Commitment and weakness of will in game theory and neoclassical economics

This paper draws on theoretical and experimental studies in game theory and on the neoclassical concept of intertemporal inconsistency in choice to argue that the motivational theory shared by neoclassical economics and noncooperative game theory is mistaken in assuming that commitment never takes place in human decisions. The paper first gives two parallel examples, one of intertemporal inconsistency in a financial decision, and the other of noncooperative (subgame perfect) equilibrium in a game in extensive form. So far as one decision-maker is concerned, the two decisions are isomorphic, and both can be associated with weakness of will. By contrast, the cooperative analysis of the game (along the lines originally suggested by von Neumann and Morgenstern) predicts a different decision associated with commitment to a particular conditional sequence of "behavior strategies", i.e. a particular pure strategy. In effect the cooperative analysis assumes perfect strength of will. The paper then argues that strength of will and rationality of decisions are independent dimensions of a decision process and reviews some experimental evidence that suggests that both traditions are mistaken in their extreme assumptions about commitment or strength of will: neoclassical economics and noncooperative game theory in assuming that commitment never takes place, and cooperative game theory in assuming that it always does. The evidence indicates moreover that commitment is more likely in at least one context of value judgments than in its absence. The value context is reciprocity. It is suggested that a focus of research on the circumstances that favor commitment - rather than on modifications of an assumed utility function to accommodate non-self-regarding motivations - might lead to a more fruitful behavioral economics and behavioral game theory.