Reflections of a Bashful Bayesian: A Reply to Peter Lipton

Abstract

I am amazed by the degree to which Peter Lipton’s gentle irenic approach has lessened, if not eliminated, the gap between us. Prior to this discussion, we both agreed on certain fundamental points. (1) Both of us rejected the doctrine, enunciated by Gilbert Harman [Harman 1965], that all forms of nondemonstrative inference fall under the rubric of Inference to the Best Explanation. (2) We agreed that the Hypothetico-Deductive method is fatally flawed as a characterization of scientific confirmation. (3) We agree basically — though there may be some differences of detail — about the problems associated with Induction by Simple Enumeration. (4) Both of us considered Carl G. Hempel’s [Hempel 1965] Deductive-Nomological model an unsatisfactory characterization of scientific explanation. (5) Nevertheless, we agreed on the crucial importance of Hempel’s distinction between potential and actual explanations. (6) We agreed that, when scientists embrace hypotheses or theories, they have more than confirmation in mind; they are also concerned with informational content [Hempel 1965] p.23. This particular point will turn out to have immense import in our discussion; in fact, in my view it is the primary locus of our residual differences.