What to Maximize if You Must

The assumption that decision makers choose actions to maximize their preferences is a central tenet in economics. This assumption is often justified either formally or informally by appealing to evolutionary arguments. In contrast, this paper shows that in almost every game, payoff maximization cannot be justified by appealing to such arguments. We show that in almost every game, for almost every distortion of a player's actual payoffs, some extent of this distortion is beneficial to the player because of the resulting effect on opponents' play. Consequently, such distortions will not be driven out by any evolutionary process involving payoffmonotonic selection dynamics, in which agents with higher actual payoffs proliferate at the expense of less successful agents. In particular, under any such selection dynamics, the population will not converge to payoff-maximizing behavior. We also show that payoff-maximizing behavior need not prevail even when preferences are imperfectly observed.(This abstract was borrowed from another version of this item.)