Date Added: Dec 2010
Most applications of game theory assume equilibrium, justified by presuming that learning will have converged to one; or in settings where that is implausible, that equilibrium approximates people's strategic thinking without learning. Yet recent experimental work suggests that initial responses to many kinds of games deviate systematically from equilibrium, and that certain nonequilibrium models can then out-predict equilibrium models of thinking. Even when learning converges to equilibrium, such nonequilibrium models of initial responses allow better prediction of history-dependent limiting outcomes. This paper reviews recent theoretical and empirical work on nonequilibrium models of strategic thinking and illustrates their applications in economics.