FACULTY OF ARTS AND SCIENCES
UNIVERSITY OF PITTSBURGH
PITTSBURGH, PENNSYLVANIA 15213

DEPARTMENT OF PHILOSOPHY

29 November 1971

Professor Wilfrid Sellars,
353 Cathedral of Learning
University of Pittsburgh
Pittsburgh, Pennsylvania 15213.

Dear Wilfrid:

I said I would try to state why I have doubts about the adequacy of your treatment of belief in "Some Problems About Belief," and, linked with that, why I am unconvinced that one can transform all psychological verb constructions into that constructions, either beliefs that or intentions that, and concentrate the intentionality of wanting, fearing, worshipping in the implied belief that. (I would choose belief in, rather than belief that, as the more central case.)

This latter disagreement puts me in the Aristotelian camp, when the issue is presented as you present it in "Aristotle's Metaphysics," and in chapter 1 of Science and Metaphysics. (It may even make me a Platonist, but I am very happy about that.) In Science and Metaphysics you say that a complex like "this white thing" is essentially incomplete, a sentential fragment, and its complexity is derivative of the sentence complexity, or predicative character, of "this is white." In its occurrence in cases like "He fears this white thing," it appears that "this white thing" is occupying a place that could also be occupied by a that clause, so it functions there not as a sentential fragment, an incomplete thing, but as a sentence-substitute, as complete as a sentence would be. What is more, it is not easy to "stretch" it out into predicative form. One cannot transform it into a "fears that" construction. In conversation you indicated that yod would treat it as "He fears this thing and he believes it is white," or possibly "He fears this thing when (as long as) he believes it is white."

-2-

The former is more plausibles while the latter would be the reading for the (unlikely) "He fears this, as a white thing" (he has leukophobia).

I shall try to show what is lost when this doctrine of the ubiquitousness of belief in intentional contexts is combined with your account of the connection between the transparent and the opaque senses of belief. This is most easily done by looking at the distinctive implications of statements such as :

(a) "Jones fears Smith, as an examiner." ("admires" works even better here)

or

(b) "Jones wants this cube, as a die" (for his collection of dice).

What is interesting about these is that one can substitute co-referential expressions for "Smith" and "this cube," as long as one keeps the as phrase. Contrast this with the sentence:

(c) "Jones fears Smith."
(d) "Jones wants this cube."

Here "fears" and "wants" create straightforwardly s-intensional (substitution-intensional) contexts, as is also the case in:

(e) "Jones fears the examiner."
(f) "Jones wants this die" (for his dice collection), where substitution of "secretary of the orchid fanciers' club," on the basis of the fact that the examiner=Smith=the secretary of the orchid fanciers' club, would be illegitimate in (e) as in (c).

The fully explicit (a) displays not s-intensionality but q (qua)-intensionality. I suggest that it can explain the apparent s-intensionality of (c) and (e), if we suppose that a perspicuous rendering of any intentional sentence will contain an as-phrase, and that, when it is omitted, one assumes that the noun in the omitted as-phrase would duplicate the noun which is the direct object of the intentional verb. Thus (c) will be more perspicuously rendered as

-3-

(c) "Jones fears Smith, as Smith" (perhaps "the actual Smith," as Kaplan puts it; or, better, "Smith as such")

and (e) by

(e₁) "Jones fears the examiner, as examiner." (as such")

This would show why substitution of co-referential terms in (c) and (e), but not in (a), is illegitimate, because it is tantamount to alteration of the crucial as phrase noun. The advantage of the fully spelled out as-construction is that it provides a place for a term which allows substitution of co-referential expressions, and a place which is opaque, so it enables one to capture both the intentionality (description-relativity,) of the mental attitude and the fact that it is an attitude to something, in this case real, which can be referred to in several ways.

What would be your treatment of such a case? I am uncertain here, and I will welcome corrections, but, as I see it, you would transform (a) into something like

(a_s) "Jones shudders at Smith and/when Jones believes Smith is dangerous and/when Jones believes Smith is examiner."

I use "shudders at" as an abbreviation for the disjunction of behavioral dispositions which would be the residue of fear, when belief in the object's dangerousness is subtracted. I assume that this behavioral statement would be straightforwardly relational or transparent, so that we can substitute for "Smith" in the first occurrence in (a_s). Thus the intentionality of fearing is shifted to the intentionality of the belief which is a constituent of the fear.

