Experience-weighted Attraction Learning in Normal Form Games

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 (belief learning) as special cases, and hybridizes their key elements. Using three sets of experimental data, the authors show that reinforcement and belief learning are generally rejected in favor of EWA. EWA is able to combine the best features of these approaches, allowing attractions to begin and grow flexibly as choice reinforcement does but reinforcing unchosen strategies substantially as belief-based models implicitly do.