Abstract

Fifty years ago, Herbert Simon complained that the available models of rational choice were not feasible decision procedures for agents like us. These models involved variants on the theme of maximizing expected utility: the rational action for an agent is the one that is most likely to bring about outcomes that the agent prefers. Simon ’s complaints about these models included the now-familiar notions that human beings do not manage probabilities well, that we have at best radically incomplete utility functions, and that we lack the cognitive resources to calculate the expected utilities of even a few alternatives