---

Philosophical Studies 3 (1952): 83-95. Author's note. This is a revised version of a paper read at the Ann Arbor meeting of the American Philosophical Association, May 1952. [ed. note: I have formatted the paper in a Wittgenstein quasi-Tractatus style as suggested by Sellars' numbering sequence (AC)]

---

Mind, Meaning, and Behavior

by WILFRID SELLARS
UNIVERSITY OF MINNESOTA

• 1. In one sense of "exist" it is beyond question that both minds and bodies exist; in another the question whether both minds and bodies or either or neither exist is the crux ofa legitimate and intricate philosophical puzzle -- the Mind-Body problem.
  • 1.1 Accordingly, the philosopher, after agreeing, with common sense that there are both minds and bodies, mental events and physical events, goes on to ask whether mental facts are "reducible" to physical facts, or vice versa, or whether both are "reducible" to facts which are neither.
    • 1.11 Where the reduction in question is taken to be explicit definition (in that broad sense which includes the Principia definition of the number Two in terms of logical primitives, not to mention the definition of Oxford University in terms of its colleges), we find philosophers exploring such alternatives as these:
      1. neither mentalistic nor physicalistic concepts are definable in terms of the other, nor both in terms of a third type of concept;
      2. mentalistic concepts can be defined in terms of physicalistic concepts, or vice versa;
      3. both mentalistic and physicalistic concepts are definable in terms of a third type of concept.
    • 1.12 Materialism and Neutral Monism have in common the claim that mentalistic concepts can be defined in terms of non-mentalistic concepts. Neutral Monism, however, claims that concepts relating to physical objects can be explicitly defined in terms of concepts relating to sense data. The materialist denies this.
      • 1.121 The most widespread form of Neutral Monism today is phenomenalism. Neutral Monisms differ according to their account of the logical form to be given definitions of mentalistic apd physicalistic concepts in terms of concepts relating to sense data.
        • 1.1211 Only if these concepts relating to sense data are non-mentalistic can Neutral Monism hope to realize its program of reducing mentalistic and physicalistic concepts to concepts which are neither.
  • 1.2 But before we can decide whether concepts relating to sense-data are or are not mentalistic, we must fist determine how the class of mentalistic concepts is to be delimited.
    • 1.21 I shall without further ado state my agreement with the classical thesis (represented, among others, by Descartes and Brentano) that the distinguishing feature of mental facts is intentionality or aboutness. Thus, some typical mentalistic concepts are believes, doubts, desires, fears, expects, etc.
    • 1.22 The question as to the status of concepts relating to sense-data thus turns out to be the question as to the role of intentionality, if any, in such concepts.
      • 1.221 It is immediately clear that the concepts of the sense qualities red, sweet, C#. etc. are not concepts of intentional acts.
        • l.2211 The phrase "concepts relating to sense data" is ambiguous. For while the concepts of the sense-qualities are not mentalistic concepts, it may well be the case that the concept of the givenness of sense-data, of their datumness, so to speak, is a mentalistic concept.
        • 1.2212 Thus, when Neutral Monism claims that mentalistic and physicalistic concepts are definable in terms of concepts relating to sense data, it must be using the phrase "concept relating to sense-data" in the sense of "concept definable in terms of sense qualities (and relations)" unless it is prepared to hold that the datumness of sense-data is a nonintentional fact.
        • 1.2213 Further-more, since it is not plausible to claim that sense qualities can be defined in physicalistic terms. Materialism must be interpreted as the claim mentalistic concepts can be defined in terms of physicalistic concepts together with concepts of sense qualities, neither of these latter two types being definable in terms of the other.
    • 1.23 Thus, the Mind-Body problem is, at bottom, the problem whether intentional concepts relating to minds can be reduced to nonintentional concepts, whether concepts of sense qualities, or physicalistic concepts or both, and if so, in exactly what sense of "reduced."
• 2. It is often wise to draw back pour mieux sauter. We shall be following this advice if we glance at the dialectics of a familiar problem in moral philosophy; the familiar one of the relation of Ought to Is.
  • 2.1 Let us notice two similarities between the Ought-Is problem and the Mind-Body problem.
    1. In both cases one asks about the reducibility of one type of concept to another.
    2. In both cases sentences employing a concept whose reducibility is in question characteristically have two verbs.
