V: BEHAVIORISM, PHYSICALISM, AND THE IDENTITY THESIS

It is the purpose of this chapter to consider the views on intentionality defended by Chisholm and Sellars, in relation to three philosophical theses which have, in one version or another, received important philosophical attention in recent years. These theses are logical (or philosophical) behaviorism, physicalism, and the so-called identity thesis. In particular, it will be useful to examine in some detail the relation of Chisholm's and Sellars' views on intentionality to both the identity thesis, and to a certain version of the thesis of physicalism which we shall call the revised thesis of physicalism, and it is to these tasks to which the greatest part of the current chapter will be devoted. Nonetheless it will be useful to begin, in sections I, II, and III, by examining two philosophical theses which, we shall see, we have some cause to reject, namely, logical behaviorism and a certain version of the thesis of physicalism.

I. Logical Behaviorism (1): Chisholm's Views

One of the most important claims which Chisholm makes for the version of the thesis of intentionality defended in "Sentences about Believing" is that if it is true, then at least a certain version of the thesis of behaviorism is false. (Vide in particular [Chisholm and Sellars 523] and [Chisholm (17) 515-517].) In order to examine this claim, it will be useful to have a formulation of behaviorism. Let us accept the following formulation, which we shall call the thesis of logical behaviorism:

(1) Every sentence about the thoughts and propositional attitudes of persons means the same as some sentence about the observable behavior of persons or dispositions or propensities to such observable behavior.

Our immediate question, then, is whether the truth of Chisholm's claims in "Sentences about Believing" is incompatible with the truth of sentence (1).

Chisholm's claims in "Sentences about Believing" involve the idea that every psychological sentence either exhibits some logical properties or contains some term such that we can say whatever we wish about non-psychological phenomena without the use of language which exhibits these logical properties or contains such terms. If this claim is correct, then if sentences about observable behavior and dispositions or propensities to observable behavior are nonpsychological sentences, then such sentences are not inter-translatable with psychological sentences, and thus sentence (1) would be false. If on the other hand, such sentences are merely a special sort of psychological sentence, then while the truth of Chisholm's claim would not entail the falsity of sentence (1), the truth of sentence (1) would be trivial. For logical behaviorism would amount to no more than the claim that certain psychological sentences mean the same as others.

If the argument at the end of Chapter I is correct, however, then it is not required that we make reference to any special logical properties exhibited by psychological sentences in formulating Chisholm's thesis of intentionality. In particular, it was argued that any sentence which exhibited such properties would, given our construal of 'technical term', contain such a technical term. On the basis of this consideration and because of the fact that Chisholm's more recent formulations of a thesis of intentionality seem to be substantially weaker than that in "Sentences about Believing," it was argued at the close of Chapter II that we could regard the following as an acceptable formulation of the thesis which Chisholm wished to defend:

(2) There is a class of terms such that each member of this class satisfied the following three conditions: (a) no psychological phenomenon can be adequately described without using at least one of these terms; (b) no one of these terms is required for describing any nonpsychological phenomenon, or for formulating the language of the physical sciences; and (c) none of these terms can be defined without making reference to another of them.

Given the truth of (2), any psychological sentence will contain at least one of these terms. (It is clear that in order to draw this conclusion, we must regard every psychological sentence as "adequately" describing some psychological phenomenon. Because our present purpose is to assess the relation of Chisholm's views to the thesis expressed by sentence (1), and not to assess the adequacy of Chisholm's views, we shall for the moment overlook the occurrence of this tendentious, and somewhat vague, requirement.) The problem which a defender of the thesis expressed by sentence (2) would face would be on a par with the problem which, it was argued, was crucial to Chisholm's argument in "Sentences about Believing"--namely, the problem of finding a way of determining whether a given term is a technical term or not. If, however, we accept sentence (2) as true, then it seems clear that we must reject the thesis expressed by sentence (1). For given that we accept that sentences about observable behavior and dispositions to behave are nonpsychological sentences, such sentences either will contain no term of the sort mentioned in sentence (2), or will be expendable in favor of a sentence which does not contain any such term; for according to condition (b), no such term is required in order to say whatever we might wish to say about any phenomenon which is non-psychological.

The incompatibility between (1) and (2) may be seen somewhat more clearly from the following considerations. Let us suppose that P(t) is a psychological sentence, containing a technical term, t, and that B(t') is a sentence about observable behavior containing a (possibly different) technical term, t'. If behavioral sentences are not simply a certain sort of psychological sentence, and if sentence (2) is true, then B(t') is equivalent to some non-psychological sentence, N, containing no such term. If, however, sentence (1) were also true, then B(t') might be equivalent also to P(t). Granted the transitivity of the relation of equivalence which we might claim to be relevant to such considerations (say, equivalence in meaning), this would result in the equivalence of P(t) with N. But if sentence (2) is to be accepted, then this result is ruled out. Thus it seems that Chisholm's thesis is indeed incompatible with any non-trivial version of logical behaviorism, any version, that is, according to which behavioral sentences are themselves non-psychological.{1}

It seems clear that at least one of the motives which led Chisholm to defend the claims advanced in "Sentences about Believing" is a desire to show that the sort of thesis of logical behaviorism which is expressed by (1) is incorrect. Thus Chisholm writes that "when we analyze the kind of meaning that is involved in natural language, we need some concept which we do not need in physics or in 'behavioristics.'" [Chisholm and Sellars 523] Thus Chisholm's arguments in defense of the claims in "Sentences about Believing" may be accepted as arguments in defense of a view which is incompatible with logical behaviorism, as it is expressed by (1). It was argued in Chapter I, however, that Chisholm's claims in "Sentences about Believing" could not be defended satisfactorily. For this reason, it does not seem that Chisholm has succeeded in showing that logical behaviorism is incorrect on the basis of anything claimed in that article.

The only other sorts of argument which it seems that Chisholm has propounded in defense of his apparent conviction that logical behaviorism is incorrect are arguments in favor of the alternative formulations of a thesis and definitions of intentionality examined in Chapter II. Thus one might argue that in spite of the fact that the indefensibility of "Sentences about Believing" prevents it from being effective in showing logical behaviorism to be incorrect, some alternative thesis and definition of intentionality might be effective in doing so. For it was argued, in Chapter II, that the claims Chisholm makes in these more recent attempts to formulate a thesis and definition of intentionality are, in fact, defensible. It was also argued, however, that these latter claims were, in certain respects, considerably weaker than the claims made in "Sentences about Believing." For this reason, if one were to argue that the defensibility of these later claims can be used to show that logical behaviorism is incorrect, one would first have to show that the truth of these later claims is incompatible with the truth of logical behaviorism, as expressed by (1).

It is important to see just how we might go about trying to show this. Since logical behaviorism, as formulated by (1), involves a claim that every sentence which is about thoughts or propositional attitudes is equivalent in meaning to some sentence about overt behavior or propensities or dispositions to behavior (we may call a sentence of this second sort a behavioristic sentence), it will clearly be sufficient to show that (1) is false if we show that there is at least one sentence about thoughts or propositional attitudes which is not meaning-equivalent to a behavioristic sentence. Now it seems clear, given either of the two recent definitions of intentionality discussed in sections III through V of Chapter II, that there are at least some sentences which are both about thoughts or propositional attitudes and intentional, by the definitions in question.{2} Thus a typical sentence to the effect that a certain person believes something to be the case will, on either of the definitions under consideration, be intentional.

Let us then take P to be such a sentence--a sentence, that is, which is both about some propositional attitude or thought, and is intentional. Let us suppose, further, that (1) is true. If (1) is true, then there is a behavioristic sentence, which we shall call B, which means the same as P. But if it means the same as P, then it entails P, and since both of the recent definitions of intentional which we are considering provide that a sentence is intentional if it entails an intentional sentence, it follows that B is itself intentional. The thesis of intentionality which Chisholm advances in connection with each of these two recent definitions is that every intentional sentence is psychological. Thus, if each of our assumptions is true, and if Chisholm's more recent thesis is true, then it follows that our behavioristic sentence, B, is a psychological sentence. Thus it seems that if Chisholm's thesis is true, and if we accept either of his recent definitions, and if there is any intentional psychological sentence, then any behavioristic sentence which is meaning-equivalent to it must also be psychological--that is, must be about some psychological phenomenon.

As we have seen in connection with our discussion of the incompatibility of (1) with Chisholm's claims in "Sentences about Believing," the thesis of behaviorism which we are concerned to defend would be trivialized if we allow behavioristic sentences to be psychological, for then (1) would assert no more than that every psychological sentence means the same as some psychological sentence of a special sort. If the foregoing argument is correct, however, then it seems that we have generated the following result. If Chisholm's recent version of the thesis of intentionality is true, then logical behaviorism can be true only if behavioristic sentences are also psychological, at least in some cases. For it seems that if Chisholm's thesis, as recently formulated, is true, then any sentence which is meaning-equivalent to a psychological sentence which is also intentional must be itself psychological. Thus if that version of Chisholm's thesis is true, then any behavioristic sentence which is meaning-equivalent to a psychological, intentional sentence must itself be psychological. Since it seems that a large number of psychological sentences will, in fact, be intentional, it follows that if Chisholm's thesis is true, and if (1) is true, then a large number of behavioristic sentences will have to be psychological.

A defender of the recent version of Chisholm's thesis might take this sort of argument as an indication that since Chisholm's thesis seems defensible, logical behaviorism seems, in the same degree, indefensible. Similarly, a defender of logical behaviorism might take this sort of argument as an indication that the truth of Chisholm's thesis is still an open question, and that there is good reason to suppose that upon more probing examination, it would turn out to be false. It will be useful, therefore, in an attempt to settle an imagined dispute of this sort, to re-examine the nature of the claim which, in formulating logical behaviorism by (1), we have attributed to the logical behaviorist.

