—a word with a meaning—“leads us back to or retains us in the logocentric circle”.)14 The collapse of a large number of alternative philosophical accounts of truth is a very different thing from the collapse of the notion of truth itself,15 just as the collapse of a large number of different philosophical accounts of certainty is a very different thing from the collapse of the ordinary notion of certainty, as Wittgenstein tried to tell us in his last work. When a French philosopher wants to know if the concept of truth, or the concept of a sign, or the concept of referring, is consistent or not he proceeds by looking at Aristotle, Plato, Nietzsche, and Heidegger, and not by looking at how the words “true”, “sign”, “refer” are used. But this shows us more about French philosophy than it does about truth, signs, or reference.

What is interesting is that there is a way of making relativism consistent which was once more or less popular in philosophy, at least in disguised forms, and which is no longer popular at all: first-person relativism. If I am a relativist, and I define truth as what I agree with, or as what I would agree with if I investigated long enough, then, as long as I continue to agree with my own definition of truth, the argument that my position faces a standpoint problem, or is self-refuting, does not immediately arise. First-person relativism is virtually unfindable on today's philosophical scene, and it may be worthwhile to ask ourselves why.

One reason, a superficial but important one, is the widespread attention, even if not always accompanied by comprehension, that has been paid to Wittgenstein's private language argument. The private language argument is certainly difficult to understand, but part of its thrust is uncontroversial; what Wittgenstein might say, if he were addressing a relativist of the kind just described, would be something like this: “You speak as if language were your invention, subject to your will at each moment. The language games we play are alterable by our will only to a very limited extent. They are cultural formations, which have an enormous amount of inertia. Rightness and wrongness in a language game is internal to that language game; it's not something that was invented by you, and it's not something that refers to you.” A language game, Wittgenstein tells us, “consists of the recurrent procedures of that game in time”.16 The procedures in question existed long before me, and will continue to exist long after me.

Now, in one sense, it might be said that this is not an argument. (I don't pretend, of course, to have summarized the private language argument here.)17 The first-person relativist might say that even in natural language we sometimes speak of things as if they were non-relative which we later discover are relative. We do not ordinarily say that two events are simultaneons or non-simultaneous relative to my frame; yet, after learning of the special theory of relativity, we know that simultaneity is in fact relative to a frame, although there are good reasons why it is not necessary to take account of this in ordinary contexts. Similarly, the first-person relativist might claim to have made a philosophical discovery that truth, as we ordinarily speak of it in natural language, refers to a relational property. It is true that that relational property doesn't presuppose the relativist's own existence: when you speak, truth is truth relative to you; and when John speaks, truth is truth relative to John; and when Joan speaks, truth is truth relative to Joan; and so on.

The reason people don't take this line, I think, is that it is simply not convincing as a description of the way we use the word “true”. If I hear someone say that palladium is a rare earth, I do not take the assertion to mean that the speaker would believe that palladium is a rare earth if he or she investigated long enough (what's that to me, after all?). I take the speaker to be making an assertion that is to be checked—by us or by a third party—by studying palladium. It could of course be that after we have looked at palladium we will still not be able to agree on whether the assertion was right. But the language game of classifying elements in this way could not function at all if in every case we were unable to come into agreement on whether something was a rare earth.

Mary Warnock once said that Sartre gave us not arguments or proofs but “a description so clear and vivid that when I think of his description and fit it to my own case, I cannot fail to see its application”. It seems to me that this is a very good description of what Wittgenstein was doing, not just in the private language argument, but over and over again in his work. Suppose we simply describe how we use the word “true” carefully and attentively, suppose we give its “phenomenology”. We will find that the idea that it stands for the property of “being what I would believe if I continued to investigate” simply wrong.

I don't think that the private language argument, influential as it is, is the only reason for the decline of first-person relativism. Solipsism has never been a popular philosophical position, and first-person relativism sounds dangerously close to solipsism. Indeed, it is not clear how it can avoid being solipsism.18

Consider, for example, a statement about a human being who is no longer alive, a statement whose truth value we are no longer in a position to determine, say, “Caesar was shaved on the day he crossed the Rubicon”. It is part of what Cavell has called our “acknowledgment” of other human beings that we treat statements that they are able to verify as having just as much right to be called meaningful and true (or false) as statements that we ourselves are able to verify. If a statement is such that no human being could ever, under any conceivable circumstances, verify it, then, indeed, we begin to wonder—or some of us begin to wonder—whether the statement even has a truth value; but the statement that Caesar had a shave on the day that he crossed the Rubicon is one that at least one human being, namely Julius Caesar himself, was in a perfectly good position to verify. I no more doubt that that statement has a truth value than I do that the statement that I shaved this morning has a truth value. For a relativist, however—and now it doesn't make any difference whether we mean a first-person relativist or a cultural relativist—it is very likely that the statement about Julius Caesar has no truth value. It may very well be the case, in fact it probably is the case, that all traces of that event or non-event have long ago been obliterated. Worse, the question of whether it does have a truth value is just a question about what the relativist (in the case of first-person relativists) or the culture (in the case of cultural relativists) will come to believe, if they do their best (by their own lights, or those of the culture) to investigate the question. Julius Caesar is a logical construction out of the actual and potential beliefs of presentday people, on such a story.

