How (Not) to Justify Induction

Abstract

A conceptual analysis of the problem of induction suggests that the difficulty of justifying probabilistic reasoning depends on a mistaken comparison between deductive and inductive inference.

Inductive reasoning is accordingly thought to stand in need of special justification because it does not measure up to the standard of conditional absolute certainty guaranteed by deductive validity. When comparison is made, however, it appears that deductive reasoning is subject to a counterpart argument that is just as threatening to the justification of deductive as to inductive inference. Trying to explain induction in such a way that it satisfies a special justificatory requirement in contrast with deduction is

therefore not the way to justify induction. An alternative approach is sought in a style of justification developed by Aristotle for the law of noncontradiction and by Kant for the conclusions of transcendental reasoning that with variations can be used to justify both deduction and induction. This strategy upholds a

principle when the principle must be presupposed even to raise doubts about the principle's justification.