Abstract

Most philosophers associate the concept of evidential relevance with the practice of confirmation, but the distinction between confirmation and positive relevance has not been adequately analyzed in the literature. This is in large part due to the predominate Bayesian position that cases of confirmation are co-extensive with cases of positive relevance. Yet there seem to be cases in which evidence is intuitively relevant to a hypothesis, but in which the evidence does not boost its credibility. For example, a positive instance of a falsified universal hypothesis does not boost its credibility, while it seems wrong to deny that the positive instance of a claim is relevant to it. ;This dissertation is an investigation of the nature of evidential relevance. I consider, and reject, construing Hypothetico-Deductivism and Hempel's instance approach as accounts of mere evidential relevance. Much attention is given to Glymour's "bootstrapping" approach, which I argue is a promising approach to evidential relevance, but that the account is beleaguered with difficulties, most importantly a variety of "irrelevance" counterexamples. Attempts to revise the account through natural axiomatization, relevance logic, J. Earman's "Pincer Strategy", and other means are examined and found inadequate. ;I propose and defend an evidential relevance relation which similar in spirit to Glymour's. An important difference is the elimation of the bootstrapping feature. The result is a relation of evidential relevance that retains all the virtues of Glymour's account, yet avoids its failures. ;I apply my new account to Woodward and Bogen's distinction between "data" and "phenomena". I argue that, although the distinction is a useful one, as it is presented by these authors it is ambiguous. By connecting data with phenomena through my new relevance relation, the distinction becomes sharpened and other shortcomings of the account are surmounted. ;Finally, I consider the strengthening conditions needed to promote evidential relevance to the ranks of confirmation. My proposal amounts to the inclusion of a weak uniformity of nature postulate into the auxiliary set of the relevance relation. The new confirmation relation is then tested against historical cases and is found to be satisfactory