Self-Consciousness

Philosophical Review

Volumn 106, Issue 1, 1997, Pages 69-117

  Bealer, George ^a  

^a UNIVERSITY OF COLORADO   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0003040715     PISSN: 00318108     EISSN: 15581470     Source Type: Journal    
DOI: 10.2307/2998342     Document Type: Article

References (66)

1. 'See, for example, Hilary Putnam, "On Properties," in Mathematics, Matter and Method: Philosophical Papers, vol.1 (New York: Cambridge University Press, 1975), 305-22.
2. Sydney Shoemaker, "Some Varieties of Functionalism," Philosophical Topics 12 (1981): 93-119.
3. Brian Loar, Mind and Meaning (Cambridge: Cambridge University Press, 1981).
4. Ned Block, "Can the Mind Change the World," in Meaning and Method: Essays in Honor of Hilary Putnam, ed. G. Boolos (Cambridge: Cambridge University Press, 1990). In a moment I will say more about the notions of second-order properties and ontological functionalism. Other versions of functionalism will be dealt with in due course; see, for example, note 2.
5. David Armstrong and David Lewis hold a "functionalist identity theory" according to which mental properties are contingently identical to the first-order "realizations." As we shall see, the argument from self-consciousness can be extended to show that this cannot be right.
6. In "Method in Philosophical Psychology" (Proceedings and Addresses of the American Philosophical Association 50 (1975): 23-53).
7. Paul Grice entertains this position (as well as the more familiar ontological functionalism).
8. I give my reasons in "Mind and Anti-Mind," Midwest Studies in Philosophy 9 (1984): 283-328.
9. "Mental Properties," Journal of Philosophy 91 (1994): 185-208; and section 1 below.
10. Analogous distinctions hold for relations. See also note 24. Note that in this characterization I used the ontological notion of a primitive property rather than the notion of an undefinable property. The reason is that there is a well-established sense (associated with the project of conceptual clarification) in which some ontological primitives may be said to be definable. For example, someone could say that even though phenomenal colors are ontological primitives, any one of them (for example, Hume's missing shade of blue) is definable in terms of the others (see the close of section 1). Because this is a well-established usage, the term 'ontological functionalism' is preferable to 'definitional functionalism'. Incidentally, in contemporary second-order logic (that part of the simple theory of types known as "the functional calculus of second order"), a property expressed by a formula A(f) is often called a second-order property even if the formula A involves no quantification over properties. This usage departs from the original Russellian usage and is not relevant to functionalism.
11. In "Mental Properties" I give my reasons for rejecting the identity thesis.
12. In "Theories" (in The Foundations of Mathematics, ed. R. B. Braithwaite (London: Routledge and Kegan Paul, 1931), 212-36).
13. F. P. Ramsey gave a technique, not for denning "theoretical" terms, but rather for eliminating them by means of existentially quantified predicate variables. To my knowledge, the idea of using existentially quantified predicate variables to construct the kind of definition described in the text is first found in R. M. Martin's "On Theoretical Constants and Ramsey Constants," Philosophy of Science 31 (1966): 1-13. The idea was subsequently taken up by Lewis, Grice, Harman, Loar, Shoemaker, Block, Cummins, Jackson, and many others.
14. But there are limitations on how comprehensive A may be. If A includes too much information about the standard mental properties, functionalism is subject to immediate refutation. For example, suppose A includes a clause stating that, say, the property of being in pain is not a first-order physical property (having firing C-fibers, etc.), as is required by the denial of the identity thesis. And so on for the other standard mental properties dealt with by A. In that case, there could be no first-order physical properties R that satisfy A, contradicting the materialistic version of functionalism. And more generally, suppose A includes a clause stating the functionalist thesis that the properties of being in pain, thinking, etc. are not first-order properties. In this case, there could be no first-order properties R satisfying A, and so the associated Ramsified definitions would be mistaken. The way around these problems is, presumably, to confine A to a description of the characteristic interaction of the standard mental properties with one another and macroscopic physical properties.
15. In "Some Varieties of Functionalism" Shoemaker gives this view a precise formulation and explicitly advocates it. In their remarks about "physical realizations" of mental properties a great many functionalists seem to be endorsing this materialistic version of functionalism.