First I note that the truth conditions for (a_s) are different from those for (a), whether we opt for "and" or "when." The "and" version would be true when the belief that Smith is examiner were not the ground of Jones' fear of Smith, and the "when" version could also be true if Jones feared Smith as blackmailer, not as examiner, but did continuously hold the belief that he was both blackmailer and examiner. Perhaps a better approximation to the sense of (a) could be given, but I can think of no adequate translation into that-constructions.

-4-

Second, I am not clear how you will get cross reference, in the belief attribution, to Smith as the thing shuddered at. You must quantify over individual concepts to express the beliefs, but you need cross reference not to the concept Smith, but to Smith himself, if "shudders at" is a behavioral relation. Will you construe it also as involving the concept Smith? If so, there will be no need to transform fears into beliefs, since fearing behaviour has now become intentional.

Thirdly, I note that there is a problem as to the sort of belief which is attributed to Jones in (a_s). It seems to meet your definition for transparent balief, in that Jones does ^tbelieve that Smith is dangerous, since he ^obelieves Smith as examiner is dangerous, and Smith as examiner is materially equivalent to Smith.

⁺Bfa= _df(Ei) iMEa. ^oBfi

seems satisfied, where a is "Smith," and the "i" in question "Smith as examiner." But if Jones transparently believes that Smith is dangerous, then he believes the secretary of the orchid fanciers' club is dangerous, and he shudders at him, so we can deduce that he fears the secretary of the orchid fanciers' club, and what is more, fears him as such. This is because the same analysis will be given for

"Jones knows Smith is secretary, and examiner, and fears him as examiner"

as for

"Jones knows Smith is examiner, and secretary, and fears him as secretary.

as long as he believes-that Smith is the secretary. But this is precisely the implication we wanted to block.

Perhaps there is a way in which you could so connect the various beliefs that it becomes evident that, although Jones both believes Smith is the secretary and also believes him to be the examiner, it is only in the latter capacity that Smith is said to be feared. I have not managed to find a satisfactory way, except by recognizing the complexity of as-constructions as separate from and not to be reduced to the complexities of that-constructions. (A counterfactual to the effect that Jones would not fear him, did he not believe him examiner, is too strong, since Jones fearing Smith in both capacities is compatible with "Jones believes Smith is secretary and fears him as examiner.")

-5-

As I see it, there are two things rolled into one in Quine's notion of a transparent belief context.

1. That the speaker commits himself to the actual existence of the object about which the believer believes something (E-extensionality).
2. That any true description of that object can be used in designating it in the belief report. (S-extensionality)

The second condition can fail when the first is satisfied, as in the case in the examples I have given. There is no reason to think that names matter only when they fail to denote things the speaker believes exists as your analysis clearly recognizes.

The first condition can fail when the second is met, in cases like

"Here they worshipped Diana as huntress, not Diana as earth mother," where one could substitutes for Diana, "the goddess sometimes worshipped as moon goddess." Mythical roses too, can smell sweetly under other names.

I suspect that your use of the ME relation as the mediator between opaque (intentional) and transparent (securing real reference) senses of belief hinders you too from recognizing all the varieties of mixes of s-intensionality, and e-intensionality, and I cannot see that it enables you to deal satisfactorily with q-intensionality.

You have two relations between individual concepts relevant to substitutivity -- ME, and =, where the latter is tantamount to synonymy. The former can be used only when the belief is 'of' (in Quine's and-your sense) existent objects, so only the latter can be appealed to in cases like worship of Diana. Yet such cases do admit of substitutivity on some kind of identity, as long as the as modifier is retained. Perhaps the real problem lies in the notion of identity. Was it correct of me to say that Smith as examiner ME Smith? Is Scott as author of Marmion materially equivalent to Scott as author of Waverly? I think my whole trouble could be put in the question -- how would you analyze

1. "I admire Scott as author of Marmion but not as author of Waverly."
2. "Diana was worshipped at Ephesus as earthmother and at Aphrodisia as hundress."?

-6-

A last query: The fundamental intentional context for you is not believe, nor intention, nor any other propositional attitude, but the semantic attitude of taking "red" as a red. This seems to be a paradigm of q-intentionality, and so I am puzzled as to why you want all the derivative cases of intentionality to take a propositional form, when the parent case does not??

With hopes of further enlightenment,

Annette Baier

AB:r