On knowledge of the unknowable

Abstract

If it is an unknown truth that p, it is an unknowable truth that it is an unknown truth that p (cf F. B. Fitch, 'A logical analysis of some value concepts', The Journal of Symbolic Logic 28, 1963, pp. 135-42, at p. 138). It follows, by classical logic, that if all truths are knowable then all truths are known. This hardish fact makes life difficult for the verificationist who wishes to assert that all truths are knowable, but to deny that all truths are known. He might try rejecting classical logic (cf. my 'Intuitionism disproved?', ANALYSIS 42.4, October, 1982, pp. 203-7). Dorothy Edgington has recently suggested a different way out (cf. 'The paradox of knowability', Mind 94, 1985, pp. 557-68). She admits that there are unknowable truths in the sense of Fitch's argument, but interestingly revises the verificationist principle, to give it this form: if, in a situation s, it is a truth that p, then there is a possible situation s' in which it is known that, in s, p. Although the truth (for some values of '/?') that, in s, it is an unknown truth that p cannot be known in s, it might be known in some possible situation other than s. I have argued elsewhere that this account faces difficulties of principle in explaining how the knower in s' can pick out the situations in thought with adequate specificity (cf. 'On the paradox of knowability', Mind forthcoming). I waive those objections here, to examine Edgington's application of her principle to a particular case of knowledge in one situation that something is an unknown truth in another. I shall argue that her account does not generalize to the full range of cases which it needs to cover.