RORTY ON IDENTITY, PRIVACY, AND THE CATEGORIES (1965)

by Steve Bayne

There is a semantical argument that says we cannot identify mental and physical events because to do so is to commit a category confusion, it being an apriori fact that we cannot identify things falling under mental categories with things falling under physical categories. It was considerations related at least to this that compelled Strawson to distinguish two kinds of predicates - those belonging to the body and those belonging to minds - thereby encouraging him to recognize a category of 'person' itself unanalyzable but capable of accomodating both sort of predicates.

Richard Rorty in a very influential paper ("Mind-Body Identity, Privacy and Categories," Review of Metaphysics. 1965. pp. 24-54 (reprinted in the volume I shall be using, _Materialism and the Mind Body Problem. ed. David M. Rosenthal. Prentice Hall 1971. pp 174-199) argues against this semantical argument. He argues that there is no categorial distinction between the empirical and the conceptual, and that because our categories reflect a given stage of language development there is no apriori argument against the identity theorists, those who would identify brain states and mental states. According to him the dualist who argues in the way we have indicated doesn't fully appreciate that our categories are merely linguistic notions subject to emendation by science. In one pithy remark he makes his general point clear:

To argue against the Identity Theory on the basis of the way we talk now is like arguing against an assertion that supernatural phenomena are identical with certain natural phenomena on the basis of the way in which superstitious people talk. p. 175

Rorty identifies two forms of the identity theory, the "translation form" and the "disappearance form." Let's pursue the one he accepts, the disappearance form. Given this position, (A) and (B) are of the same relevant form:

A. What people used to call 'demonical possession' is a form of hallucinatory psychosis.

B. What people now call 'sensations' are identical with certain brain processes (Rosenthal p. 177)

These two cases are examples of the form "X's turn out to be nothing but Y's." Where "X" stands for "Unicorn horns," we might say that the form "Unicorn horns turn out to be nothing more than narwhal horns" might serve, if true, to justify eliminating talk of unicorn horns in favor of "narwhal horns." No longer are we existentially committed to unicorn horns. But contrast this with similarly substituting for "X" the word 'chairs' (ala Eddington), and arriving at something like"Chairs are nothing but clouds of molecules." This case differs from the first. Here we would not recommend *eliminating* talk of "chairs." In the first case, says Rorty, we are identifying observables with other observables, but here we are identifying observables with theoretical terms. But now there is a problem. Look at (B). Which of these forms of "X's turn out to be nothing but Y's" does (B) belong to?

People may have been mistaken in their identification of certain horns as unicorn horns but would they be mistaken in identifying pains with sensations? The answer is "No." This rules out assimilation to the first kind of identification, but the second kind is no more useful, for neither is it the case that brain processes are theoretical entities (p. 182). I happen not to agree here with Rorty. There is much in such processes that is theoretical, but I will let this pass. What Rorty wants to do is link sensations and unicorn horns to the same sort vis a vis the identity theory. He needs to overcome the counterintuitive nature of the claim that I do not report sensations anymore than I report demons or unicorn horns. Let's see how he goes about the attempt.

We know that the unicorn horn case and the sensation case differ in the sense that once we find out that unicorn horns are nothing but narwhal horns we desist reporting unicorn horns, but we would hardly desist from saying I have a toothache merely because we find out the neurological basis of such pains. Here is the dilemma for the identity theorist as Rorty sees matters:

...either the identity theorist claims that talk about sensations presupposes some empirically disconfirmed belief (and what could it be?) or the "identity" which he has in mind is the uninteresting sort of identity which holds between tables and clouds of molecules (mere "theoretical replaceability"). (p. 182)

Rorty attempts to escape the dilemma by making a very subtle move, but first there is a not so subtle move grounded firmly in his pragmatism. What he does is look at (A) and (B) (the unicorn horn and sensation cases, respectively) and make the claim that in fact we might as well eliminate tables as well as unicorns but for the fact that it is "monstrously inconvenient" to do so. This is unsubtle and unsurprising in a way. But now for the subtle move.