    • 2.11 Just as "Jones desires to go downtown" has the two verbs "desires" and "to go," so"Jones ought to pay his debt" has the two verbs "ought" and "to pay."
    • 2.12 Just as "Smith believes it is raining" is not a truth function of "It is raining," so "Jones ought to pay his debt" is not a truth function of "Jones pays his debt."
  • 2.2 If we suppose that to ask whether Ought is reducible to Is is to ask whether Ought is (contextually) definable in descriptive terms, we find a clash between Ethical Naturalism which claims that it is, and Ethical Non-naturalism ("Intuitionism") which claims that it is not.
    • 2.21 Let us put this by saying that for the former Ought is logically reducible to Is, while for the latter it is not.
    • 2.22 Consider now a position according to which, while Ought is not logically reducible to Is, nevertheless the only way in which moral obligation enters into the causal explanation of human history is via facts of the form Jones thinks that he ought to pay his debt.
      • 2.221 "Ought," in other words, enters into the antecedent or consequent of causal laws only as a subordinate element in a mentalistic context -- as, e.g., "entails" occurs in "Jones believes that responsibility entails indeterminism."
      • 2.222 In traditional terminology, obligation enters into the causal order only as an element in the intentional object of a mental act.
      • 2.223 If, as seems proper, we so use "ethical assertion" that while "Jones ought to pay his debt" is, of course, an ethical assertion, "Jones thinks he ought to pay his debt" is not, then the above claim can be rephrased as the claim that although the normative is not logically reducible to the descriptive, one can nevertheless explain the history of moral agents without making ethical assertions.
      • 2.224 Let us agree to put this by saving that although Ought is not logically reducible to Is, Ought is causally reducible to Is.
  • 2.3 Traditional moral philosophers, however, Naturalists and Non-naturalists alike, have tended to assume that Ought can be causally reducible to Is only if Ought is logically reducibleto Is.
    • 2.31 Thus, Ethical Naturalists have tended to assume that it can only be possible (which they think it to be) to explain the history of moral agents without making ethical assertions in characteristically ethical language, on condition that Ought is logically reducible to Is.
    • 2.32 While Ethical Non-naturalists have tended to assume that it is reasonable to deny (as they do) that Ought is logically reducible to Is, only if one is prepared to deny that Ought is causally reducible to Is. This latter, of course, they are prepared to do, since they characteristically insist that the existence of moral concepts and beliefs in the human mind cannot be accounted for in purely descriptive terms. Human thinking on ethical matters is, as they see it, ultimately grounded in and controlled by objective values and obligations.
    • 2.33 The moral philosophy we have been adumbrating combines a thesis characteristic of Ethical Naturalism with a thesis characteristic of Ethical Non-naturalism; the causal rcducibility of Ought with the logical irreducibilitv of Ought. Is it a form of Naturalism? of Non-naturalism?
      • 2.331 Would we not dodge these alternatives, and point out that the value of a system of classification is threatened when one of its presuppositions is abandoned?
• 3. Let us return to the Mind-Body problem. But first some general considerations.
  • 3.1 Let us now speak of terms rather than concepts. Thus, let us discuss the logical properties of ". . . believes . . . " rather than the status of the concept of Belief.
    • 3.11 Among the expressions of our language we find some which, in their primary use, appear in sentences which have sentences or quasi-sentences as component parts. By a quasi-sentence I mean an expression (containing a verb) which becomes a sentence when the mood of the verb is changed. Thus "It is necessary that all giants be tall" contains the quasi-sentence "all giants be tall.
      • 3.111 Let us call such expressions "connectives." Both "believe" and "and" are connectives. Thus, "It is raining" is a component part of both "It is raining and the temperature is low" and "Jones believes that it is raining."
    • 3.12 Some connectives are, in a familiar sense, truth functions. Others are not. The truth of "Jones believes that it is raining" is compatible with both the truth and the falsity of "It is raining." The falsity of "It is unusual that Jones has come to the meeting" is compatible with both the truth and the falsity of "Jones has come to the meeting."
  • 3.2 The connectives which belong to the primitive expressions of the language form developed in Principia Mathematica are all truth functions. Let us call a language built on the pattern laid down in Principia Mathematica with a certain set of primitive descriptive predicates, a PM language.
    • 3.21 In addition to its primitive expressions, a PM language will include expressions definable in terms of its primitive expressions.