16. This principle is formally consistent with the identity thesis (for example, that being in pain is identical to having firing C-fibers). For example, principle I does not formally imply that a person cannot be self-consciously aware that he has firing C-fibers. It only requires that if he is, then having firing C-fibers = being in pain (or thinking or ⋯). Of course, principle I together with the negation of the identity thesis (for example, being in pain ≠ having firing C-fibers, etc.) implies that a person cannot be self-consciously aware that he has firing C-fibers. Paul Churchland (in"Reduction, Qualia, and the Direct Introspection of Brain States," Journal of Philosophy 82 (1985): 8-28) holds that people with prior exposure to physiological theory and its terminology could be self-consciously aware of their brain states. I find mis claim incredible. But for the purpose of this paper we may simply sidestep the issue by restricting principle I and our applications of it to people who have had no prior exposure to physiological theory and its terminology.
17. Incidentally, someone might claim that the contents of a being's self-conscious awareness are propositions that involve only species-specific mental properties. But this is wholly implausible. If a being in another species is self-consciously aware that it is in pain, this proposition (that it is in pain) is a predicative proposition formed from the property of being in pain. Likewise, when I am self-consciously aware that I am in pain, this proposition (that I am in pain) is a predicative proposition formed from the property of being in pain. To see that they are one and the same property, note that both propositions have as a logical consequence the existential proposition that something is in pain. This proposition-that something is in pain-is something the being and I would agree on. This and many other crucial logical relations would be lost if the analysis were based on species-specific properties. (For more on this, see responses 1 and 2 below.) In any case, the appeal to species-specific properties here would violate the spirit of ontological functionalism: if the mental lives of crea-tures in diverse species are functionally identical, ontological functionalists are committed to holding that they are identical simpliciter. Moreover, much the same argument as I am about to give could be mounted against ontological functionalism even if one supposed that the contents of self-conscious awareness were always propositions involving species-specific mental properties.
18. I am making the following standard supposition, which holds in all classical intensional logics (Frege's, Russell's, Carnap's, Church's, Kaplan's, Montague's, etc.): in principle P the proposition denoted by the 'that'-clause 'that he is in pain' is the same as the proposition expressed by the unem-bedded antecedent 'he is in pain'. In each case, the proposition is formed from that which is expressed by the predicate (verb phrase) 'is in pain'. In their Ramsification of P our functionalists should therefore use one and the same predicate variable 'R1' to replace both the embedded and the unembedded occurrence of 'is in pain'. (For further discussion of this point, see response 2 below.) If, instead, embedded occurrences of psychological predicates were not replaced by predicate variables, circularity would result. Alternatively, if principles like P were just deleted from A, the resulting Ramsified definitions would be too weak. (See response 4 below.)
19. For example, Sydney Shoemaker's "Some Varieties of Functionalism." See also my "Mental Properties."
20. Here is a proof of this principle: Predicates (that is, verbs) in a given language do not shift their meaning (versus extension) from world to world-relative to any possible world, F means what it means in the actual world. At the same time, relative to any possible world, the gerund being F, which is the nominalization of the predicate F, denotes the property expressed by the predicate F. This is why x is F iff being F is a property of x is necessarily true for ordinary F It follows that, relative to any possible world, being F denotes what it denotes in the actual world. Hence, being F is rigid.
21. On the Armstrong-Lewis theory, the following relativized statements are supposed to hold: in human beings, pain = firing C-fibers; in Martians, pain = firing D-fibers; etc. But the analogous thing certainly does not hold for our canonical gerundive idiom for referring to properties: in human beings, the property of being in pain = the property of having firing C-fibers; in Martians, the property of being in pain = the property of having firing D-fibers; etc. This is wholly counterintuitive, and the above argument shows why. (As shown in note 10, when someone is self-consciously aware that he is in pain, the content-that he is in pain-is a predicative proposition formed from the property of being in pain. And this is so regardless of species.)
22. This is exactly what Russellians would do. Fregeans would do some-thing logically equivalent. They would, for example, rewrite P with a 'that'- clause in the antecedent: if it is true that a person is in pain and engaging in introspection, the person will be self-consciously aware that he is in pain. Then they would replace both occurrences of 'is in pain' with one and the same predicate variable 'R1'. In fact, functionalists who are Fregeans would want to deal with all unembedded occurrences of predicates in A in some such manner. The reason is that they want their predicate variables R1 to range over intensions (that is, the sort of entities expressed by predicates F), not over extensions (that is, the sort of entities predicates F refer to, namely, the set of things F is true of).