Sentence (1) involves a claim of the following sort. It is asserted that every sentence which is of a certain type, T1, means the same as some sentence which is of another type, T2. These two types of sentence are determined, according to the claim expressed by (1), on the basis of what kind of phenomenon the sentences in question are about. Thus, to make the claim more explicit, it is asserted that every sentence which is about a certain kind of phenomenon, K1, is meaning-equivalent to some sentence which is about a (possibly) different kind of phenomenon, K2. But it is intuitively plausible that if two sentences mean the same, then they are about exactly the same things. But if this is true, then even if we put Chisholm's claims completely aside, it seems that if (1) is true, then the behavioristic sentences which are meaning-equivalent to psychological sentences must be about psychological phenomena, and that every psychological sentence which is meaning-equivalent to some behavioristic sentence must be about not only some psychological phenomenon, but some behavioristic phenomenon as well. Thus any behavioristic sentence which is meaning-equivalent to a psychological one will be, itself, a psychological sentence.

The argument in the foregoing paragraph rested on the idea that if any two sentences mean the same, then they are about the same things.{3} It seems, therefore, that if we accept this claim, then we do not need Chisholm's recent version of the thesis of intentionality in evaluating logical behaviorism, for the strongest result that we seem to be able to generate about logical behaviorism by using Chisholm's thesis seems to be that if (1) is true, then behavioristic sentences are (at least many of them) a certain sort of psychological sentence. But since if the argument of the foregoing paragraph is correct then we can generate this result without reference to Chisholm's thesis, it does not seem that Chisholm's thesis represents a useful way of attacking logical behaviorism. For it seems at best questionable whether Chisholm's thesis is true, and the idea that if any two sentences are meaning-equivalent they are then about the same thing(s) does not seem open to question in the same way or to the same degree. It seems, therefore, that the most damaging results which we could generate about logical behaviorism by using Chisholm's somewhat elaborate claims, can be generated also by using a simpler, more straightforward, and less questionable claim.

It will be useful, however, to discuss briefly three ways in which the attack on logical behaviorism advocated in the last two paragraphs might be countered. (i) It might be argued that (1) does not correctly capture what logical behaviorists would wish to maintain, and that a different formulation, which would not be open to the argument presented above, should be sought. It is unclear, however, what alternative formulation would be both recognizable as a version of logical behaviorism, and not subject to the sort of argument presented above. For it would seem that a version of behaviorism which is to count as logical behaviorism must invoke a notion of meaning-equivalence, and that the occurrence of this notion will immediately subject any such version to the sort of argument presented above. In any event, before such a counter could be evaluated, much less found persuasive, a defender of logical behaviorism would have at least to indicate a reasonable alternative formulation, and it is not clear what alternative would be acceptable.{4}

(ii) It might be argued that although (1) does represent an acceptable formulation of logical behaviorism, it is nonetheless possible to understand (1) as involving a sense of 'about' or a sense of 'means the same as' (or both) such that the idea that any two meaning-equivalent sentences are about the same thing(s) is rendered unacceptable. Thus although this sort of counter would not dispute that in some senses of 'about' and 'means the same as' which are, for certain purposes, at least as acceptable as any others, this claim is correct, nonetheless it is not such a sense of these terms which is invoked in (1), understood correctly. The trouble with this move seems obvious. It is simply unclear how a defender of logical behaviorism who invoked this type of argument would effectively characterize his senses of 'about' and 'means the same as', aside from claiming simply that meaning-equivalence of two sentences is not, given those senses, sufficient to infer that the two are about the same thing(s). Without some plausible account of such senses, it seems clear that this move will be unsatisfactory.

(iii) The third sort of argument which might be offered here is that although (1) is a satisfactory formulation of logical behaviorism, and the meaning-equivalence of two sentences does entitle one to infer that they are about the same thing(s), nonetheless the idea that (1) cannot be true unless the behavioristic sentences in question are themselves psychological sentences is compatible with the spirit of the thesis of logical behaviorism. That is, a defender of logical behaviorism might reply to the argument presented four paragraphs back, that it is no deficiency of the thesis a logical behaviorist would wish to defend that the behavioristic sentences which could be used to instantiate sentence (1) must all be psychological sentences, and that it is no deficiency of that thesis that if the thesis is true, then all psychological sentences are also behavioristic sentences.

If a logical behaviorist were to take this course, then it does not seem that either the above argument attacking logical behaviorism independently of Chisholm's claims, or Chisholm's claims themselves, could be used effectively to attack logical behaviorism. If the remarks made above concerning this sort of move are correct, however, then if this sort of move is adopted, the defender of logical behaviorism is faced with significant difficulties in defending his position. For if one accepts that two meaning-equivalent sentences must be about the same thing(s), then if (1) is true, it follows that sentences which we should ordinarily count as not about psychological phenomena are in fact about such phenomena, and sentences which we should ordinarily count as not about behavioristic phenomena are in fact about such phenomena.

Two sorts of moves seem open, at this point, to a defender of logical behaviorism who adopts the third of the strategies envisaged above. The first is to argue that we should not suppose that the way in which we would ordinarily determine what a sentence is about will do in the context of defending logical behaviorism. To adopt this move, however, is in effect to commit oneself to the need for constructing an alternative way of determining what a sentence is about. This move, like that discussed as the second of our possible strategies above, requires that some suitable way of determining what a sentence is about be formulated before we can be in a position to evaluate the merits of the thesis of logical behaviorism. In the absence of a means for determining what a sentence is about which diverges, in the required ways, from our ordinary intuitive way, it does not seem that our envisaged thesis of logical behaviorism has been made sufficiently precise to render it subject to effective evaluation. Since no suitable alternative way of making such a determination seems to suggest itself, it does not seem that this move can provide an acceptable strategy for a defense of logical behaviorism.

The other way of coming to terms with the difficulties encountered in the third of our envisaged defenses of logical behaviorism, is to claim that our ordinary intuitive way of determining what a sentence is about is, indeed, satisfactory, but that in spite of this sentences which are about thoughts and propositional attitudes really are also about behavioristic matters, and that many sentences which are behavioristic really are also about psychological phenomena. It is not clear, however, how this claim might be defended. For if we were to try to defend it by appeal to the truth of (1), then we would be begging the question which we had before us--namely, the question of providing a satisfactory evaluation of logical behaviorism. And if we were to reject any appeal to the putative truth of (1) in defending this claim, it is not at all clear what sorts of considerations we would be able to appeal to. It seems, therefore, that the thesis of logical behaviorism cannot adequately be defended by means of any of our three envisaged strategies. Unless, then, some alternative way of defending that thesis against the arguments presented above can be advanced, it seems that we are justified in rejecting logical behaviorism, as expressed by (1), as false.

If the above arguments are correct, then, as we have seen, it is unnecessary to appeal to any of Chisholm's claims in presenting an argument which will entitle us to reject logical behaviorism, and any appeal to Chisholm's claims seems, moreover, to be superfluous, because the considerations which would entitle Chisholm, given his more recent claims, to reject logical behaviorism, seem to presuppose the considerations which have led us, independently of Chisholm's claims, to reject logical behaviorism. Thus if it is not the case that meaning-equivalent sentences must be about the same thing(s), it seems unclear that any of Chisholm's recent claims are incompatible with logical behaviorism. If, on the other hand, meaning-equivalent sentences must be about the same thing(s), then Chisholm's claims are not needed in framing what seems to be an adequate argument in defense of rejecting logical behaviorism. Thus while Chisholm's more recent claims might be taken as defensible (as has been argued in Chapter II), it is not the case that these claims are incompatible with logical behaviorism except in conjunction with the additional claim that all meaning-equivalent sentences must be about the same thing(s).

It should be remarked that it is not intended that the above arguments should be taken as ruling out the possibility that there are, in fact, sentences which are both about psychological phenomena and about behavioristic phenomena. Thus, on Chisholm's view, it seems clear that a sentence like

George said "The cat is on the mat."

is, implicitly, about some psychological phenomenon, for semantical sentences are, on Chisholm's view, implicitly about psychological phenomena. But it is also clear that such sentences are also about the behavior of the person in question. Thus on Chisholm's view, there are a number of sentences which are both behavioristic and psychological. On the other hand. it does not seem to be the case that sentences about a person's beliefs, hopes, desires, thoughts and other psychological states and dispositions are, ordinarily, about any behavioristic phenomena. This possibility also does not seem to be ruled out, however; there could be sentences which are conjunctions of two sentences, the first of which is about a person's thoughts, say, and the second of which is behavioristic, and the resulting conjunction would intuitively seem to be both psychological and behavioristic. There are, however, clearly cases of sentences which intuitively seem to be about the behavior or propensities or dispositions to behavior of persons, but not about thoughts or propositional attitudes, and there clearly seem to be sentences which satisfy the converse condition. That there are such cases seems clearly all that is needed for the arguments presented above against logical behaviorism.

II. Logical Behaviorism (2): Sellars' Views

Before turning to a brief discussion of Sellars' views concerning logical behaviorism, it is worth taking note of a thesis which, though importantly different from the thesis expressed by (1), seems like the latter to warrant being called a type of behaviorism. Thus rather than requiring, as our thesis of logical behaviorism does, that every sentence about thoughts or propositional attitudes must be meaning equivalent to some sentence about behavioristic matters, we may require instead that

(3) For every sentence about thoughts or propositional attitudes there is some law of nature such that the conjunction of the two entails some sentence about behavior or propensities or dispositions to behavior, and such that the behavioristic sentence in question taken in conjunction with that law of nature entails the psychological sentence in question.

It seems clear that the thesis expressed by (3) is weaker than that expressed by (1), for (3), unlike (1), does not require that any psychological sentence be meaning-equivalent to some behavioristic one. On the other hand, if (1) is true, then (3) must clearly be, since if (1) is true than (3) is satisfied for every sentence about thoughts or propositional attitudes no matter what law of nature is chosen. It will be useful to call the thesis expressed by (3) the thesis of nomological behaviorism, in order to contrast it with logical behaviorism as expressed by (1).