    • 3.22 A recurring question in philosophy A. P. (after Principia) has been "Can the language we speak be 'reconstructed' as a PM language?"
      • 3.221 An obvious stumbling block in the way of such a "reconstruction" is the presence in ordinary discourse of connectives which are not truth functions. To this category belong, as we have noticed, certain expressions characteristic of normative, modal, and mentalistic discourse.
      • 3.222 That it is not absurd to hope that this goal may be achieved is suggested by the fact that "It is unusual that Jones has come to the meeting" can plausibly be regarded as identical in meaning (vagueness aside) with "Jones has come to the meeting and Jones was absent from 75% of the preceding meetings" a sentence of a kind which Principia Mathematica was designed to "reconstruct."
        • 3.2221 Might not the same be true of "It is necessary that Jones has come to the meeting," "It is fitting that Jones has come to the meeting" and "It is believed that Jones has come to the meeting"?
  • 3.3 Can our language insofar as it contains mentalistic expressions be "reconstructed" in PMese?
    • 3.31 Let us call "Philosophical Behaviorism" the thesis that the mentalistic expressions of our language can be defined by PM techniques in terms of the basic PM resources of our language.
      • 3.311 By "basic PM resources of our language" I mean those expressions which can reasonably be reconstructed as expressions belonging to the basic syntactical categories of PM.
      • 3.312 No connective which is not a truth function, and hence no mentalistic expression, belongs to the basic PM resources of our language.
      • 3.313 Caution would require that the following qualification be added to the definition of "Philosophical Behaviorism (3.31) after ". . . the basic PM resources of our language," namely: with the possible addition of modal expressions, should these be regarded as indispensable, and as incapable or definition in PMese.
        • 3.3131 Fortunately, to explore the question at hand it is not necessary to commit ourselves on the issue whether an adequate language of science must contain modal connectives (in particular, the causal modalities) so regarded. In the remainder of this paper, therefore, I shall assume that the requirements of scientific discourse (including the formulation of subjunctive conditionals and the definition of disposition terms) can be met by an extensional logic. In short, I shall assume that the maior part of the "extensionalist" program can be carried out, and concentrate attention on that part of the program which concerns the Mind-Body problem.
    • 3.32 Let us use the term "(Mind-Body) Dualist" to refer to philosophers who reject Philosophical Behaviorism and insist that at least one mentalistic expression in our language must be construed as a primitive connective.
      • 3.321 Is one committed to Dualism if one rejects Philosophical Behaviorism? (Assuming, of course, that Mentahsm is out of the question.)
      • 3.322 Before we attempt to answer this question, let us note that our rejection of Ethical Naturalism did not entail an acceptance of Ethical Non-naturalism -- for we saw that both are complex theses involving a logical claim and a causal claim. We rejected the logical claim of Ethical Naturalism, but accepted its causal claim; we rejected the causal claim of Non-naturalism, but accepted its logical claim.
        • 3.3221 The common presupposition of Naturalist and Non-naturalist is causal reducibility implies logical reducibility. We rejected this presupposition.
      • 3.323 Notice the parallel provided by the Mind-Body problem.
        • 3.3231 Dualists buttress the denial that mentalistic expressions are logically reducible to PMese with the claim that a causal account of the world must make use of mentalistic expressions.
          • 3.32311 This is, of course, true even of the Epiphenomenalistic variety of Dualism -- though it ceases to be so if "physical world" is substituted for "world."
        • 3.3232 On the other hand Philosophical Behaviorists buttress the assertion that the world is causally explainable in PMesc, with the claim that mentalistic expressions are logically reducible to PMese.
    • 3.33 May it not, however, be possible to hold that while mentalistic expressions are not logically reducible to PMese, nevertheless a causal account of the world (including psychological phenomena) can, in principle, be given in PMese.
      • 3.331 Let us use the function "m has A(O)" to say that in mind m there occurs a mental act of kind A of which the intentional object is O. Thus, A might (for the moment) be illustrated by believes and O by it is raining, so that "m has A(O) says of mind m that it believes that it is raining."
      • 3.332 Now Behaviorism as a substantive thesis is the claim that in an ideally complete psychology it would turn out that to each mentalistic function "m has AO)" there corresponded a PM function "φb" such that