I shall try to make my claim plausible by sketching a general theory of the conditions under which a term may cease to have a referring use without those who made such a use being convicted of having held false beliefs. (p. 183)

What does this mean?! It means that he wants a theory that will allow us to admit reports of sensations while not affirming that we refer to sensations in making those reports. The truth of the report remains but not the purported referent of its terms, or at least not of all of them. Why then is it that we easily eliminate use of 'unicorn horn' but not 'pain' in favor of something like 'C-fibers'?

The basic reason is that we *habitually* and non-inferentially report pain but not unicorn horns, just like we don't drop the term 'temperature' even though it "reduces" to the "kinetic energy of molecules." It is all an issue of convenience: Since "it is simpler to stop inferring to the existence of demons altogether than to continue making such inferences" (p. 185) we drop talk of demons in scientifically advanced societies. The same cannot be said for dropping 'temperature', say in terms of 'mean kinetic energy'. All the materialist alleges, according to Rorty, is that "at no greater cost than an incovenient linguistic reform, we *could* drop such terms" as 'sensations'. (ibid) But the question now becomes "What, then, are we reporting when we report having a 'pain'? " The answer is that we are reporting something like "My C-fibers are firing. But this sounds rather odd, and Rorty know it.

Are we to imagine that a child who says, "I am in pain" is reporting on his C-fibers? The oddity is largely owing, says Rorty, to the fact that we do not noninferentially report C-fibers, and this - recall - was a requirement of 'X's being said to be nothing more than 'Y's. Let's quote Rorty at this crucial point:

So the proponent of the disappearance form of the identity theory must show that reports of brain-processes are neither incapable of being non-inferential nor, if non-inferential, necessarily made in the way just imagined (with the periscope-microscope gadget) or in some other peculiar way. (p. 187)

But instead of defending the diappearance theory directly he searches for higher dialectical ground, asking "But now we must ask who bears the burden of proof" Here I think it is very possible that Rorty leaves himself open to the charge of circularity. What he is attempting by shifting the burden of proof is to place the opponent of the disppearance theory in a position where he has to show that we do not non-inferentially report our C-fibers firing. So we ask ourselves under what circumstances we could be understood to be directly reporting the firing of C-fibers. Here is his answer:

For this will in fact be the case if, when we were trained to say, e.g., "I'm in pain" we were in fact being trained to respond to the occurrence of within ourselves of a stimulation of C-fibers. (ibid).

But if Rorty requires that in fact we are reporting C-fibers firing, he is assuming that sensations are nothing but C-fibers firing. In other words he can shift the burden onto the opponent of the disappearance theory only by assuming that the disappearance theory is true. This is how I read him here.

Setting aside my circularity objection, we can at least say that it appears to Rorty at this point that his disappearance form of the identity thesis needs to overcome only one more possible objection against it: Even if one is successful in making reports of C-fibers noninferentially reportable, one still must contend with the argument that the elimination of the mental will not go through because unlike C-fibers mental events are essentailly private. I think it is important to note that privacy is something merely epistemic for the dualist on Rorty's account and contains not even so much as a purported ontological basis (p. 188).

In attempting to overcome the privacy argument against identity theory, Rorty relies on the argument as conceived by Kurt Baier as paradigmatic. Basically the idea is that because it will *never* be possible to construct a C-fiber detector capable of overruling first person reports there will always remain an epistemic hiatus between "private" sensations and "public" C-fibers, so the identity argument collapses. Here is what Baier is quoted as saying:

To say that one day our physiological knowledge will increase to such an extent that we shall be able to make absolutely reliable encephalograph-based claims about people's experiences, is only to say that, if carefully checked, our encephalograph-based claims about 'experiences' will always be correct... But ...However good the evidence may be, such a physiological theory can never be used to show to the sufferer that he was mistaken...for such a mistake is inconceivable ...Talk about brain processes therefore must be about something other than talk about experiences. Hence, introspective reports and brain process talk cannot be merely different ways of talking abouut the same thing. ("Smart on Sensations," _Australasian Journal of Philosophy. (1962). pp. 64-65)