The thesis expressed by (3) has two important advantages over the thesis of logical behaviorism. The first is that it seems clear that (3) is not subject to the sorts of difficulties raised against (1) in the paragraphs above. For it is clearly not required that two sentences which are such that each, in conjunction with some law of nature, entails the other, need be about the same thing(s). The second advantage is that it seems clear how (3), as against (1), might be defended, as a weakened (and possibly plausible) thesis of behaviorism. Thus what would be required in order to argue for the truth of (3) would be the discovery and formulation of certain suitable laws of nature. Similarly, it seems clear how one might try to argue effectively against the truth of (3); what would be required would be evidence that no laws of nature satisfying (3) could be discovered, or evidence that no such laws of nature are ever likely to be discovered.

An evaluation of the merits of this alternative thesis of behaviorism would take us far beyond the bounds of the present discussion. What is important for our purposes is that it seems clear that the thesis expressed by (3) is not the thesis against which it is reasonable to suppose that Chisholm is arguing. For not only does it seem clear that none of Chisholm's claims, either in "Sentences about Believing" or in his later works, would provide any reason for doubting the truth of (3), but it also seems clear that if one were to set out to attack such a thesis, one would have to make reference to the plausibility of the idea that there should, or that there should not, be any laws of nature of the sort mentioned by (3). But it seems clear that Chisholm nowhere expressed any such concern. We may conclude, then, that it is the thesis expressed by (1) which is under attack by Chisholm, and that although an alternative thesis of behaviorism along the lines indicated by (3) may be defensible, the thesis which is Chisholm's concern is not.

Let us turn, then, to Sellars' views concerning behaviorism. In "Empiricism and the Philosophy of Mind," Sellars distinguishes between two different views, both of which may be reasonably regarded as forms of behaviorism. These two views are called by Sellars methodological behaviorism, on the one hand, and philosophical, or analytical, behaviorism, on the other. [Sellars (5) 183-184] The second of these two views is characterized by Sellars as involving "the thesis that common sense psychological concepts are analyzable into concepts pertaining to overt behavior," [Sellars (5) 194, emphasis original] and this characterization seems to indicate fairly clearly that the view that Sellars is here discussing under the heading of 'philosophical behaviorism' is the view which we have called logical behaviorism. There is a temptation to think that the other view which Sellars discusses, what he calls methodological behaviorism, may be the thesis which we have expressed by sentence (3). But this is not accurate. For Sellars envisages the person who adopts methodological behaviorism as espousing the adoption of a certain maxim, namely, that "whether or not the mentalistic concepts of everyday life are definable in terms of overt behavior, . . . [the methodological behaviorist will] ensure that this is true of the concepts that . . . [he] shall employ." [Sellars (5) 184] What Sellars is calling methodological behaviorism does not, therefore, involve any appeal to the existence and possibility of discovering certain sorts of natural laws, but rather to the fruitfulness of setting certain restriction on the sorts of notions allowable in the formulation of a given area of scientific research.

In expanding on his notion of methodological behaviorism, Sellars makes clear that the restrictions which a methodological behaviorist would impose on notions allowable in the science of psychology are not to be conceived as ruling out certain sorts of theoretical concepts which might not be definable, strictly speaking, in terms of behavioristic notions. Thus in this way, the maxim quoted above is (as Sellars points out) somewhat misleading. For it should not be required that a methodological behaviorist may invoke no notions not definable in terms of behavioristic notions, but rather that he may invoke no notions unless they are "introduced in terms of a basic vocabulary pertaining to overt behavior." [Sellars (5) 185] Expanded in this way, it seems clear what the point of Sellars' notion of methodological behaviorism is. Since the theoretical notions invoked by Jones in the course of formulating his theory are notions whose original application is determined on the basis of behavioristic matters, Jones' theory will count as a theory which satisfies the requirements of methodological behaviorism. Thus if Jones' theory is acceptable, and if it is correctly regarded as yielding fruitful results, then this fact provides at least a partial justification of methodological behaviorism, seen as a sort of requirement to be set on scientific research which will provide acceptable and fruitful results. Since it has been argued that Jones' theory does, in fact, satisfy these conditions, we may regard methodological behaviorism as at least to this extent defensible.{5}

In the course of the passages cited, however, Sellars seems clear that he is rejecting what he calls philosophical, or analytical behaviorism--what we have called logical behaviorism. While no arguments in favor of such a rejection seem to be provided in "Empiricism and the Philosophy of Mind," Sellars does discuss the matter, although in brief, in a number of other locations. Thus, Sellars compares at length the irreducibility of psychological notions to behavioristic notions, which he espouses, to the putative irreducibility of ethical normative notions to factual descriptive notions (vide [Sellars (2) 47-58], and [Sellars (7) 85-88].), and compares this latter irreducibility to the irreducibility of semantical discourse to discourse involving neither semantical nor psychological resources, nor resources for talking about abstract entities{6} (video [Sellars (4) 445-453].). Similarly, in his correspondence with Chisholm, Sellars is clear that one way of looking at his rejection of the idea that psychological discourse is reducible (in the sense of analyzable into) behavioristic discourse, is to claim that the former involves a unique mode of discourse. Thus writing about one sort of intentional discourse, semantical discourse, Sellars claims that his 'means'-rubric,

'. . .' means ---.

"is the core of a unique mode of discourse, which is as distinct from the description and explanation of empirical fact, as is the language of prescription and justification." [Chisholm and Sellars 527]{7}

Sellars is also clear, however, that to say this much is not to present an argument, properly so-called, in favor of a rejection of behaviorism. Thus, in response to Chisholm on a certain point, he writes as follows:

I quite agree, then, that it is no more of a solution to our problem to say that semantical statements are "unique," than it would be a solution of the corresponding problem in ethics to say that prescriptive statements are "unique." What is needed is a painstaking exploration of statements belonging to various (prima facie) families, with a view to discovering specific similarities and differences in the ways in which they behave. [Chisholm and Sellars 531, emphasis original]

Thus Sellars seems to be explicit that on his view, we could not defend the view that psychological discourse is irreducible to behavioristic discourse without appeal to such a detailed examination of cases of sentences belonging to these two (prima facie) families of discourse.

In spite of the absence, in Sellars' writings, of such a detailed examination, Sellars' conviction that a rejection of logical behaviorism is warranted seem to be quite firm. In expressing this conviction, moreover, Sellars continually appeals to the idea that psychological and semantical discourse is simply a unique mode of discourse, as compared, say, to ordinary descriptive discourse involving facts which are not facts about intentional items. It is not difficult to see, however, how this sort of conviction is related to the argument which we have urged above in favor of the rejection of logical behaviorism. For if logical behaviorism is viewed as an assertion of (1), and if we regard sentences about thoughts or propositional attitudes as, in effect, about a different sort of subject matter from sentences about behavioristic matters, then it is just this difference in subject matter which will provide reason for a rejection of logical behaviorism. In connection with the argument which we have urged, we shall be led to reject logical behaviorism because of a conviction that if two sentences are equivalent in meaning, then they are about the same thing(s). Similarly, in connection with Sellars' way of presenting the matter, we shall be led to reject logical behaviorism out of a conviction that if two sentences are characterizable as about different sorts of things, then they are not analyzable, one into the other. Given Sellars' way of expressing the point, the one sentence will belong to a distinct mode of discourse, as compared with the family of sentences to which the other sentence, in virtue of its subject matter, belongs.

If this is a correct understanding of the basis of Sellars' conviction that logical behaviorism is false, then we may understand Sellars' views on this matter as reasonably reconstructible in terms of the argument presented above in rejection of logical behaviorism. Since Sellars' remarks on this subject are somewhat obscure and are not as explicit as the formulations of the considerations which led us, above, to reject logical behaviorism, it is difficult to be certain that such an understanding of Sellars' position is correct. Since, however, it seems unclear that there is any alternative understanding of the nature of Sellars' position that is as clear and as justified, we may regard such an understanding as acceptable for our present purposes. We may conclude, therefore, that Sellars' rejection of logical behaviorism, being implicitly founded on considerations similar to those invoked in our rejection above, is well founded.

III. Formulating a Viable Thesis of Physicalism

The thesis of physicalism may of course be formulated in such a way as to be parallel to the thesis, examined above, of logical behaviorism. If this is done, then much that has been said about Chisholm's and Sellars' rejection of logical behaviorism, as well as about our own reasons for such a rejection, may be said, mutatis mutandis, about such a thesis of physicalism. Thus the thesis of physicalism might be advanced as the claim that every psychological sentence means the same as some sentence of the physical sciences, possible of physics. If this formulation is adopted, then it is clear that the thesis of physicalism may be viewed as a broad version of the thesis of logical behaviorism, and if this is taken as granted, then logical behaviorism entails this formulation of the thesis of physicalism.

If this version of the thesis of physicalism is adopted, however, then it will conflict with our views in just the way that the thesis of logical behaviorism was seen to, in section I. For if a sentence of the physical sciences is identified as a sentence which is about physical phenomena --phenomena, that is, which are treated in the physical sciences--then psychological sentences will be about something different from that which sentences of the physical sciences are about.{8} But if sentences of the physical sciences are not to be picked out in this way, then before we can evaluate the merits of such a thesis of physicalism, we must first discover and evaluate some appropriate alternative way of picking out sentences as instances of sentences of the physical sciences.

The primary concern of this section is to formulate a version of the thesis of physicalism which would not be rejected on the basis of reasons such as those which led to our rejection of logical behaviorism. In particular, we shall be concerned to see whether an alternative thesis can be formulated which is, in suitable respects, weaker than the thesis examined above, and which can be viewed as defensible. The similarities between the thesis of physicalism which we shall be concerned to formulate and the thesis of nomological behaviorism expressed by sentence (3), will be obvious. In order, however, not to confuse the following discussion with what was said of nomological behaviorism above, we shall not refer to the latter thesis in what follows.