        (A) m has A(O) ≡ φb

        where the values of "m" and "b" are pairs of minds and bodies which "belong" to each other. It follows that for every law involving mentalistic function there would be an equivalent law in PMese.
        • 3.3321 Let us refer to this claim as the thesis of Scientific Behaviorism. It must by no means be confused with the thesis of Philosophical Behaviorism. The latter differs in two respects:
          1. It is a stronger thesis. For the '≡' of schema A it substitutes the '=' appropriate to statements of analysis.
          2. It is not prospective. It claims that what we now say by using mentalistic expressions we can now say by using PM expressions. To be sure, it adds "in principle." The reference, however, is not to future developments or to an ideal knowledge, but to the disjunctive complexity of the appropriate PM expressions. Mention is also made of "open texture;"
        • 3.3322 We shall also refer to Scientific Behaviorism as the thesis of the causal reducibility of mental events to bodily events, where causal reducibility does not preclude logical [ir]reducibility (cf. Burks).
      • 3.333 In the argument to follow, I shall be assuming that the thesis of Scientific Behaviorism is true. I shall attempt neither to establish its truth, nor even to make it plausible. To those readers who are disinclined to accept it, I can only say that they may find the following pages interesting as a philosophical counterpart of "If Napoleon had won the battle of Waterloo . . ."
        • 3.3331 If Scientific Behaviorism were true, what would follow for the Mind-Body problem?
        • 3.3332 If one were to assert

          m has A(O) ≡ φb

          and yet deny the logical reducibility of "m has A(O)" to "φb" would one not be committed to Dualism in its epiphenomenalistic form?
          • 3.33321 Can the joint thesis of the causal reducibility but logical irreducibility of the mental to the bodily be held otherwise than as Epiphenomenalism?
  • 3.4 As our first step toward answering this question, let us examine that crude form of Scientific Behaviorism according to which the PM functions correlated with mentalistic functions concern the linguistic utterances of the body and their role in its economy.
    • 3.41 Consider, for example, the claim that the following equivalence obtains

      (B) m believes it is raining ≡ b tends to utter "es regnet"

      • 3.411 What is the import of such a statement as "b tends to utter 'es regnet'"? Clearly the utterance "es regnet" is not being considered here as a mere sequence of squeaks and whistles such as a parrot might emit. It is conceived to be a meaningful sequence of. sounds.
      • 3.412 The natural way of making this fact explicit is by reformulating (B) to read

        (B')

        m believes it is raining ≡

        b tends to utter "es regnet" and "es regnet" means it is raining

        • 3.4121 If we explore the right hand side of (B'), the first thing we note is that the second clause (" 'es regnet' means it is raining") contains the connective "means" which is clearly not a truth function.
        • 3.4122 Next we note that to say of an utterance that it "means it is raining" clearly conveys information about how the utterance is being used.
          • 3.41221 Thus,

            (B'-R) b tends to utter "es regnet" and "es regnet" means it is raining

            asserts that b tends to utter "es regnet" and conveys psychological information about b's use of "es regnet."
          • 3.41222 But granted that (B'-R) conveys psychological information about b's use of "es regnet," does it follow that (B'-R) makes a psychological assertion about b's use of "es regnct"?
        • 3.4123 Let us agree, for the moment, to make this inference. In other words, let us agree that "b's utterances of 'es regnet' mean it is raining" makes a psychological assertion about b's utterances of "es regnet."
          • 3.41231 If we continue along the lines of the analysis above, we note that Scientific Behaviorism is committed to an equivalence of the form