23. See, for example, Jon Barwise and John Perry, "Semantic Innocence and Uncompromising Situations," Midwest Studies in Philosophy 6 (1981): 387-404;
24. Bealer, Quality and Concept (Oxford: Oxford University Press, Clarendon Press, 1982).
25. Bealer and Uwe Monnich, "Property Theories," Handbook of Philosophical Logic 4 (1989): 133-257. See also many of the authors mentioned in note 30.
26. Incidentally, someone might try to hold the following: When F is a nonpsychological predicate, the proposition denoted by the 'that'-clause that x is F is formed from a concept of the property of being F; but when 'F' is a psychological predicate, the proposition denoted by the 'that'-clause that x is F is formed, not from a concept of the property of being F, but rather from a concept of a "realization" of the property of being F. This logical theory, however, is so disunified and unmotivated that it may not be taken seriously.
27. In Inquiry (Cambridge: MIT Press, 1984), Robert Stalnaker seems to advocate a two-step picture, but one that is designed to do without mental representations. Our main problem threatens both steps. Concerning step one, although Stalnaker does not tell us how he would actually define mental state-types (belief, desire, self-consciousness, etc.), it seems inevitable (given that he does not avail himself of mental representations) that he would need to resort to some sort of Ramsification to avoid standard circularity problems (for example, defining belief in terms of desire and desire in terms of belief). But if he does, it seems that the arguments in the text would carry over mutatis mutandis. Concerning step two, Stalnaker advocates an ideal-rational-agent version of the reliable-indicator analysis. This analysis has two unacceptable consequences: (1) Believing a proposition entails believing every necessarily equivalent proposition. (2) Believing a proposition entails that one believes that one believes the proposition (and that one believes that one believes that one believes the proposition, ad infinitum).
28. Much the same problem would arise if one proceeded in stages: first, giving functional definidons of all mental properties besides self-consciousness and, second, using these defined notions in a functional definition of self-consciousness. The problem is that there are principles akin to T in which 'is self-consciously aware' is embedded within itself, for example, a principle whose consequent is this: x is self-consciously aware that he is self-consciously aware of something. How would one Ramsify this clause? If one wrote 'x is related by R4 to the proposition that he is related by R4 to something', our main argument could be repeated to show that the relation of self-conscious awareness is a first-order relation, thereby contradicting ontological functionalism. If instead one left the consequent of the relevant clause untouched (that is, 'x is self-consciously aware that he is self-consciously aware of something') or if one wrote 'x is related by R, to the proposition that he is self-consciously aware of something', the definition would be viciously circular. Of course, one might try to avoid the problem by just deleting the relevant principle from the theory A, but this proposal would be subject to the problem I am about to discuss in the text.
29. Some of these principles are like P in that they specify "dynamic" characteristics of standard mental properties. Others are "static," simply restricting the range of the relation of self-consciousness awareness. I itself is one such principle: for all F, if a person is self-consciously aware that he is For that he Fs something, then either F = being in pain or F = thinking or ⋯. Unlike P, I gives a necessary not sufficient condition for self-consciousness.
30. Personally, I find this idea wholly implausible: scarcely anything could ever be better justified to someone like you or me than the fact that he is self-consciously aware of various things; it would be altogether unscientific of him to omit principles concerning this undeniable phenomenon. Moreover, in normal cases, when a person knows that he is self-consciously aware, say, that he is thinking, this is neither a theory in any standard sense of the term, nor "theory-laden."
31. There is an explicitly two-tier proposal that also omits order restrictions: for example, x is self-consciously aware that P iffdef there exist properties r and properties R both satisfying A such that r are "realizations" of R and x R4 P. This proposal is defeated by problems analogous to those cited at the end of section 2 and in note 45. There is also a more complicated two-tier proposal that weaves R into the propositions comprising the ranges of r, but it too is defeated by such problems.
32. See, for example, Jerry A. Fodor, Psychosemantics (Cambridge: MIT Press, 1987). I believe that token-in-a-box functionalism cannot literally be right, for it is inconsistent with the possibility of disembodied subjects-a possibility that many functionalists accept and for which supporting arguments can be given (see my "Mental Properties"). If the token-in-a-box theory were taken literally, the "boxes" and the "tokens" would have to be functioning bodily parts. Given that a disembodied subject would have no functioning bodily parts, a literal formulation of language-of-thought functionalism must therefore be stated, not in terms of "tokens" and "boxes," but rather in terms of abstract relations to language-of-thought sentences (versus sentence tokens). An alternative response to the possibility of disembodiment would be to offer token-in-a-box functionalism, not as a general theory of mind, but rather as a theory of the human mind, assuming that human beings are essentially embodied. But this restricted token-in-a-box theory would falter over the problem I am about to discuss in the text, namely, the problem of specifying the contents of (human) Mentalese psychological predicates. The reason is that human thoughts about mental properties are not typically about human-thinking, human-desiring, human-self-consciousness, etc.; rather they are about thinking, desiring, self-consciousness, etc. simpliciter. (See note 10.)