Intuitively, the following sort of claim might be proposed as a candidate for a plausible weaker version of the thesis of physicalism. We might claim that the language of the physical sciences alone is adequate to say whatever we wish to say about the world without any supplementation from language which does not belong to the physical sciences. Such a view seems to be what Cornman has in mind in framing a view which he intends to conflict with Chisholm's thesis [Cornman (1) 44], and also seems close to what Nagel has in mind as the view which he is discussing [Nagel 340]

The problem is that so expressed, physicalism seems not yet to be sufficiently rigorously formulated to admit of fruitful argument concerning its truth or falsity. For it is not at all clear what is involved in claiming that with certain linguistic resources at our disposal, we can or cannot say that we wish to say about the world. For one might wish to make the formulation of this second version of physicalism more explicit by simply making it equivalent to the formulation of the first version. If physicalism is to present any problems which are of a nature different from those presented by logical behaviorism, this move will of course not do. If, on the other hand, one wishes to provide a different means for making this version more explicit, then one seems to face becoming involved in debates in which one must become clear on what is intended by locutions such as 'S provides the same description of the phenomena as S. ', 'S describes the same phenomena as S. ', 'S cannot be used to describe phenomena in the same way as S. ', and so forth.{9} It is worthwhile, then, to search for an alternative means by which, independently of falling back on our first version, we can make the second version of physicalism sufficiently precise.

One way of so doing seems to involve a consideration of a certain sort of reductionist program, and, in particular, a reductionist program that might be suggested by a certain form of the identity thesis. It will be useful, then, to consider a certain form of the identity thesis, and the sort of reductionism that seems, on the basis of the relevant identities, to be suggested.

The identity thesis may be viewed as the thesis that it is not ruled out by logical or philosophical considerations that a result of future research should be the identification of mental phenomena of certain sorts with certain physical phenomena. In particular, one might hold that the identification would be between certain psychological states of a person and certain physical states of a person, as Nagel argues. The chief difficulty with such a thesis arises in connection with the problem of whether it is acceptable to suppose that the properties which psychological phenomena exhibit are the sort of properties which physical phenomena could exhibit, and vice versa. (Cf. [Smart (3) 92-105], [Smart (4) 163-171], [Rorty (2) 32-35], [Shaffer (1) 813-816] and [Shaffer (2) 160-162].) If the solution of this problem were a negative answer, then the identification in question would, to use Cornman's term, be a cross category identification. [Cornman (4) 492] Since, if this were the case, there would be certain properties which the denotata of the two sides of the identity statements could not share, the identity statements would have to be false. And on this basis, the identity thesis would have to be false, as formulated above, for on this basis logical or philosophical considerations would rule out the possibility that future scientific research could yield such identifications.

It is problems of this sort which will draw our attention in the fourth section of this chapter. For the time being, however, we are concerned rather with the type of reductionism which such a form of the identity thesis might be seen to suggest, in the hope of thereby arriving at a version of physicalism more precise than that expressed by our second formulation.

Since, on this view of the identity thesis, it is the results of future scientific research which warrant our making certain identifications, the following picture is suggested. It is clear that the identity statements in question constitute certain statements of coextentionality: certain statements to the effect either that this particular item is, in fact, identical with that, or that all items of this sort are, in fact, identical with one or another of the items of another particular sort. The identification in question, whatever its form, will be a contingent identity, in the sense that it could be the case, as a matter of fact, that the identities did not hold, and, in particular, it could be the case that the results of future scientific research should yield not these but different identifications, or, perhaps, none at all. For this reason, it is clear that no meaning-equivalence is involved in the identifications, and thus that objections of the sort which Chisholm has raised to the first sort of physicalism and to logical behaviorism would not, thus far, apply.

Since, however, we are supposing that the identifications are warranted by certain results of future scientific research, we may wish to claim more for them than we do for identities such as those between Cicero and Tully. In particular, we may wish to claim that certain statements are not only true, and contingently so, but true on the basis of laws of nature, or lawlike. The force of such a claim need be no more than to assert that the identifications are, in some appropriate sense, well established and reliable in roughly the degree to which certain natural laws are.

It is important to see just what sort of statements might be regarded as well established in this way. It is compatible with the formulation of the identity thesis which we have been considering that the only sort of identifications which we might be able to come to make are between certain individual psychological phenomena and certain physical phenomena. If this were so, then we would be unable to make identifications between such phenomena which relied upon certain general features of each sort of phenomenon. We would be unable, that is, to frame true identity statements of the sort which would assert that any psychological phenomena of this specifiable sort is identical to a physical phenomenon of that specifiable sort.{10} If we were able to establish the latter sort of general identity, as opposed to certain particular identities, then the general identity statements would be, on the view presented, well established to roughly the degree that certain natural laws are. If that were the case, then it seems open to suggest that we could take the class of sentences which describe psychological phenomena of a certain sort, s, and replace them by sentences which describe no more than the physical phenomena to which, according to our general identity statements, the psychological phenomena in question are identical. And although this move would not be warranted by any meaning-equivalences, it would be warranted in the way in which the assertion of certain natural laws is warranted.

Given all this, we might provide a more precise formulation of the second version of physicalism in the claim that such psychological sentences, sentences, that is, about the psychological phenomena which figure in the well established general identifications, are replaceable by non-psychological sentences which describe the physical phenomena which are identified with the psychological ones. Although the replacing sentence would, of course, have the same truth value as the original sentence, the meanings of the two sentences would not necessarily be the same; sameness of meaning would simply not be a consideration, given this version of the thesis of physicalism. Since, however, the general identities are well established in roughly the way that laws of nature are, we might claim for the replaceability something more than the mere extensional equivalence of two sentences which differ only in the occurrence in each, in the same purely referential position, of two referring expressions which independently of any nomological matters refer to the same thing. It will be useful to refer to such equivalence between sentences, which is weaker than meaning equivalence but stronger than mere extensional equivalence, as nomological equivalence.

To be certain of being clear on what is involved in this formulation of the thesis of physicalism, it will be helpful to be more specific in several respects. Let us suppose that the identity thesis is true in the version presented, and that certain identifications, which constitute what we may call an identity theory,{11} have been established. Let us further suppose that the identifications generated by our scientific results include a large number of general identities. We may consider the following as an instance of such a general identity

(4) (x)(x is a psychological phenomenon of sort s --> (Ey)(y is a physical phenomenon of sort s' & x = y)) . (x)(x is a physical phenomenon of sort s' --> (Ey)(y is a psychological phenomenon of sort s . x = y)){12}

(4) is, then, accepted as both true and well confirmed as the consequence of certain scientific laws, although it is not claimed that it is anything like analytic. If, then, we have a sentence which means the same as

(5) George exhibits a psychological phenomenon of sort s.

we may, on the basis of our thesis of physicalism, replace this with

(6) George exhibits a physical phenomenon of sort s' .

with the guarantee that the replacing sentence has the same truth value as the original. It will be the equivalence between sentences (5) and (6) which we shall call a nomological equivalence.{13}

The thesis of physicalism with which we began our discussion in this section involved the idea that every psychological sentence should be meaning-equivalent to some non-psychological sentence. As we have developed our weaker thesis of physicalism, no question of the meaning-equivalence of such sentences need be raised, for the thesis of physicalism may now, as revised, be constructed on the idea that every psychological sentence should be simply nomologically equivalent to some non-psychological sentence. A formulation of the thesis of physicalism which rests on nomological equivalence seems to provide a way of spelling out what might be involved in the claim that physicalism involves the idea that we should be able to say all we might wish to say about the world by means of no more than sentences belonging to the language of the physical sciences, and it seems to provide a way of spelling this out without invoking any notion of meaning- equivalence. For in any case where we wish to describe a phenomenon by using some psychological sentence, the nomological equivalence of that sentence to a certain non- psychological sentence belonging to the language of the physical sciences suggests that we might be able to use that non- psychological sentence instead. As a first approximation to a formulation of this revised thesis of physicalism, then, we may offer the following:

The Revised Thesis of Physicalism:
We may say all we might wish to say about the world by using no more than sentences belonging to the language of the physical sciences; and the sentences which, although belonging to the language of the physical sciences, we may use to say what we wish to say about psychological phenomena, will be nomologically equivalent to psychological sentences which do not belong to the language of the physical sciences.

For the present, we may let this formulation stand, although we shall be somewhat more rigorous in formulating this revised thesis of physicalism in the course of section VII.

There is one point concerning this initial formulation of a revised thesis of physicalism which deserves special attention. We have characterized our general identities as involving identifications between psychological phenomena of certain sorts and physical phenomena of certain sorts, and on this basis, supposing our identities to be well established in roughly the way that laws of nature are, we have characterized the revised thesis of physicalism in terms of a nomological equivalence between psychological sentences and physical sentences, that is, sentences belonging to the language of the physical sciences. Since this nomological equivalence is conceived of as being established on the basis of certain general identities, it will seem most reasonable to suppose that whatever the psychological sentences in question are about, the nomologically equivalent physical sentences will be about also. If this is correct, then if following Chisholm we are to classify a sentence as psychological if it is about some psychological phenomenon, the thesis of physical will commit us to the idea that certain sentences of the physical sciences are psychological sentences. For this reason, it will be useful to be somewhat more explicit concerning the sorts of sentences which, according to our revised thesis of physicalism, are nomologically equivalent. The following formulation seems appropriate for our present purposes: we may speak of every sentence which, as things stand now,{14} is taken to be about some psychological phenomenon as being nomologically equivalent to some sentence which belongs to the language of the physical sciences.