The Smart answer is to say that a situation like this will never arise; Rorty's answer is to discount this case as "not the interesting case" (p. 190). The interesting case is one where the machine has worked for years but suddenly one day a subject claims he has no pain when the "gizmo" says otherwise. What to do, what to do? Rorty replies that it is "fairly clear" what we do. What we do is say that the subject doesn't know what he is talking about...that "he is not using the word 'pain' in the way in which his fellows use it." But is this really the "interesting" case? Imagine the subject being told he is *not* in pain and that his or her writhing is the untoward reality of a verbal misunderstanding! Also, if Rorty's is the "interesting" case, one can imagine knowing one is in pain at some future time only by looking at the "gizmo." Keep in mind that nothing has changed as far as the subject's physiology is concerned, only the advance in electronic experience detection, and also keep in mind that Rorty's machine *must* in theory be capable of detecting the presence or absence of *any* sensation, so he can't just say: "Well he feels something but it just isn't pain" This can be adjusted for. Wouldn't the subject with a long memory, recalling a past time where 'pain' was used to refer to something seemingly different, at least, be inclined to say that it was the physiologists, not he, who were linguistically challenged? Would we refuse anaesthetic to a patient based on our disposition to narrowly interpret errors of fact and errors of judgment, or would it be that we simply wouldn't know what to do with anaesthetic in such a case? (As an aside, consider that there may be an "is/ought" question here that need examination) A final comment on this phase of the argument.

What is the *evidence* for what Rorty finds "fairly clear" ? In other words, what is the relative epistemic defeasibility of the claim that I am in pain made by me as compared to that of the claim that I don't know how to use my words? And what ever happened to the not so "interesting" case from Baier? So what if it isn't interesting? What is Rorty's reply? It is "fairly clear" that in this particular instance Rorty's "linguistic turn" here is a sleight of hand. But if I am to convince the reader this is not merely hyperbola of a debased and contentious sort, let's take a bried look at a crucial nexus in Rorty's argument, because it is here that exhausted readers will be inclined to "throw in the towel" and simply follow Rorty's in drawing conclusions.

The place I have in mind is where Rorty simply tells us without argument, as we have noted, that the subject, Jones, whose testimony contradicts the gizmo's result considers as his "first move) (p. 190) that he is misusing 'pain'. How can he verify that he is *not* using 'pain' incorrectly? Doing so requires, according to Rorty, that he be able to distinguish "misuse of language" and "mistake in judgment." But in this case, Rorty alleges, the distinction can't be made! And to prove it he complicates the situation even more by entertaining the possibility that before being tested for pain Jones is burned but avows that there is no pain. He may testify to some sensation (but this is complicated by the fact that sensations must be detectable eventually tout court in order for the disappearance theory to work, but I will ignore this). Suppose, then, that contrary to Jones the gizmo "says" that he *is* in pain. Now, says Rorty, we cannot tell whether the error is judgmental or a matter of language.

In order to make the determination we would need a convention telling us that under these circumstances we are to say Jones doesn't know the meaning of pain. But even Rorty recognizes that convention won't solve the problem since the only way to prove that Jones is unable to uninferentially know pain is to provide cases where he denies having it but in fact does (191). If Jones were to hop around exhibiting pain behavior while denying he is in pain he may be said to violate a criterion upon which the convention applies, and this is the only sense at this point of the discussion where much sense can be made of Rorty's puzzling remark "So now we would have a public criterion" mangled as it is together with the idea of a convention (ibid). Finally Rorty says,

The dilemma is that either a report about one's sensations which violates a certain public criterion is a sufficient condition for saying that the reporter does not know how to use 'pain' in the correct way, or there is no such criterion. (191-192)

Suppose there is such a public criterion. Then, according to Rorty, just because one can't be mistaken doesn't imply that pain reports cannot be overridden. But wait! If they can't be mistaken, how *can* they be overridden?! Secondly, suppose there is no criterion. Then, says Rorty, we have no way of *ruling out* that he doesn't know the meaning of 'pain'; but then again we cannot rule it in, either. But the real problem with this line in Rorty's argument is when he says:

...since the apriori probability that he does not is a good deal higher than the a priori probability that the psychophysiological theory of Jone's era is mistaken, this theory has little to fear from Jones. (192)

But this must give us pause, for it is the very issue of whether any psychophysiological device in *any* era could defeat Jone's avowal that we began with, and Rorty has simply brought us back to from whence we began. Twice now we have been drawn into a circular argument. But there is a final formulation of how it may turn out to be reasonable to overrule a first person report of pain using a machine.