33. For an example of a conceptual role approach, see Gilbert Harman, "Conceptual Role Semantics," Notre Dame Journal ofFormal Logic 23 (1982): 242-56.
34. For a world-word causal approach, see Dennis Stampe, "Toward a Causal Theory of Linguistic Representation," Midwest Sludies in Philosophy 2 (1977): 42-63;
35. Fred Dretske, Knowledge and the Flow of Information (Cambridge: MIT Press, 1981).
36. For a sophisticated descendant of the world-word causal analysis, see Jerry Fodor, Psychosemantics; my discussion of the causal theory will apply mutatis mutandis to this view.
37. For an example of a "two- factor" approach, see Ned Block, "Advertisement for a Semantics for Psychology," Midwest Studies in Philosophy 10 (1986): 615-78.
38. Strictly speaking, the definition of the content-of relation for Mentalese primitives would be a general "definition-by-cases" in which the definition just given in the text is associated in the obvious ways with one of the cases: x is the content of y iffdef (1) if y is a logical constant such that it is nomologically necessary that y behaves in Mentalese sentences in the Belief and Desire Boxes in the way 'there exists' behaves in good arguments in rational-decision theory, then x = the operation of existential generalization and (2) ....
39. Incidentally, conceptual role functionalists need to be careful in how they characterize the notion of a second-order property. Their notion should be relative, not absolute. Specifically, they should relativize (or parameterize) their notion to an antecedently given class of nonpsychological entities, among which may be nonpsychological entities of arbitrary order. For example, suppose existential generalization is a second-order intension. Would this force conceptual role functionalists to hold that mental properties are of some order higher than two? No. For, on the charitable reading, they mean that mental properties are of order two, relative to an antecedently given class of nonpsychological entities. This brings up a related point. In Russell's own ramified theory, propositions themselves have orders, and this affects how he would classify relations holding between individuals and such propositions. Functionalists have suppressed this (unnecessary) complication and, at least in this regard, they treat propositions on a par with individuals. This is certainly coherent, and in this paper I am following them on this point.
40. The following alternate analysis avoids the circularity but at the expense of getting the wrong relation: the content of 'B' =def the relation R such that, for each Mentalese sentence s, the subject's bearing R to s would in normal conditions cause i B s to be tokened in the subject's Belief Box. After all, the belief relation is not a relation between subjects and Mentalese sentences; it is a relation between subjects and the propositions that are supposed to be the contents of those Mentalese sentences.
41. Incidentally, the following analysis also fails: the content of 'B' =def the relation R such that, for some Mentalese sentence s and proposition p that is the content of s, the subject's bearing R to p would in normal conditions cause i B s to be tokened in the subject's Belief Box. The analysis fails because relations such as considering and entertaining would also satisfy the definition (see note 27 for further explanation).
42. def there is a unique relation R such that, for each Mentalese sentence s and proposition q drat is the content of s, the subject's bearing R to q would in normal conditions cause i B s to be tokened in the subject's Belief Box, and x is related by R to p. There would be no objection to this definition (or that in the text) if it could be rewritten as an inductive definition in which the offending occurrences of 'R' could be eliminated and in which the only surviving occurrences of predicate variables had as values first-order "realizations." But the causal theory is incompatible with this idea: the belief relation is not being built up inductively; rather it is picked out all at once as an independently existing, causally efficacious relation.
43. Note in this connection that causal theorists would have to deem both the thinking and the considering to be causes of the believing. This is relevant to the point made at the close of note 25.