The revised thesis of physicalism which has been described rests on the idea that we should be able to establish general identifications of the sort which the identity thesis claims are not ruled out on the basis of any philosophical or logical considerations. Thus our revised thesis of physicalism cannot be true unless the requisite version of the identity thesis is also true. On the other hand, it clearly seems open, at this stage of the discussion, that we should discover that the identity thesis is acceptable and nonetheless reject our revised thesis of physicalism. Thus we might regard the identity thesis as acceptable, and nonetheless discover that in spite of the general identifications we might establish, we could not find natural laws of the sort required to generate the sorts of nomological equivalences required for the truth of the revised thesis of physicalism. For this reason, preparatory to our discussing the revised thesis of physicalism in greater detail, it will be useful to discuss the merits of, and possible objections to, the identity thesis. It is to this task that we shall turn in the following three sections of the current chapter, returning, in sections VII and VIII, to a discussion of the merits of the revised thesis of physicalism itself.

In particular, it will be part of our concern in section IV to argue that there is nothing in the views of either Chisholm or Sellars which necessitates a rejection of the version of the identity thesis on which the revised thesis of physicalism rests, and that Sellars' views in fact seem to presuppose the truth of the version of the identity thesis in question. If these arguments are correct, then if either Chisholm or Sellars is to reject the revised thesis of physicalism, it will have to be on some other basis than an objection to the acceptability of the identity thesis. Sellars' views concerning our revised thesis of physicalism will be discussed toward the end of the final section of this chapter, after the merits of that thesis have been discussed in some detail. Granting in advance what we shall argue in the next section, that nothing in Chisholm's views is incompatible with the identity thesis, we may in closing the current section discuss whether there is anything in Chisholm's views which would be incompatible with our revised thesis of physicalism, aside from some possible incompatibility involving the presupposed identity thesis.

Because the revised thesis of physicalism does not involve the claim that sentences which, given the current state of empirical knowledge and our current uses of language, are taken to be about psychological phenomena are equivalent in meaning to sentences which are not, it seems clear that the sorts of objections which we imagined Chisholm to raise against logical behaviorism and the stronger thesis of physicalism will not, as those objections stand, be applicable here. Nonetheless, it seems clear that Chisholm is concerned to reject the idea that we can say all we wish to say about intentional phenomena by using no more than terms which are required for the language of the physical sciences. Thus, as we have seen in section I, Chisholm writes that "when we analyze the kind of meaning that is involved in natural language, we need some concept which we do not need in physics or in 'behavioristics.'" [Chisholm and Sellars 523] We may formulate our concern, then, as follows: Is there anything in the claim expressed by sentence (2) which is incompatible with the truth of the revised thesis of physicalism?

The revised thesis of physicalism asserts that we can say whatever we wish about any phenomena whatever by using no linguistic resources other than those required in formulating the physical sciences. This claim is clearly not susceptible of being evaluated, however, unless we have at least some moderately clear intuitive idea of what linguistic resources are (or might be) required in formulating the physical sciences. In order, however, to discuss the relationship between this thesis and Chisholm's thesis (as expressed by sentence (2)), it is not required that we become more specific concerning the language of the physical sciences, since both theses make reference to the linguistic resources required for the physical sciences. It is merely required, therefore, that we be clear that in both cases we are concerned with the linguistic resources required to formulate that area of empirical knowledge which practitioners of the physical sciences are (or would be) prepared to count as part of their discipline.

Chisholm's thesis, as expressed by sentence (2), requires that there be a class, C, of terms, such that if we are adequately to describe some psychological phenomenon, we must use at least one of these terms. The revised thesis of physicalism requires that we should be able to describe any psychological phenomenon adequately using no terms other than those necessary for the physical sciences. Chisholm's thesis also requires, however, that no member of C is necessary for the physical sciences. These three requirements, taken together, seem to be incompatible with the claim that there are any terms which can, in fact, be used adequately to describe psychological phenomena. For if there are any such, then Chisholm's thesis requires that none of them be necessary for the physical sciences (nor interdefinable with any such), while the revised thesis of physicalism asserts that at least some of them do belong to the class of terms necessary for the physical sciences. Thus if we take for granted the idea that we can adequately describe psychological phenomena, then Chisholm's thesis is incompatible with the revised thesis of physicalism.

Clearly, an understanding of what it is to give an adequate description of something is crucial for an understanding of what is involved here. We might wish to avoid this problem by speaking in terms of our ability to say all we might wish to about same sort of phenomenon, instead of in terms of giving an adequate description of it. But this move will not help. For then, it seems, we should have to reformulate Chisholm's thesis to require that there is a class of terms such that if we are to say all we might wish to about psychological phenomena, then we must use at least one of these terms, none of which are required for the physical sciences; while the revised thesis of physicalism would require that we can say all we might wish to about such phenomena using no more than terms required for the physical sciences. Given this formulation, however, we would still be left with the problem of giving an account of what is involved in our being able to say all we night wish to about a given type of phenomenon.

The importance of this point for our present concern is that it is of course possible that the conditions which Chisholm might impose if we are correctly to say that we can, given certain linguistic resources, adequately describe or say all we might wish to about certain phenomena might differ from the conditions which a defender of the revised thesis of physicalism might impose. In particular, we might imagine a defender of physicalism being prepared to settle for less stringent requirements than Chisholm. If this were indeed the case, then it does not seem that there would be any incompatibility between Chisholm's thesis and physicalism, for what Chisholm would be claiming in asserting that we can adequately describe or say all we might wish to about psychological phenomena by using certain linguistic resources would simply be different from what the defender of physicalism would be claiming in making (what superficially appears to be) an assertion of the same form.

If a defender of physicalism, therefore, wished to make his thesis compatible with Chisholm's, then we may imagine such a defender of physicalism holding out for a sense of 'adequately describe' or 'say all we might wish to' which is weaker than that advanced by Chisholm. Thus it would seem that in the strongest sense of 'adequately describe' and 'say all we might wish to', we might be said to be able adequately to describe, or say all we might wish to say about a certain sort of phenomenon given certain restricted linguistic resources if, and only if, any sentence which though not formulable given only those restricted resources might be used to describe such phenomena is equivalent in meaning to some sentence which can be formulated using nor more than those restricted linguistic resources. Similarly, weaker senses of 'adequately describe' and 'say all we might wish to' could be generated by means of appeal to weaker notions of equivalence. In particular, one sort of weaker sense of 'adequately describe' and 'say all we might wish to' might rest on the idea of nomological equivalence which was invoked in our formulation of the revised thesis of physicalism Thus if one set out to show that physicalism is, in fact, incompatible with Chisholm's thesis, it would have to be shown that there is a sense of these expressions which Chisholm can adopt, such that adopting any weaker sense would render physicalism indefensible. It is unclear, however, that there is such a sense, and, in any event, it is unclear that Chisholm would wish to adopt a sense of these expressions which was itself sufficiently weak to generate this result.

The foregoing argument is not, of course, intended as a counter to Chisholm's claim that his thesis is, in fact, incompatible with (at least some versions of) physicalism. What has been argued, rather, is that it is not by any means clear that there are not plausible formulations of physicalism with which nothing that Chisholm would wish to claim is incompatible. In particular, it seems that there is no reason to suppose that anything which Chisholm would wish to claim would be incompatible with what we have called the revised thesis of physicalism, construed as involving a sense of the expressions 'adequately describe' and 'say all we might wish to about' which is weaker than the sense which Chisholm's thesis would invoke, and nonetheless sufficiently strong to generate a plausible and interesting version of physicalism. Thus if the foregoing arguments are correct, then there is no reason to suppose any incompatibility between Chisholm's views and the revised thesis of physicalism unless there is reason for Chisholm to reject the version of the identity thesis required for the revised thesis of physicalism. We shall therefore turn at this point to a consideration of the identity thesis, as discussed earlier in this section.

IV. The Identity Thesis (1) Vis-a-vis Intentionality

Because of the dependence which we have noted of the revised thesis of physicalism on the identity thesis, it is important to become clear whether this latter thesis is compatible with the views advanced by Chisholm and Sellars. For this reason, the current section will be devoted to an examination of this question, and sections V and VI will take up the question of whether, independent of the specific views of Chisholm and Sellars, the identity thesis is defensible. Thus matters relating to the merits of, and possible objections to, the identity thesis will be raised in this section only if they derive from considerations connected with either Chisholm's or Sellars' position on intentionality.

The version of the identity thesis which we have considered is along the lines of that defended by Smart and Place. (Cf. [Place], [Smart (3)] and [Smart (4)].) A specific point at which it differs, however, is that in our discussion the identity thesis is not, as it is in the writings of Smart, Place, Cornman, Rorty and others, restricted to matters pertaining to those mental events which are called sensations. Since whether sensations, which are generally regarded as non- conceptual mental phenomena, are intentional is a controversy related to those which we are considering,{15} and since it is primarily psychological phenomena which both Chisholm and Sellars would agree to be intentional with which we are concerned, we shall avoid so restricting the identity thesis. In particular, since sensations are not primarily our concern and thoughts are, we shall be concerned with the identity thesis as it applies to thoughts, whether or not it applies also to sensations.{16}

The thesis we are considering, then, may be formulated as follows:

The Identity Thesis:
There are no logical or philosophical considerations that rule out the possibility that it should be a result of future scientific research that psychological phenomena, in particular the occurrence of thought episodes, should be strictly identical with certain physical phenomena. {17}

We may then consider two sub-versions of this thesis, the identity simpliciter, that the identities which are under consideration include a substantial number of what we have called general identities, and the particular identity thesis, that the identities yielded should consist only of particular identities.

So long as it is clear that the identity statements involved are not intended as anything like analytic, and it seems clear that on all versions of the identity thesis this condition is satisfied, it does not seem likely that Chisholm's thesis of intentionality will in any way conflict with the identity thesis. For it surely is open for psychological terms to comprise a family which is closed in the sense discussed at the close of Chapter II (cf. sentence (2), above), and nonetheless for it to be empirically warranted to identify the psychological phenomena which psychological sentences are about with certain physical phenomena--physical in the sense that the phenomena in question are cases of phenomena which are described by the language of physical science.

