UTILITY-THEORY WITH PROBABILITY-DEPENDENT OUTCOME VALUATION

This paper provides an axiomatic characterization of a class of nonexpected utility models of decision making under risk. It formalizes the notion that the valuation of outcomes may depend on the probability measure in whose support the outcomes are contained and defines a transformation of the space of risky prospects taking account of these probability-dependent valuations. The von Neumann-Morgenstern axioms are imposed on the preference relation among the transformed risky prospects to obtain a representation that is nonlinear in the probabilities. Some implications of the resulting models are pointed out using local expected utility analysis. © 1992 Academic Press, Inc.