Expected Utility for Probability Measures

Abstract

One of the first axiomatizations of expected utility for nonsimple probability measures was given by Blackwell and Girshick (1954). They applied ≻ to the set of discrete probability measures on a set C of consequences, using Al, A3 and a denumerable generalization of A2. Although utilities need not be bounded when only simple measures are used, Blackwell—Girshick utilities must be bounded. Fishburn (1967, 1970, 1975a), DeGroot (1970), and Ledyard (1971) later extended the expected-utility form to other sets of probability measures.