As we have noted in the foregoing section, the idea that the identifications involved in the identity thesis are cross category identifications--that the items picked out by one of the terms of the purported identification are in some appropriate sense "categorially" different from the items picked out by the other--is perhaps the most important objection to the truth of the identity thesis. Thus unless this sort of objection can be dealt with in some effective way, the identity thesis must, it seems, be rejected. Our current concern, however, is more restricted than a defense of this thesis. Thus while we shall have to come to terms with this objection in the discussion which follows, for the time being we are concerned only with the question of whether this sort of objection follows from considerations involved in Sellars' or Chisholm's positions. In particular, it will be useful to see whether this sort of objection is involved in anything which would follow from the truth of Chisholm's thesis.

It will be useful here to consider a specific case, and in particular a case of a general identity, in order to see what is involved in the idea that such an identity might be cross-categorial we may let the following be our identity statement:

(7) (x)(If x is an occurrence of the thought that it is raining then x is identical to some pattern of firings of nerve firings satisfying condition C) and (x)(If x is a pattern of nerve firings satisfying condition C then x is identical to some occurrence of the thought that it is raining)

Now patterns of nerve fibers firing may be described in mathematical terms as rapid or slow, as involving certain potentials of electric charges, as requiring certain threshold potentials, and so forth; while occurrences of the thought that it is raining may be described as true or false, as warranted or not, as cogent or not, and so forth. But in both cases, it does not appear that the ways in which we might describe the thought that it is raining would be applicable to patterns of nerve firings, nor do the ways in which we might describe these latter seem applicable to thoughts.

It is of course clear that there are certain predicates which apply to the one sort of thing as well as to the other: both sorts of phenomena nay be located in time, may have a certain duration, may be said to be states of persons or properties of persons, both may be embedded in suitable causal frameworks, and so forth. Thus it might be open to argue that for the purposes of defending a satisfactory version of the identity thesis--say, a version which reflected traditional materialistic views in philosophy{18}--it need not be required that the identifications in question be constructed so as to require that identical items have all their properties in common, but only that they have certain specifiable sorts in common. Thus Nagel suggests that the identification of psychological with physical phenomena should require not that identified phenomena have all their properties shared, but a certain proper subset, and he argues that the properties which we must consider in the attempt to establish such identifications are those which would be considered in establishing identifications which are embedded in and dependent on some scientific theory or theories. [Nagel 345-350] For such a proposal to be satisfactory, it is of course necessary that we be given some means of telling, independently of which sorts of properties are (as things stand) ascribed or ascribable to both sorts of things, which sorts of properties must be ascribed to both if we are to establish this weaker "theoretical" sort of identification. Sections V and VI of the present chapter will be concerned to discredit the sorts of objections which, in being advanced against the identity thesis, have motivated the suggestion that a weaker sort of identity than so-called strict identity must be invoked in defending a satisfactory version of the identity thesis. For this reason, it does not seem that it will be necessary that we consider proposals such as that advanced by Nagel. We may thus return to our primary concern, namely, the relationship of the identity thesis (seen as involving the possibility of strict identifications) to Chisholm's and Sellars' views.

In the case of the identity thesis as applied to thoughts, it seems clear that the properties which Chisholm would wish to insist are not ascribable to physical phenomena but are ascribable (and in some essential way) to thoughts are properties of being about this or that. An occurrence of the thought that it is raining might be said to be about the weather, for example, but it seems clear that it is not the case that we can sensibly claim that a certain pattern of neural firings is about anything whatever. While this result would still not conflict with the thesis of intentionality as defended by Chisholm, it does seem to conflict with the view that if two things are identical then they have all their non-modal properties in common.{19}

There is one way in which it might seem, however, that Chisholm's thesis of intentionality might conflict with the identity thesis so described. For while the problem of cross-categorial identifications seems to be independent of the sorts of considerations involved in Chisholm's thesis, this thesis does involve, in one form, a claim about what certain sentences describe. If our identity statement (7) is supposed to be true, then both

(8) George has the thought that it is raining.

and

(9) George exhibits a pattern of neural firings satisfying condition C.

would, as a matter of fact, be about the same thing, namely some one item satisfying the conditions specified by the two open sentences following the quantifiers in (7). If this were true, then we could use the non-intentional sentence (9), containing no technical terms, to describe some psychological phenomenon. This clearly runs counter to the thesis as stated in "Sentences about Believing."

One move which could be made if we wished to avoid this apparent conflict between Chisholm's thesis and the identity thesis would be as follows. We might deny that any clear sense can be made of the idea that a given sentence is or is not descriptive of this or that sort of phenomenon, thereby leaving it open to claim that even though (9) is about whatever (8) is about, nonetheless (9) need not be held to describe whatever (8) describes. In particular, one might hold that although (9) is about a given psychological phenomenon, namely the one described by (8), it does not describe that phenomenon.

If Chisholm's thesis is formulated in the way expressed by sentence (2), these difficulties do not seem to present themselves. For if the thesis is to the effect that we cannot avoid using any given member of the circle of psychological terms without using another in its stead, we have avoided any reference to the matter of what sentences describe and are about. We may be explicit in formulating our thesis by reference to meaning equivalence, that is, by using meaning equivalence as our means of determining whether we have, in a given case, successfully avoided using a given term. And we may take as initial members of our class of psychological terms certain terms which seem clear cases of psychological terms in a given language. Thus the application of the thesis, so formulated, is simply a problem of determining when some sentence containing such a term is meaning-equivalent to some other sentence, and when not. We need not, on this formulation, be concerned with what sentences describe which phenomena and what sentences do not For this reason, the move suggested above, involving our allowing (9) to be about a psychological phenomenon which it nonetheless does not describe, seems to capture Chisholm's intent. For if the formulation of Chisholm's thesis presented at the close of Chapter II does, as has been argued, provide an accurate account of Chisholm's intent, then it is reasonable that given his original formulation in terms of what sentences describe, we should be able to avoid any problems avoided by the later formulation. In particular, since this later formulation does not conflict with the identity thesis in the fashion under discussion, it is reasonable that we should seek to avoid such conflict as it might be generated by the original formulation.{20}

It appears, then, that there is no conflict between Chisholm's thesis of intentionality and the identity thesis. For although the problem of cross category identification must be met if we are to defend the latter, this problem is independent of Chisholm's thesis; and the problem concerning what sentences describe can be avoided, either by providing a different formulation from Chisholm's which nonetheless seems to be adequate to his intent, or by taking a certain stance regarding the putative counterexamples generated by the identity thesis.

Sellars' position on the identity thesis seems, however, to be more involved. In a number of places ( [Sellars (5) 193], [Sellars (14) passim], [Sellars (10) 32-37], and [Sellars (8) 669] ), Sellars seems clearly to reject the identity thesis, although he clearly wishes, in these same passages, to argue that something like what he calls the identity thesis is true. It will be helpful to begin by considering the identity thesis in relation to Sellars' insistence that intentional items must, in order to play their intentional role, exhibit some determinate factual character. Thus on Sellars' view, if a thought is to be truly characterizable in terms of the intentional role(s) it plays, then it must be truly characterizable also in terms of some empirical characteristics, if only in terms of certain causal relations it has to empirically identifiable phenomena such as overt behavior. If this is so, however, then it seems to be the case that it is open to discover that the intentional item referred to in a given case, say, by sentence (8), is the same as a certain empirically specifiable episode which has the causal properties truly ascribable to the thought in question. And it seems reasonable to suppose that such an episode might be one which is truly describable by a sentence such as (9).

So far as Sellars' position on intentionality is concerned, then, it does not seem to conflict with the identity thesis as we have formulated it. Something like the identity thesis must, in fact, be true if Sellars' position on intentionality is. For if it is a requirement for something to play an intentional role that it have empirical properties in virtue of which it can, independently of an intentional role which it may play, be identified as of a certain sort, then it must be the case that all intentional items can be described in virtue of that determinate factual character.

Let us consider the case of a particular intentional item, say a thought, which plays an intentional role like that played by the English sentence 'It is raining.'. For any item to play such a role, it must exhibit certain empirical features in virtue of which it is independently identifiable as of a certain sort. Now supposing that there are no sentences other than the English 'It is raining.' which play exactly the role which tokens of that sentence-type play, we may conclude that we have a fairly precise account of what the determinate factual character which items playing precisely that role in spoken language have in common. For the determinate factual character in question is simply the class of physical properties which individuate sentences as tokens of the 'It is raining.' type.{21}

Now the parallel question is suggested, for the case of our thought episode, as to whether the same sort of results might be generated. In particular, could we, if we supposed ourselves restricted, say, to the thoughts of persons who are natively English speakers and have no command of any other spoken language, specify a class of, say, neurophysiological characteristics adequate to pick out all and only occurrences of the thought in question?{22} The class of characteristics could be put in the form of a disjunction, and could be imagined as immensely complex. But the question is whether it would be possible to find such a class at all. If we could not, then, at least for cases of occurrences of the thought that it is raining, we should be able to specify each occurrence in terms of empirical characteristics and independently of the intentional role played, but the only characteristics which all occurrences would have in common would be that they each played the same intentional role. In this case, we should be able, with regard to this particular thought, to establish many particular identities with, say, occurrences of patterns of neural firings, but we should be unable to establish any general identity of that sort.{23}

In the case of a thought episode, the Jonesian theory on the basis of which Sellars wishes to construct a notion of a thought requires that the episode exhibit the following features. It will be a state of a person, it will play a certain specifiable intentional role, one analogous to the role played by some specifiable overt verbal episode, and it will be causally linked (initially solely as a theoretical entity by means of observational evidence and certain theoretical inferences) to observable behavior. The fact that we are able to classify a certain thought as being, say, one which plays a role analogous to that played by the English 'It is raining.', depends upon its being causally linked to observable behavior patterns which involve the occurrence of the overt verbal episode 'It is raining.' and of others which are suitably related. After a certain stage of development, we are able to pick out occurrences of the thought without reference to observational matters, by a process of noninferential reporting. But this non-inferential reporting rests, so Sellars contends, on the causal connection between a given thought and the thought that one is having that thought.

