Abstract

The dissertation gives an account of the principles guiding the rational revision of belief. I develop a non-probabilistic account of belief revision. My central thesis is the claim that there are two quite different kinds of rational belief change; two methods suited to two different sorts of situation. I call these methods updating and supposing. This claim, presented in Chapter Two, is argued on the basis of results proved in Chapter One. Chapters Three and Four are applications of the distinction between updating and supposing to two areas of current philosophical debate: the semantics of conditionals; and the theory of deliberation dynamics. The discussion of conditionals appeals to the temporal asymmetry in knowledge. Why is it that we know more about the past than about the future? Chapter Five offers a new answer to that old question, based on an account of recording devices as physical systems stable under perturbation. ;I hope to have shown that many of these philosophical questions--questions about belief revision, decision theory, and conditionals, that have traditionally been framed in probabilistic terms--can be most fruitfully discussed without ever mentioning probability