Off to Rome next week (May 13, h 16, Aula Elio Matassi) to give a PhD seminar talk:
One, but not the same: Bayesian confirmation theories and their implications
Abstract. According to a Bayesian view of confirmation, the relation of support between evidence and hypotheses can be analyzed in probabilistic terms, where probabilities are meant to represent quantitative credences of a rational agent. Bayesians largely agree that the qualitative notion of confirmation (disconfirmation) amounts to an increase (decrease) of the probability of the hypothesis once the evidence is given. However, as soon as we move from this purely qualitative idea to ordinal assessments of stronger / weaker evidential support, a plethora of substantially divergent explications can be plausibly put forward. This is a fairly well-known fact, forcefully pointed out in now classical work by Fitelson (1999) and Festa (1999). Importantly, the very same issue of measure sensitivity arises, it turns out, for a number of further explicanda of interest in formal epistemology, such as coherence, explanatory power, accuracy, and informational entropy. I will present recent developments concerning the axiomatic characterization of alternative candidate measures of Bayesian confirmation and discuss how this plurality of accounts can be interpreted. Against both monist and deflationist attitudes found in the literature, I will argue that a considered choice of constraints motivates a tempered form of pluralism concerning Bayesian confirmation theories. Examples discussed will include a novel explication of the classical philosophical idea of partial entailment.