If we adopt this picture, then we have in effect accepted the idea that every occurrence of the thought that it is raining, in addition to sharing with every other such thought the intentional property of playing a certain role, must also share with every other such thought the property of being involved in certain specifiable causal connections with observable events. For unless we accept this, we are deprived of any means of picking out a given case of a thought playing a given intentional role: this identifying is possible because we can pick out cases in which events which are supposed to cause such a thought are present, and cases in which events which are supposed to be caused by such a thought are present. If we adopt Sellars' views concerning the nature of thought episodes, then, it is required that we accept the idea that thought episodes which play the same intentional role have certain causal properties in common, where a causal property is understood as the property which a thing is asserted to have by the statement that it bears a certain causal relationship to things of a certain specifiable sort. If, moreover, we accept the idea that no two things have causal properties in common unless they have certain empirically determinable noncausal properties in common, then we are led to accept the idea that thought episodes which play the same intentional role also have in common certain empirically determinable properties which are independent of the fact that the episodes all play a specifiable intentional role.

Several remarks may be made about this line of reasoning. First, it seems clear that the assumption that if two things have causal properties in common then they have certain empirically determinable non-causal properties in common, although plausible, is not itself an assumption which follows from any part of Sellars' position on intentionality. For as we have noticed, Sellars' position seems compatible with the idea that thoughts are states of, say, a Cartesian soul, and in this case, while thoughts would nonetheless be supposed to exhibit some determinate factual character inasmuch as they exhibit certain empirically determinable causal properties, they might be conceived of as exhibiting no other empirically determinable properties whatever. It seems clear, however, that unless there is some reason to reject this assumption, it seems a reasonable one to make. Secondly, it seems clear that the only empirically determinable properties which a thing exhibits which might be relevant to determining what intentional role (if any) it plays are its causal properties. For on the basis of Jones' theory, it is the causal properties of thoughts in relation to items of overt behavior which allow them to play some intentional role: similarly in the case of semantical episodes, it is the causal properties of such episodes to other items of overt behavior which allow those episodes to play an intentional role. Thus if we are, along the lines of the foregoing paragraph, able to conclude that intentional items which play the same intentional role also have certain empirically determinable non-causal properties in common, then we are warranted in concluding that such items have in common certain empirically determinable properties which are independent of their intentional properties--independent, that is, of the fact that they play the intentional role they do.

It seems clear, however, that we do not need to adopt quite so strong an assumption in order to defend the conclusion that intentional items which play the same intentional role have in common certain empirically determinable properties which are independent of their intentional properties. Thus rather than assuming that if two things have certain causal properties in common then they have certain empirically determinable non-causal properties in common, we might instead argue that it is implausible to suppose that the causal properties of an intentional item which are involved in the item playing its intentional role exhaust the totality of its empirically determinable properties. This claim seems clearly warranted in the case of semantical episodes, for it is clear that such episodes have causal effects, as well as other empirically determinable properties, which have no bearing on the intentional role played by such episodes. In the analogous case of thought episodes, it seems reasonable to expect that if such episodes do exist--if, that is, Jones' theory is a good one--then we may expect to be able to discover empirically determinable properties (whether causal or not) which are independent of the intentional roles played by such episodes; empirically determinable properties, that is, which are not imputed to thoughts by Jones' theory. For if we are unable to assign any properties to thoughts independently of those properties which Jones' theory imputes to thoughts, then it is unclear how we might be able to test the adequacy of Jones' theory, in the way in which we ordinarily suppose that theories are testable. For in that case, we would be unable to make any true statements about thought episodes other than those which are warranted by Jones' theory, and we would be unable to determine whether or not a certain thought episode had occurred independently of whether or not we were warranted on the basis of Jones' theory in saying it had occurred. Thus it is not required that we assume that items having causal properties in common must therefore have at least some noncausal properties in common in order to draw the desired conclusion;{24} nor is it required, in order to draw this conclusion, that we rest our argument on the somewhat questionable distinction between empirically determinable properties which are causal, as opposed to empirically determinable properties which are not.

If such a class of (either causal or non-causal) characteristics is, in principle, specifiable such that all and only occurrences of the thought that it is raining will satisfy the class, then it is possible, on that basis, to establish a general identity statement of the form of sentence (7). For if we were restricted, no matter what the results of future empirical research, to particular identities of this or that occurrence of the thought that it is raining with one or another physical phenomenon, then we should in principle be unable ever to have come to be able to pick out cases of the thought that it is raining in the way Sellars supposes we have. If Sellars' position on intentionality is acceptable, then, it follows that we must accept a version of what we have been calling the identity thesis simpliciter.{25}

If the above argument is correct, then while Chisholm's position as regards intentionality is independent of and compatible with the so- called identity thesis, Sellars' position on intentionality requires that some form of the identity thesis be acceptable. Thus, so far as Chisholm's position was concerned we were not required to offer a defense of the identity thesis, for in order that Chisholm's views be correct it was not required that the identity thesis be correct. In particular, we did not find it necessary to consider the argument that the identity thesis involves claiming that certain identities, which are cross-categorial, could be established. In order for Sellars' position to be acceptable, this objection to the identity thesis must be overcome. If this objection can be dealt with effectively, then we shall be in a position to consider the revised form of the thesis of physicalism which was formulated in the foregoing section.


[Table of Contents][Continue to Rest of Chapter 5]

Notes

{1} This conclusion seems to be close to that drawn by Putnam. Thus Putnam writes as follows: "Psychological states are characterizable only in terms of their relations to each other (as well as to behavior, etc.), and not as dispositions which can be 'unpacked' without coming back to the very psychological predicates that are in question." [Putnam (5) 70] For Putnam's rejection of logical behaviorism seems to rest, like Chisholm's, on the idea that psychological terms form a closed family of expressions with respect to terms which are non-psychological in some appropriate sense. Putnam's conclusion, however, is in no way defended, as Chisholm's is, by appeal to something like a thesis of intentionality. [Back]

{2} It was argued, in fact, that as things stand in the English language (and related languages), all contingent psychological sentences will be intentional by either of these two recent definitions of intentionality. Chisholm, however, neither makes nor attempts to defend or to consider such a claim, and for our purposes it is sufficient to grant simply that there is at least one sentence which is both about psychological phenomena (thoughts or propositional attitudes) and is also intentional. [Back]

{3} It seems clear that so-called indexical sentences, like 'I threw the ball to him.', present problems with regard to such a claim. For one might wish to assert that two uses of such a sentence in circumstances in which the two occurrences of 'I' and 'him' referred to different persons, might be occurrences of the same sentence- type, and thus that the two sentence-tokens, although not about the same things, would nonetheless mean the same thing. Since the issues involved in such examples are both controversial and irrelevant to the argument under consideration, we shall side-step this sort of example by restricting our consideration to sentences which are not indexical in this way. [Back]

{4} It seems clear that the following formulation might be offered, as a formulation of the thesis which certain philosophers, possibly Chisholm among them, might give of the thesis which they are concerned to evaluate:

(1') Every sentence about thoughts or propositional attitudes can be translated, without remainder, into a sentence which is about behavior or propensities or dispositions to behave.

Such a claim would, it seems, ordinarily be understood as involving the idea that the behavioristic sentence into which a given psychological sentence can be "translated" without remainder satisfies two conditions. First, it is meaning equivalent to the psychological sentence in question: and second, it does not involve any psychological component. It seems clear, that if (1') were to be adopted as an alternative formulation of logical behaviorism, then if it is to be acceptable it must be spelled out just what is involved in the idea that the behavioristic sentence is a translation with no remainder--that is, that it has no psychological component. Since on the face of it, it would seem that no behavioristic sentence could not be equivalent in meaning to a psychological sentence unless the former had, as part of its meaning, that it is about some psychological phenomenon, it seems unclear how such an alternative formulation could be defended against the difficulties which, it has been argued, face (1) as a formulation of logical behaviorism. For this reason, it does not seem that (1') exhibits any advantages, as a formulation of logical behaviorism, over (1), and therefore the former will not be discussed in what follows. [Back]

{5} Cf. [Ryle 327], at which location Ryle characterizes behaviorism along similar lines, as having originally been "a theory about the proper methods of scientific psychology." Ryle goes on to find fault with such a view, at least as it has been applied by empirical psychologists in recent centuries. The point of Sellars' notion of methodological behaviorism is, however, intended to apply to theoretical and conceptual innovations of the sort which might originally have led to our present-day kind of psychological discourse, and not to apply to scientific enterprises undertaken, in recent times, on the basis of such psychological discourse. [Back]

{6} On Sellars' view, discourse about abstract entities is to be understood as discourse which is implicitly about certain classes of intentional items, items, that is, which may be appropriately described as being about something. (Cf. [Sellars (1) passim].) Since this view does not relate directly to our present concerns, we shall not discuss it here. [Back]

{7} Thus compare Sellars' claims that the concept of a dot-quoted item (an item which plays the same role as the linguistic expression exhibited within dot quotes does in our language) "ultimately involves (as does the concept of a pawn) the concept of what it is to satisfy a norm or a standard, the point of the norms or standards pertaining to conceptual 'pieces' . . . [being] to bring it about that as items in the natural order they picture the way things are." [Sellars (8) 664, emphasis original] This sort of claim seems to indicate the fact that on Sellars' view discourse about semantical items (and intentional items generally) have, like the language of prescription and justification, a normative component which is irreducible to strictly descriptive language. While the force of such a claim seems to be obscure without some satisfactory account of what Sellars claims the nature of descriptive discourse is, it seems clear that he is appealing to the idea that discourse about the intentional properties of intentional items is irreducibly about a subject matter different from any other. [Back]

{8} This does not rule out, of course, that there are sentences which are about both sorts of items. What is important here, as in section I, is that there are many sentences which are about the one sort of phenomenon but not about the other. [Back]