The case Rorty puts forth is one where public criteria for one's knowing how to use 'pain' change. So when the "appropriate" brain state is produced a machine murmurs 'pain' in a child's ear and in this way inculcates the accepted criterion. The "appropriate" brain state now is correlated with the murmurings of a machine not those of people whose brains are at issue. The "appropriate" brain state for 'pain' attribution is then singled out by meter readings, rather than avowals. We are in a curious situation where our "inability to be mistaken does not entail our inability to be overridden" (p. 194). How convincing is this?

The subject, I submit, may not understand the meaning now assigned to 'pain' but this does not mean that he is using the same word before and after all the scientific progress Rorty assumes; indeed the word *has* changed, despite its identical morphology. The original question was whether Jones might not misunderstand the word he was using in contradicting the machine reading, or whether it was a judgmental error on his part, possibly. Rorty's point was that since the distinction cannot be made Baier's case is inconclusive. Now we have a purported case where we can conclusively show, if Rorty is right,that the subject is overruled as a linguistic show of strength, rather than that he is mistaken. This falls short of the mark,however.

If the machine cannot be shown to contradict the subject then Baier type cases are not diffused. Since the criteria have changed and Rorty admits the subject is not mistaken, then there is some reason to believe that the machine is not contradicting the subject (on ground of equivocation). Further, being "overruled" tells us only that there is every likelihood that the progress Rorty assumes never actually took place, that the brain state under the description of what is detected by the machine is not the brain state being reported, if indeed contrary to the dualist hypothesis - the very thing under dispute - it is a brain state being reported at all.

Lest I be thought unfair to Rorty, it must be admitted that the criticism I lodge above of equivocation is anticipated by Rorty (p. 194) but his reply is not encouraging to his case. His appeal is more a strategic reliance on our presumed fatigue, I suspect, than content. He says,

...Baier's explanation of the final epistemological authority of first-person reports of pains by the fact that this "logic" is " a function of this type of subject matter" rathther than... convention is an explanation of the the obscure by the more obscure. (ibid)

Rorty also relies - without elaboration - on citing Putnam on the problem related to equivocation.

But there is a more compelling criticism of his line of reasoning. Baier's interesting case is simple and direct; Rorty's is convoluted and depends on a number of additional assumptions. There is greater initial probability that Baier's case proves its point than that Rorty's proves his; moreover, the straightforward Baier case is something to which Rorty returns but only to say that "the body of curent sceintific theory foundering upon the rock of a single over-riding report - can probably never arise" (p. 107). Viewed one way, "probably" is something of an admission, for it leaves open the very possibility that identity theory should exclude, a point that didn't escape Smart - but which seems to require a mere contingent identity of identicals in contradiction to "Marcus's Theorem" that if two things are identical they are necessarily identical. Returning now to the issue of whether the Identity Theorist can be shown to be guilty of a conceptual confusion, Rorty relates the following possible objection to invoking language change in the course of defending the identity thesis:

Now a critic may object that this strategy is subject ot a *reductio ad absurdum*. For the same fact about linguistic change would seem to justify the claim that *any* statement of the form (s) "What people call 'X's may be discovered to be 'Y's" is *always* sensible and unconfused. Yet this seems paradoxical, for consider the result of substituting, say "neutrino" for "X" and "mushroom" for "Y." If the resulting statemtent is not conceptually confused, what statement is? (p. 197)

One may question here whether we are talking about a conceptual confusion or an implicit contradiction, a problem of concepts or a problem of essences. Later Kripke would illuminate this territory with the assistance of his theory of rigid designation. This would become the next significant episode in this dispute. (2001)