            "Es regnct" uttered by b means it is raining ≡ ψ("es regnet," b)

            where the right-hand side says of b that it has certain habits with respect to utterances of "es regnet."
          • 3.41232 Consider now the case of a Frenchman who utters "Il pleut." We are in any case authorized by our assumptions to write down the equivalence

            "Il pleut" uttered by b means it is raining ≡ θ("il pleut," b)

            But clearly the Scientific Behaviorist is committed to the thesis that if "es regnet" uttered by Germans has the same meaning as "il pleut" uttered by Frenchmen, then the habits of the latter with respect to "il pleut" share a common generic feature with the habits of Germans with respect to "es regnet." Let us represent this common feature by "K('. . .,' b)." Then we can write down the equivalences

            "Es regnet" uttered by b means it is raining ≡ K("es regnet," b)
            "Il pleut" uttered by b means it is raining ≡ K("Il pleut," b)

            or generally,

            ". . ." uttered by b means it is raining ≡ K(". . .,"b)

          • 3.41233 Notice, therefore, that

            "It is raining" uttered by b means it is raining ≡ K("It is raining," b)

            • 3.412331 Now, when I say

              Jones' utterances of "es regnet" mean it is raining

              I am mentioning "es regnet" and using "It is raining" to convey what is meant by "es regnet" as uttered by Jones. According to Scientific Behaviorism, if what I say of Jones' utterance is true, then the utterance "It is raining" which I use is the manifestation of habits generically identical with Jones' habits with respect to "es regnet."
              • 3.4123311 Notice that since utterances convey information about language habits by virtue of being manifestations of these habits, an utterance may (potentially) convey more information than is appreciated (actually conveyed) at a given stage of human knowledge. Thus (to exaggerate), " 'Es regnet' uttered by Jones means it is raining" might today convey only that Jones has the 'same' habits with respect to "es regnet" as the speaker has with respect to "it is raining," but might convey in the future that Jones has habits of (specific) kind and with respect to "es regnet."
            • 3.412332 And when the Scientific Behaviorist is in a position to propose specific equivalences of the form

              ". . ." uttered by b means *** ≡ F(". . .," b)

              they will be subject to the condition that they can only be true if they are "pragmatically consistent," that is, if the "***" used on the left-hand side is a manifestation of the kind of habit mentioned on the right-hand side. By virtue of this fact they will be more than "mere" material equivalences.
            • 3.412333 Yet they are neither laws of nature nor, in any usual sense of the term, logical equivalences. They are validated not by showing that the left-hand side can be constructed out of the same (PM) primitives as the right-hand side, but rather by knowing the circumstances in which it is correct to use the left-hand side.
              • 3.4123331 As an illuminating parallel it can be pointed out that although "x is here" said by Smith who is at s is, in a strong sense, equivalent to "x is at s," nevertheless it is not, in any ordinary sense, logically equivalent to it.
        • 3.4124 Now the truth of the matter, of course, is that while

          (C) b's utterances of "es regnet" mean it is raining

          conveys psychological information about b's utterances, it does not make a psychological assertion about b's utterances. We must abandon the inference momentarily sanctioned in 3.4123.
          • 3.41241 Semantical assertions, such as (C), convey psychological information about language users, but they are not psychological assertions.¹ They do this by virtue of the feature of their use pointed out above (3.412331).
          • 3.41242 When Jones says to me

