Philosophy 257 Behavior and Other Minds

Asst. Prof James Pryor Dept. of Philosophy

Dummett's Argument for Anti-Realism

1. If realism about the truth-conditions of S is correct, then understanding of S must consist in a "grasp" of S's truth-conditions (i.e., in knowledge of S that they are its truth-conditions).
2. Understanding of S can't consist in knowledge of S that it has certain truth-conditions.
3. So realism about the truth-conditions of S has to be rejected.

Let's turn then to a closer examination of Dummett's argument:

1. A realist will say that a sentence S for which we have no effective procedure for establishing whether or not it's true, nevertheless has a definite truth-value. The realist says that such a sentence S has verification-transcendent truth-conditions, that is, truth-conditions which might obtain undetectably, or in the absence of any possible evidence.

2. If 1, then the realist has to say that your understanding of S consists in or could be explained by knowledge of S's truth-conditions, that is, knowledge of S that they are its truth-conditions.

   Dummett writes:

   We have to be clearer about what it means to say that the meaning of a sentence consists in its truth-condition... Philosophical questions about meaning are best interpreted as questions about understanding: a dictum about what the meaning of an expression consists in must be construed as a thesis about what it is to know its meaning. So construed, the thesis becomes: to know the meaning of a sentence is to know the condition for it to be true. (Seas of Language, p. 35)

   A theory of meaning is a theory of understanding; that is, what a theory of meaning has to give an account of is what it is that someone knows when he knows the language, that is, when he knows the meanings of the expressions and sentences of the language. (Seas of Language, p. 3)

   I do not believe the realist has to accept these claims. See Principle 3 in our discussion of truth.

3. The manifestation constraint. Any knowledge which partly constitutes or explains a speaker's understanding of S has to be fully manifestable in that speaker's "use" of S--his linguistic behavior.

   For some of Dummett's arguments for this claim, see: Seas of Language, pp. 21, 37, 46; Truth and Other Enigmas, pp. 216-17, 224; and Frege: Philosophy of Language (2nd ed.), pp. 461-2 and 467-8.

   The manifestation constraint seems to assume that linguistic abilities can be given a behaviorist analysis. Does it assume this? If it does, why should we believe that linguistic abilities can be given any such analysis? (Cp. the ability to add, or the ability to rotate images in one's mind, and the ability to remember who won the World Series in 1932.)

   A typical passage where Dummett insists on the manifestation constraint is: There is no content to an ascription of an implicit knowledge of what that condition is, since there is no practical ability by means of which such knowledge may be manifested. (Seas of Language, p. 46) Certainly it means something to ascribe implicit knowledge to a person even if that knowledge is not manifested, and we have no good evidence for the ascription. Such an ascription "has no content" only in the sense that there is no apparent reason to believe it. Is Dummett assuming here that an ascription of implicit knowledge can be true only if it's possible to have good evidence for its truth? Why should we accept that, at this stage in the argument?

4. In some cases, your knowledge of what S means can be manifested in your ability to say what S means, but this can't generally be the case. For some S, your knowledge of what S means has to manifested non-verbally.

   For some of Dummett's arguments for this claim, see Seas of Language, p. 36, 44-5, 57-8, 101, 270-1; Truth and Other Enigmas, pp. xxxiii-xxxv, 217, 224; and Frege: Philosophy of Language (2nd ed.), pp. 460-61.

   This is a sketchy and unconvincing part of Dummett's argument, partly because it's not clear what motivates the manifestation constraint, in the first place. In different places he appeals to different considerations, none of which are fully convincing.

5. Knowledge of S that it has verification-transcendent truth-conditions cannot be fully manifested non-verbally. Only a grasp of such things as: when it's appropriate to assert S, what counts as evidence in S's favor, and so on, can satisfy the manifestation constraint.

6. The acquisition constraint. Any knowledge which partly constitutes or explains a speaker's understanding of S has to be acquirable on the basis of what anyone who masters our standard linguistic behavior with the sentence S thereby learns.

7. Knowledge of S that it has verification-transcendent truth-conditions doesn't satisfy this acquisition constraint. We could not have acquired a conception of what it is for S to be true undetectably, merely on the basis of what we learned when we mastered the use of S.

   For some of Dummett's arguments for this claim, see: Seas of Language, p. 47; Truth and Other Enigmas, pp. 217-18, 362-4; and Frege: Philosophy of Language (2nd ed.), p. 467.)

8. So by 3 through 5 and by 6 through 7, our competence with or understanding of S can't consist in a grasp of S's verification-transcendent truth-conditions.

9. So by 2 and 8, we have to reject realism. We don't understand S to have verification-transcendent truth-conditions.

10. So by 9, if we understand S, then what it means has to be "epistemically constrained," that is, fully determined by the sorts of possible evidence we could have for or against it. We can have no conception of what it would be for S to be undetectably true, or true in the absence of any possible evidence.