Game theory without decision-theoretic paradoxes

Abstract

Most work in game theory is conducted under the assumption that the players are expected utility maximizers. Expected utility is a very tractable decision model, but is prone to well-known paradoxes and empirical violations (Allais 1953, Ellsberg 1961), which may induce systematic biases in game-theoretic predictions. La Mura (2009) introduced a projective generalization of expected utility (PEU) which avoids the dominant paradoxes, while remaining quite tractable. We show that every finite game with PEU players has an equilibrium, and discuss several examples of PEU games. © 2009 Springer-Verlag Berlin Heidelberg.