Towards a transcendental Justification of Induction

Abstract

With the discovery of Bayesian inductive logic optimism about the possibility of rationally justifying inductive inference has been renewed. The justification of Bayesian inductive rules is known as the Dutch-book argument (Ramsey-de Finetti theorem). The question dividing theoreticians of induction is whether such an argument can really justify Bayes' and Jeffrey's conditionalization rules (Bayesian inductive rules). In this paper, I will be interested, first, in distinguishing two senses of justification of inductive inference: on the one hand, persuasive justification of induction and, on the other, its explicative justification. Secondly, I will relate the problem of the justification of Bayesian conditionalization rules with Hume's classical problem of persuasively justifying causality and induction. Thirdly, I shall argue that the Dutch-book argument cannot persuasively justify Bayesian conditionalization rules in agreement with Hume's negative thesis concerning non-circular justification of inductive inference of a causal type. Finally, I will suggest a sort of explicative justification of Bayesian inductive rules in terms of a transcendental argument of kantian inspiration but davidsonian in its style.