Blackwell Reference Online

Extract

Rationality is standardly divided into the practical and the theoretical. Practical rationality concerns what we should do; theoretical rationality concerns what we should believe. In telling us to perform that act with the greatest “expected utility” (a maxim associated with Bayes 1763, although the idea predates him—see Jeffrey [1965] 1983:1, 20–1), decision theory concerns the practical. But, in many of its guises (see Fishburn 1981, for an extensive sampling of decision theories), the theory also apparently concerns what we should believe, since it tells us to conform our degrees of belief (intuitively, degrees of confidence that things are or will be so) to the standard Kolmogorov (1933) probability axioms. In one common approach a key goal is to lay down the minimal qualitative conditions on the rationality of preference that are sufficient to prove a representation theorem to the effect that, if an agent obeys the qualitative preference axioms, then certain quantitative facts are true of that agent: not only are his or her preferences ranked in accord with their expected utilities, but also the agent's degrees of belief obey the standard probability axioms. An issue that arises, however, is the extent to which degrees of belief are practical as opposed to theoretical. I shall focus initially, however, on a theory, due to von Neumann and Morgenstern ([1944] 1953), that takes ... log in or subscribe to read full text