Coming back to first-person relativism, what is true of Julius Caesar is true of people who are alive now. If you and I are not the first-person relativist in question, then the truth about me and about you and about the friends and the spouse of the first-person relativist is, for the first-person relativist, simply a function of his or her own dispositions to believe. This is why first-person relativism sounds like thinly disguised solipsism. But it is hard to see why cultural relativism is any better off, in this respect. Is solipsism with a “we” any better than solipsism with an “I”?

Materialism, Relativism, Metaphysical Realism

Wittgenstein introduced the image of language as an overlapping system of games. But what is often overlooked is that in one place Wittgenstein stresses that rightness or wrongness in a language game is not always determined by rules. I shall discuss the passage in question later; for the moment let me just say that the passage we will examine19 also makes a different and less emphasized point. Wittgenstein looks at disagreements about what is going on in the soul of another human being (Is he faking an emotion? Has she fallen in love?). Sometimes such a disagreement can be settled to everybody's satisfaction. But Wittgenstein describes a case which is not so settled, and where what counts is “unwägbare Evidenz” (imponderable evidence).

The phenomenon of the controversial, of what cannot be settled to the satisfaction of everyone who is “linguistically competent”, is, however, ubiquitous, and extends far beyond the psychological. Even so-called factual judgments are frequently controversial, at least in some parts of the culture—(think of the disputes about evolution between scientists and fundamentalists, or think of the impossibility of convincing some people that there are no American prisoners of war left in Vietnam). It is part of our form of life, part of the way we live and think and act and will go on living and thinking and acting, that each of us treats many controversial propositions as definitely having a truth value. Rorty will of course say that such sentences are not part of “normal” discourse, that to call them true is only to “pay them a compliment”; but the minute he leaves his study he will talk of the fundamentalist opponent of evolution, or the right-wing crank who thinks that there are prisoners of war in Vietnam, as a fool. What it is right to say in a given context cannot always be established to everyone's satisfaction; but it is nonetheless the right thing to say.

Yet the explicit recognition that language games are human activities in which what is right and wrong is not simply conventional, is not simply determined by consensus, but is something that requires evaluation, is troubling to many a contemporary sensibility.20 (Perhaps it was even troubling to Wittgenstein; perhaps this is why in the Investigations there is only this one isolated reference to this all-important fact.21 Later I shall talk about some unpublished lectures of Wittgenstein, which were taken down by others and published after his death, which shed more light on the nature of his thinking.) But the distrust of the normative in present-day philosophy is evidenced above all by the lengths to which philosophers will go to avoid admitting that truth—that is, the rightness of what is said—is a normative notion.22

We have seen that both materialists and relativists avail themselves of the counterfactual conditional when they try to explain what truth is. At first blush, one would not expect either kind of philosopher to be very happy with this way of talking. Matters of fact about non-actual situations do not fit very well into either the worldview of materialism or the anti-metaphysical prejudices of relativism. If, in spite of this, one finds both kinds of philosophers resorting to this machinery, it can be only because the price of not doing so seems too high. Philosophers of a more traditional kind would have tried to escape the problem in still another way. For such philosophers, to say that a sentence is true is not to make a normative judgment at all: it is just to say that the sentence “agrees” with something (“the facts”) or that it “corresponds” to something (“a state of affairs”). But I have shown elsewhere, using a number of theorems from contemporary model theory,23 that the notion of “correspondence” is totally empty in this context. It is possible, in fact, to interpret our language, in the sense of “interpret” used in contemporary model theory, in such a way that the sentences of any consistent theory “agree with reality” under an appropriate correspondence. Even if the truth conditions for all the sentences of our language are in some way fixed, it is still possible to find a correspondence under which every sentence of our language retains its present truth conditions (up to logical equivalence), although the references of the individual words are changed so radically that the word “cherry” ends up referring to cats and the word “mat” ends up referring to trees.24 It was in response to this theorem that a number of physicalist philosophers suggested that what singles out the reference of a term is not a matter of model theoretic correspondence in the abstract, but specifically of “causal connection” or “causal attachment”. In the previous chapter I examined the most recent form of this suggestion and saw how little it actually comes to. Beyond that, as I have also pointed out, these philosophers ignore the language dependence of the ordinary notion of causality itself. In effect, the ordinary notion of causality is a cognitive notion, and these philosophers treat it as if it were a purely physical one.

What I have been suggesting is that philosophy of language is in a bind because it is hell-bent on eliminating the normative in favor of something else, however problematic that something else may be. In analytic philosophy, this desire to eliminate the normative often goes hand-in-hand with the idea that science is “value free” and that science and only science tells us how things “really are”. In the next chapter I examine this science-ethics distinction in Bernard Williams’ recent and influential Ethics and the Limits of Philosophy.