44. Suppose that BT were modified with every relevant qualifier of the sort just indicated. This new BT principle would not be a mere causal or nomological necessity. It would be necessary tout court. (Indeed, if this new BT principle were merely a causal necessity, we would not be able to know it in the way we evidently do, that is, as something which is intuitively obvious in the way many other necessary truths are.) But if the relation between the antecedent and consequent of this BT principle is necessary tout court and if the relations between the "realizations" of mental properties really mimic the relations between the corresponding mental properties, the relation between the "realization" of the antecedent of BT and the "realization" of the consequent of BT would also have to be necessary tout court. On the usual causal picture, however, this relation ought to be only causally or nomologically necessary. (Call this the modal problem.) Since the stronger modality meshes readily with a conceptual role picture, here is another reason why language-of-thought functionalists might, in the case of Mentalese psychological predicates, be led to abandon causal models and turn to a conceptual role account.
45. In turn, this approach avoids the modal problem (note 28); for, in connection with the indicated qualified principle, A does not wrongly attribute a mere causal connection but rather the relevant stronger modality.
46. The list is long, and includes Aczel, Bealer, Chierchia, Davidson, Dunn, Feferman, Fine, Fitch, Gaifman, Gilmore, Gupta, Jubien, Kripke, McGee, Menzel, Mönnich, Martin and Woodruff, T. Parsons, Perry and Barwise, Reinhardt, Salmon, Soames, and Turner.
47. "Outline of a Theory of Truth," Journal of Philosophy 72 (1975): 690-716.
48. If you doubt this, shift to the following variant of the cogito: I am contemplating more intently than anything else the proposition that there is something I am contemplating more intently than anything else. In symbols: 'i C [(∃p) i C p]' . I can be contemplating this proposition more intently than anything else and yet, for each n ≥0, I might be contemplatingn nothing whatsoever (that is, contemplating0 no nonpsychological proposition, contemplating1 no proposition about contemplating0, and so forth).
49. The points in the text show that the following fails: x thinks p iff for some n, x thinksn, p. Of course, within standard type theories one cannot even quantify over levels n.
50. The propositional attitudes thus differ from truth (as Kripke explains the concept in "Outline"): once the extension of 'true' over the nonsemantical is fixed, its extension over the semantical is fixed as well. The extension of 'thinks' is formally different: even once the extension of 'thinks' over the nonpsychological is fixed, its extension over the psychological remains open.
51. In this connection we can see that infinitary disjunctive definitions of mental relations fail because they are circular: for example, x thinks p iffdef (x is in physical state S and p = the proposition that something is a cow) or ⋯ or (x is in physical state S' and p = the proposition that thinking is a 2-place relation that holds between people and propositions). If, to avoid this circle, one omits the offending clause, the resulting definition would then fail to provide a necessary condition. (Note that, for an analogous reason, valuative properties cannot have infinitary disjunctive definitions in terms of their "naturalistic realizations." See also note 46.)
52. This is not to say that the definitions are correct: nothing I have said rules out the possibility that they are both subject to some further kind of problem, such as inverted spectrum. That is an independent question.
53. This conclusion could be reached quite directly if one assumed the strong auxiliary view that every property is either a simple or else a complex formed in a unique way from simples and that a property is definable only if it is a complex. On this view, at most one Ramsified definition of a given property could be correct. The problem is that there are always a number of logically equivalent formulations of any given psychological theory A. Given that thinking is self-embeddable and ungrounded, we can show that for any two such formulations of SI, the Ramsified definitions of thinking based on them would not be even materially equivalent. (To see why, see note 43.) Because there can be nothing in principle that would make one of these Ramsified definitions stand out as the "right one," the only reasonable conclusion to draw is that none of them would be correct. In the text I do not use this argument, for I want an argument that is consistent with the (plausible) thesis that definable properties need not be complex.
54. See, for example, Jon Barwise, "Three Views of Common Knowledge," in Theoretical Aspects of Reasoning about Knowledge, vol.2, ed. M. Vardi (Los Altos, Calif.: Morgan Kaufmann, 1988), 365-380
55. Kripke, "Outline." I will consider an artificial one-person example; obviously there are analogous real-life examples involving chains of people.
56. In symbols: the relation mathematical equation;
57. Most functionalists should be happy with this logical framework: its quantified variables may range freely over properties, relations, and propositions, as well as particulars, and it is fully equipped to give standard propositional-attitude reports.
58. As indicated earlier, language-of-thought functionalism would be undermined for a similar reason: there would be no way to identify thinking (as opposed to thunking) as the content of the Mentalese psychological predicate 'T'.
