Hume’s problem, epistemic deductivism and the validation of induction

• Main Author: Rea, George
• Published: Durham University 2005
• Subjects: 161
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.426432

Contrary to Owen (2000), Hume's problem is, as has traditionally been supposed, a problem for the justification of inductive inference. But, contrary to tradition, induction on Hume's account is not deductively invalid. Furthermore, on a more modem conception of inductive or ampliative inference, it is a mistake to suppose that the proper construal of an argument explicating the supposed justification for such inferences should in general be non-deductive. On a general requirement for argument cogency that arguments should be suitably constructed so as to make it clear to the audience that the subject is justified, on whatever basis is cited, in regarding the hypothesis with whatever epistemic attitude the arguer purports to be so justified, arguments in general, fully explicated and properly construed, should be deductively valid. Hume’s problem does not prevent such justification because his crucial argument establishes only that our basic assumptions cannot be justified, in the sense of being 'proven', or shown by non-question-begging argument to be just. It does not establish that our basic assumptions, properly explicated, are not just, or that they are not (at least to the satisfaction of most of us) clearly so. Nor does Goodman's 'new riddle' of induction pose a serious problem for the justification of our inductive inferences, as is still commonly suggested, since Jackson figured out the solution to the riddle thirty years ago. There is an analogous problem to Hume’s for the provability of principles or claims of deductive inferability, and if my analysis of the proper construal structure of argument (in the natural sense) is correct, this will block Howson's (2000) proposed escape route. Nevertheless, as with the case of induction, the unprovability of basic claims and principles of deductive inferability does not bar their deployment in cogent justifications.