{9} Compare the dispute between Chisholm and Sellars concerning whether sentences about psychological and semantic phenomena are, or are not, descriptive sentences [Chisholm and Sellars 524-525] Here Sellars claims that semantic and psychological sentences constitute a special and "unique form of discourse," [Chisholm and Sellars 527] while Chisholm argues that they may be taken, for the purposes of the issues at hand, as descriptive in just the way that standard empirical sentences are. As Chisholm later suggests [Chisholm and Sellars 529], such debate seems fruitless in the absence of any clear idea as to what will count as a descriptive sentence and what not. Throughout most of the earlier discussions, I have adopted Chisholm's attitude here in point of rejecting the fruitfulness of regarding such sentences as, in any interestingly specifiable way, other than descriptive. [Back]

{10} Assuming that it is this contrast which Nagel wishes to mark in using the terms 'particular identity' and 'general identity', I shall adopt these terms to mark the distinction as it is presented here. [Nagel 339] [Back]

{11} This convention, using 'the identity thesis' to refer to the claim that certain identifications are, roughly, logically possible, and using 'the identity theory' to refer by contrast to identifications based on certain scientific results, is not intended to reflect common usage. This application of 'the identity thesis' does, however, seem to reflect the usage adopted by Smart (Cf. [Smart (4)] ), while the application adopted here for 'the identity theory' seems to reflect that of Kim (cf. [Kim]), Brandt and Kim (cf. [Brandt and Kim]), and Lewis (cf. [Lewis]). [Back]

{12} Some defenders of the identity thesis might wish to hold out for no more than general identities of the sort expressed by our first conjunct, but since it is not a defense of the identity thesis which is here at issue, it seems reasonable to deal with a more thoroughgoing general identification The consequences of this in what follows are evident. [Back]

{13} If only the first conjunct of sentence (4) were present in the general identity under consideration, then we would be warranted in holding not that (5) and (6) were nomological equivalent but merely that (5) nomologically implies (6) and not vice versa, that is, that the move from (5) to (6) is warranted, preserving truth value, on nomological grounds, but not the reverse move. [Back]

{14} That is, given the way our language is currently used, and given the current state of scientific and general factual knowledge. [Back]

{15} Cf. [Sellars (3) passim], [Sellars (5) 190-196], [Sellars (9) passim] and [Sellars (18) passim] Sellars' position seems to be roughly that sensations are not intentional in the sense that they are not, in a way analogous to the way in which overt verbal episodes are, about anything. Sellars also maintains, however, that sensations exhibit properties similar to, and often confused with, the intentionality of thoughts and semantical episodes. Thus Sellars argues that to view sensations as intentional, in the sense of being about something, is to confuse the two different senses of 'of' in the expressions

sense impression of a red triangle

and

thought of a red triangle.

(Vide [Sellars (11) 196-197].) Sellars argues, in effect, that 'of a red triangle' in the former locution involves the idea that the episode in question is a response to red and triangular physical objects and to sensory stimulations of the sort produced by such objects (even when there are no such objects present); whereas only as used in expressions of the latter sort does 'of a red triangle' indicate that the episode in question is about a red triangle (or a red and triangular physical object). Thus Sellars speaks of the "object language character" of the former locution, which is not shared by the latter, construed as it is as involving notions analogous to semantical metalinguistic constructions. [Sellars (11) 196-197] A related source of the view that sensations exhibit intentionality is, Sellars seems to believe, a mistaken view concerning the relationship between sensations and perceptual propositional attitudes--propositional attitudes that we are perceiving in a certain way. Sellars' conviction on this matter seems to be that sensations are items which are used in explaining the occurrence of perceptual propositional attitudes, although they do not share the intentionality of the latter. [Sellars (14) 28]

Since it is not our concern here to discuss the question of how to construe discourse about sensations, we shall not be concerned to evaluate these features of Sellars' position. It seems worth noting, however, that since Chisholm seems to wish to rule out the possibility that there should be any psychological phenomena which are not intentional, it is not clear how he might wish to deal with such discourse. [Back]

{16} A recent discussion of the identity thesis makes explicit that it is intended to apply not only to mental items ordinarily called sensations by philosophers, but to thoughts as well. Thus in discussing the identity thesis, Brandt and Kim are concerned with what they call the class of "phenomenal occurrences," which, they write, include "events like itches, tickles daydreams, thoughts, and afterimages." [Brandt and Kim 516] In commenting on their wish to include thoughts in the class under consideration, Brandt and Kim write as follows "We assume that a thought can be characterized as a purely psychical event, as a kind of occurrence which is introspectible, and that such a characterization is distinct from a semantic description of it--an account of what it is about, what role the symbols that occur in the thought have in the speaker's language system, the connection of these with expectations, and so on." [Brandt and Kim 516, fn.] The willingness which Brandt and Kim express to have their discussion apply to thoughts seems, therefore, to rest on their assumption that thoughts are characterizable in terms other than semantic terms, that is, that they are characterizable independently of their intentionality. In particular, this willingness seems to rest on the assumption that thoughts can be characterized in terms of their being "introspectible," and that this characterization is independent of any characterization in terms of their intentionality. If we accept the idea that the fact that thoughts are "introspectible" is simply the fact that we can report our own thoughts non-inferentially, then it does not seem clear how Brandt and Kim would defend their crucial assumption. In any event, in what follows, we shall be concerned to defend the identity thesis as applied to thoughts without invoking any special assumptions of this sort. [Back]

{17} The standard account of strict identity, which we shall adopt in what follows, is that two items are said to be strictly identical if, and only if, they have all their properties in common. In giving an account of Leibniz's law, Nagel seems to wish to introduce the following qualification into this standard account of strict identity: two items are said to be strictly identical if, and only if, they have all their non- intensional and non-modal properties in common. [Nagel 341] Nagel does not specify what he has in mind as intensional or modal properties, but it seems reasonable to assume that if Cicero has the property of George thinking that he denounced Cataline, or if the Evening Star has the property of being necessarily identical to the Evening Star, then these are cases of such properties. Since on most accounts, such putative intensional or modal properties would not be counted as properties at all, Nagel's qualification does not seem to be required in giving an adequate account of strict identity. (Cf. [Quine (2) 139-144].) In what follows, therefore, we shall speak of identity and identification in the sense of strict identity and strict identification, indifferently presupposing either Nagel's qualification, or that putative intensional or modal properties simply are not properties at all.

It is worth mentioning that the properties which we have called intentional properties are not, on Sellars' view, intensional or modal properties as we have construed these latter terms above. For the properties we have called intentional are those which are attributed to items by the use of predicates such as 'is a thought episode that- p', and 'is a semantical item characterizable as logically expressing the proposition that-p'. But so to characterize an item is, on Sellars' view, simply to classify it as a member of the class of <>p<>s, or as a member of the class of .p.s, respectively. It seems reasonable, therefore, to regard such properties as just as non- modal and non-intensional as those attributed to items by the use of predicates such as 'is an elephant', and 'is a citizen of the United States', which clearly seem to be used to classify items as members of a certain class. (Cf. [Sellars (8) 658-659].) In any event, this particular feature of Sellars' views aside, we shall not, in what follows, exclude from consideration those properties which we have been calling intentional properties. Thus it will be required, if two items are to be truly said to be identical, that they have all their intentional properties, if any, in common. [Back]

{18} This notion of what it would be for a version of the identity thesis to be satisfactory seems to be invoked in Rorty's defense of a certain formulation of that thesis. Cf. [Rorty (2) 45]. [Back]

{19} It seems clear in a related discussion (Cf. [Chisholm (8) 244].) that Chisholm does wish to retain this principle. [Back]

{20} Such a move is also suggested by Chisholm's insistence, in his correspondence with Sellars, that he is not using the term 'describe' in any special or technical way. [Back]

{21} It is clear that we would not be warranted thereby in concluding also that all tokens satisfying these conditions would, in fact, play such a role. For there might be utterance tokens of some other language satisfying these conditions but playing an entirely different role, or tokens satisfying these conditions but playing no role whatever, for example, if uttered parrotingly. [Back]

{22} The other restriction made in the case of an overt verbal episode to the effect that there are no other sentences in English which play precisely the same role, seems superfluous in the case of thoughts, given that we accept the idea that if two thoughts play roles analogous to those played by two overt verbal episodes which, though differing in sign-design characteristics, nonetheless play the same intentional role, then the two thoughts are tokens of the same thought-type. Cf. [Sellars (8) 656-663]. Examples of two such verbal episodes might be utterings of 'It is raining.' by an English speaker and of 'Es regnet.' by a German. [Back]

{23} Charles Taylor has recently argued that defenders of the identity thesis commonly omit any argument that general identities could be established on the basis of any future scientific research, and that unless such identities could be established, the truth of the identity thesis would be deprived of an important source of interest. [Taylor (1)] [Back]

{24} The fact that thought episodes are causally related in a certain determinable way to certain meta-thoughts might be taken as providing one way in which thought episodes have (causal) empirical properties which are independent of the intentional roles they play, in the sense that although a part of the playing of such roles which thoughts come to play depends on there being such causal links, the latter are describable independently of the way in which one would specify the roles in question. Thus such empirical properties which, on Sellars' contention, are open to be discovered independently of the properties imputed to thoughts by Jones' theory, and which are neither assigned nor predicted by Jones' theory, might represent one sort of factual feature which thoughts playing similar intentional roles may be discovered to exhibit, independently of the imputation of properties made by Jones' theory. [Back]

{25} Sellars' contention that the identity thesis is not correct, and his belief that his view does not involve accepting the identity thesis, need not be taken as incompatible with this conclusion. For it does not seem to be the case that Sellars' rejection of what he calls the identity thesis (cf. p. 266 above) involves a rejection of what is being discussed here under that description. What Sellars does wish to reject will, we shall argue, be more adequately discussed in section VIII in connection with his rejection of the revised thesis of physicalism. [Back]


[Table of Contents][Continue to Rest of Chapter 5]