59. In this argument, we would have a sequence of (simultaneously introduced) nonstandard relations thonking, dosiring,.... The nonstandard relation thonking = the relation mathematical equation. Likewise for the nonstandard relation dosiring. And so on. Grant that thinking ≠ thonking; desiring ≠ dosiring;.... Then, if A is otherwise as above, we can show: the sequence of standard mental relations would satisfy A iff this sequence of nonstandard relations also satisfies A. So the Ramsified definitions based on A would be mistaken, and the associated version of functionalism would fail. Note that if A deals with the relation of referring, the latter should be deemed a psychological relation and a nonstandard relation of roferring would also be included in the list of nonstandard relations.
60. For example, in a similar but more complex fashion, one could conversationally introduce the name of an operation * with the following features. * (thinking) ≠ thinking; * (desiring) ≠ desiring; etc. For an arbitrary complex intension w (either a proposition, a complex property, or a complex relation), *(w) is the complex intension that arises from w by replacing each simple constituent u of w with *(u). And for simple properties u, *(u) = the property of being a y such that, for some z, z is an instance of u and y = *(z). In symbols: *(u) = [y: (∃z) (z is an instance of u & y = * (z) ]. Likewise for simple relations. Then, on the assumption that thinking, desiring, etc. and their *-counterparts are simple relations, we can show that p and *(p) are necessarily equivalent for arbitrary propositions p. (Alternatively, if they are complex, we can reason as we did in note 35.)
61. This follows from the way the two relations would show up in embedded propositions. To illustrate: Suppose that thinking and thunking differ in some way, however slight it might seem. If so, it would be possible to think (consciously and explicitly) a proposition involving thinking and not at that time to be thinking (consciously and explicitly) the corresponding proposition involving thunking. For example, I am now thinking that I am thinking, but I am not now thinking that I am thunking. But I think that I am thunking iff I thunk that I am thinking. Since I am not now thinking that I am thunking, it follows that I am not now thunking that I am thinking. Thus, the proposition that I am thinking is not in the range of the thunking relation. But it is in the range of the thinking relation. So the two relations are not materially equivalent.
62. In Non-well-founded Sets, CSLI Lecture Notes, no.14 (Stanford: CSLI, 1988), Peter Aczel considers an axiom for set theory that implies that there is exactly one set x such that x = {x}. As far as I can see, there is no convincing support for this axiom. In any case, this axiom is not relevant to the point under discussion in the text, for the axiom is concerned with extensional entities, whereas we are concerned with intensional entities. Presumably, intensionality makes all the difference.
63. Ontological functionalists may also make this move, holding that the standard mental properties are "natural" second-order universals. Although the resulting view would then avoid the argument of this section, it would still be mistaken, for sections 1 and 2 show that the standard mental properties are first-order, not second-order.
64. The reflections in the text suggest another fault with the Armstrong-Lewis picture. Suppose, as their doctrine requires, that there are first-order physiological properties that satisfy A. Then, by a "diagonal" argument somewhat similar to that in the text, we can show that associated with them there would have to exist any number of deviant first-order properties also satisfying A. It follows that the Armstrong-Lewis Ramsified definitions would be mistaken. To avoid this problem, Armstrong and Lewis might try to mimic the strategy in die text by restricting the first-order physiological "realizations" satisfying A to "natural" universals. But it is wholly implausible uiat there are "natural" first-order physiological universals of die envisaged sort. Think of what it would be for first-order "realizations" R to behave with respect to one another in a fully A-like way. R2 (the "realization" of thinking) would need to hold between subjects and fine-grained propositions and would need to be ungrounded (for example, it should be possible that x R2 the proposition that, for some p, x R2 p, and it should be possible for this to occur independently of whether x R2 any other proposition). And now we are being asked to believe that R2 is a "natural" universal belonging to an actual physical science, namely, physiology! Certainly, physiologists would never have reason to posit the existence of a basic physiological relation meeting diese remarkable conditions. Anyone who seriously posits such a basic physiological relation surely is only "spreading the mind" onto die brain in an unscientific manner. (Corresponding posits of new "natural" physical properties by computer endiusiasts would likewise be unscientific "spreading of the mind" onto physical machines.)
65. Notice, incidentally, that if independent intuitive considerations show that the standard mental properties are "natural" universals, then the last remarks would lead to a new style of refutation of the identity thesis itself.
66. If correct, the considerations in this paper also seem to undermine efforts to reduce valuative properties to nonvaluative properties by means of Ramsification, for valuative properties exhibit a self-embeddability akin to that exhibited by mental properties.