            Smith's utterances of "es regnet" mean it is raining

            I can infer that Smith uses "es regnet" as I use "It is raining," even though Jones is not making an assertion about the way in which Smith uses "es regnet."
          • 3.41243 It should now be pointed out that it is not only linguistic events in the narrow sense of the use of conventional languages that are correctly said to "mean such and such." If we use the term "symbol" for items which are correctly said to mean such and such (whereas "sign" means symptom of such and such, and is not a semantical expression), then the class of symbol events is radically more inclusive than that of linguistic events in the narrower sense. It is only if "language" is taken in the broader sense of the use of symbols, that it is plausible to identify thought with the use of language.²
  • 3.5 Before we can put the results given above to good use, we must take another look at mentalistic discourse.
    • 3.51 It is a familiar fact that many mentalistic expressions are definable in terms of other mentalistic expressions. Indeed, a scrutiny of the psychological (and philosophical) literature devoted to the descriptive phenomenology of the mental suggests that an adequate basis for the definition of all mentalistic terms can be found in "act of thought" (which we shall abbreviate as "thought") and "about," together with non-mentalistic expressions.
      • 3.511 Thus, "x is a thought about O" would be the form of a basic sentence of mentalistic discourse. A thought in this sense is a mental episode. Furthermore, to say of x that it is a thought is not to ascribe any dispositional features to x.
        • 3.5111 Yet it may be the case that in order to be correctly characterized as a thought, x must be a complex state of affairs with dispositional as well as purely episodic components. Compare our remarks in 3.4124 above on semantical statements. To say of certain grunts and groans that they mean it is raining is not to ascribe any dispositional feature to these grunts and groans, although it is only if they are the manifestation of certain habits that it is correct to say of them that they mean it is raining.
      • 3.512 Mentalistic verbs relating to motivation ("desires," "chooses," "hates," etc.) would be defined in terms of the tendency of thoughts about conduct to bring about conduct
    • 3.52 As a matter of fact, further reflection suggests that our list of two mentalistic primitives is redundant, and that the single term "about" would suffice. It would be absurd to speak of a thought which was not about something. Can we not therefore define an act of thought as an event which is correctly said to be about something?
      • 3.521 And a mind as a continuant which has thoughts?
      • 3.522 But what is aboutness but meaning? Thus (3.4124) to say of an evenf e that it is about something (and hence that it is a mental event) is not to make a psychological assertion about e, even though it is to convey psychological information about e.
        • 3.5221 To say of an event e that it is about something is not to describe e. In general, to make semantical assertions about psychological events is not to describe these events, though it is to convey information the assertion of which would describe them.³
        • 3.5222 But if in saying of an event that it is about something we are not describing the event, it follows that to say of an event that it is mental is not to describe it in a way which precludes a correct description of the event in bodily terms. In other words, "x is mental" does not stand to "x is φ" (where "φ" is definable in terms of bodily states) as "x is green" to "x is red."
        • 3.5223 Indeed, it follows that every mental event must (in principle) be describable in non-mentalistic terms.
        • 3.5224 And while, of course, it docs not follow from the above alone that mental events must (in principle) be describable in terms of bodily states (for they might be describable in terms which were neither mentalistic nor definable in terms of bodily states), this does follow from the above together with the thesis of Scientific Behaviorism.
        • 3.5225 Notice, of course, that if Scientific Behaviorism is to be plausible, we must include in the class of bodily states such activities as seeing colors, hearing sounds, tasting tastes, and having images.
          • 3.52251 Prichard has correctly pointed out that seeing a color is not cognizing a color. Ducasse's insight that tasting a taste is like waltzing a waltz was vitiated only by his failure to appreciate that tasting a taste is not cognizing a taste.
          • 3.52252 To see colors, hear sounds, etc., is, in one sense of this everyday term, to be conscious, but not cognitively conscious. The latter involves aboutness, the former does not. Seeing a color is not a mental activity.
            • 3.522521 The epistemological notion of the givenness of colors, sounds, etc., must not be confused with the notions of seeing colors, hearing sounds, etc. Givenness is a form of cognitive consciousness and requires mention of aboutness in its analysis.
          • 3.52253 Seeing a color cannot be defined in physico-chemical terms, or, for that matter, in terms of overt behavior. But it is a mistake to suppose that "bodily state" (in ordinary usage) means state definable in either of these ways.
          • 3.52254 Since sensory states do not occur apart from what would readily be called bodily states, and since they are not mental states, the decision to use "bodily state" to include them would not be absurd.
          • 3.52255 If, on the other hand, one decided to exclude them from bodily states, then Scientific Behaviorism would have to be reformulated in terms of "states of the body" and "states of the sensorium," as something like the latter would have to be the continuant language for sensory events.⁴
  • 3.6 To sum up our results: If Scientific Behaviorism is correct, and if our account of sentences of the form ". . . means____" or ". . . is |about____" is correct, then every mental event can (in principle) be described in terms of expressions which are definable in terms of bodily states.
    • 3.61 We have thus shown how it is possible to accept Scientific Behaviorism, that is (3.3322) the thesis of the causal reducibility of the mental, yet deny the logical reducibility of the mental, without being committed to Epiphenomenalism (cf. 3.33321). And we have shown how it is possible to reject Philosophical Behaviorism without being committed to Dualism (cf. 3.121).
• 4. The logical irreducibility of mentalistic discourse to Behaviorese, insisted on by traditional dualisms, turns out, if the argument above is sound, to be exactly the logical irreducibility of semantical metalanguages to PMese.
  • 4.1 Now, while we often use semantical statements to convey information which could (in principle) be formulated in PMese, and while this use constitutes the application of the semantical language form, it must not be inferred that what is said by semantical discourse could, in principle, be said in PMese.
    • 4.11 Just as from the fact that the use of normative discourse conveys a great deal of information about the speaker and his community, and from the fact that the normative form of discourse gains application through functioning in motivation,⁵ it must not be inferred that what is said by normative discourse can be said in psychological and socio-psychological discourse about motivation.
      • 4.111 That which is said by "Jones ought to pay his debt" could not be said in even an ideal PMese.
    • 4.12 That which is said by " 'It is raining' is true if and only if it is raining" or even by " 'It is raining' means it is raining" could not be said in even an ideal PMese.
  • 4.2 It is indeed important to see that (in principle) the world, including human behavior, could be described and predicted without using semantical discourse. But the proper way to interpret this fact is not by propounding an "extensionalist thesis" according to which everything can be said without using semantical discourse, but rather offering a careful account of the interrelationships which would obtain between semantical discourse and an ideal behaviorese.
    • 4.21 In general, the task of the philosopher is to explore without prejudice the syntactical and pragmatical relationships which obtain between the various forms of discourse, descriptive, semantical, normative, modal, etc.
      • 4.211 Surely the hankering to give bad marks and a pseudo-conceptual status to other forms of discourse merely because they are discerned not to be descriptive discourse belongs with other left-wing deviations in the Madame Tussaud's Wax Work Museum of the analytic movement

