Towards a unified model of rational belief.

The provision of an adequate formal theory of rational belief and its dynamics is a concern that is shared by many researchers, across many disciplines. The two principal frameworks on offer are the Bayesian model, which represents the states of mind of rational agents, in a graded fashion, by means of a function mapping sentences onto corresponding degrees of confidence, and the AGM model, which represents these states of mind, in a binary fashion, by means of sets of believed sentences. Whilst it is clear that there is a relation between degrees of confidence and belief, the nature of this relationship has been the subject of a particularly vexed and long-standing debate. Quite remarkably, all existing proposals fall foul of one or another of several intuitively compelling constraints. The proposed research project will build on preliminary results obtained by the applicant, outlining a complete set of desiderata on the aforementioned mapping and developing a proposal that uniquely satisfies all of these. The account will be general enough to allow for the modelling of agents capable of entertaining sophisticated beliefs involving conditionals (e.g. If A then B) as well as so-called epistemic modalities (e.g. It might be the case that A).

Keywords:Probability, AGM, Rationality, Belief revision, Uncertainty, Formal epistemology

Disciplines:Other philosophy, ethics and religious studies not elsewhere classified, Theory and methodology of philosophy, Philosophy, Ethics, Artificial intelligence, Cognitive science and intelligent systems