In 'experience-weighted attraction' (EWA) learning, strategies have attractions which reflect initial predispositions, are updated based on payoff experience, and determine choice probabilities according to some rule (e.g., logit). EWA includes reinforcement learning and weighted fictitious play...

The authors find that subjects' behavior in an incomplete-information, repeated-game experiment is roughly a sequent ial equilibrium. The deviations from sequential equilibrium can be ex plained by the existence of a "homemade prior" probability (about 1 7 percent) that players will cooperate...