  ---

  NOTES

  1 It is my hope that this distinction between what is asserted, and what is conveyed but not asserted, by semantical statements in ordinary usage throws some light on what I was trying to say in Section IV of my "Realism and the New Way of Words," Philosophy and Phenomenological Research, vol. 8 (1948), (reprinted in Readings in Philosophical Analysis, edited by Herbert Feigl and Wilfrid Sellars, and published by Appleton-Century-Crofts, New York, 1949).

  2 Implicit in this paragraph together with 3.41231-3.41232 and 3.41241 is the semantical (not psychological) distinction between linguistic types: (linguistic functions) and token-classes which I have developed in several papers, most recently in "Quotation Marks, Sentences and Propositions," Philosophy and Phcnomcnological Research, vol. 10 (1950); and "The Identity of Linguistic Expressions and the Paradox of Analysis," Philosophical Studies, vol. 1 (1950).

  3 This paragraph, together with 3.4243 and 3.522521 is a restatement of the thesis, argued in "Realism and the New Way of Words," that the semantical (as opposed to psychological) concept of a token is the central concept of an epistemology which is to avoid both forms of psychologism distinguished in that paper (see Readings, p. 430, notes 2 and 3).

  4 It may not be out of place to point out that the account of the Mind Body problem given in Section IX of "Realism and the New Way of Words" (see also the second paragraph of note 22 to page 455 of Readings) differs from that of the present paper only in its greater obscurity. Its primary flaw was to suppose (p. 453 of Readings) that a dualism of sense qualities and brain events qua describable in physicalist terms, would be a mind-body dualism.

  5 See my "Obligation and Motivation" in Readings in Ethical Theory, edited by Wilfrid Sellars and John Hospers (New York: Appleton-Century-Crofts, 1952), p. 516.