Bayesian Conditioning Revisited

Richard Bradley (LSE) gives a talk at the MCMP Colloquium (27 June, 2012) titled "Bayesian Conditioning Revisited". Abstract: Bayesian conditioning is widely regarded as the correct way to revise your degrees of belief in circumstances in which experience leads you to believe, with certainty, that some proposition is true. But different revision rules have been proposed for other types of experience: for example, Jeffrey conditioning when your degrees of belief for some set of propositions changes without your becoming certain of any of them; or Adams conditioning, when you conditional degrees of belief change. In this talk I will argue that these rules are all characterised by two properties - Responsiveness to experience and Conservativeness (in the sense of conserving any beliefs on matters on which experience is silent) - and that what distinguishes them is just the nature of the experiences to which they respond. Furthermore they are the only rules with these properties defined on these experiences.