Predicting How People Play Games: Reinforcement Learning in Experimental Games with Unique, Mixed Strategy Equilibria

by: Ido Erev, Alvin E Roth

The American Economic Review, Vol. 88, No. 4. (1998), pp. 848-881.

View FullText article

DOI, JSTOR

Reviews [Write a review of this article]

There are no reviews of this article

Find related articles from these CiteULike users

krista, dep, yish, Glimcher_Lab (group), mathgamespatterns (group), dtl (group), vds-arg (group), eni (group), mullonc

Find related articles with these CiteULike tags

computational, decisionmaking, economics, economy, experiment, game, gametheory, learning, montecarlo, motivation, no-tag, reinforcement, reward, theory

Abstract

We examine learning in all experiments we could locate involving 100 periods or more of games with a unique equilibrium in mixed strategies, and in a new experiment. We study both the ex post ("best fit") descriptive power of learning models, and their ex ante predictive power, by simulating each experiment using parameters estimated from the other experiments. Even a one-parameter reinforcement learning model robustly outperforms the equilibrium predictions. Predictive power is improved by adding "forgetting" and "experimentation," or by allowing greater rationality as in probabilistic fictitious play. Implications for developing a low-rationality, cognitive game theory are discussed.

@article{citeulike:404918, abstract = {We examine learning in all experiments we could locate involving 100 periods or more of games with a unique equilibrium in mixed strategies, and in a new experiment. We study both the ex post (\"best fit\") descriptive power of learning models, and their ex ante predictive power, by simulating each experiment using parameters estimated from the other experiments. Even a one-parameter reinforcement learning model robustly outperforms the equilibrium predictions. Predictive power is improved by adding \"forgetting\" and \"experimentation,\" or by allowing greater rationality as in probabilistic fictitious play. Implications for developing a low-rationality, cognitive game theory are discussed.}, author = {Erev, Ido and Roth, Alvin E. }, citeulike-article-id = {404918}, doi = {http://dx.doi.org/10.2307/117009}, journal = {The American Economic Review}, keywords = {economics, gametheory, learning, reinforcement}, number = {4}, pages = {848--881}, posted-at = {2006-05-31 15:12:53}, priority = {3}, title = {Predicting How People Play Games: Reinforcement Learning in Experimental Games with Unique, Mixed Strategy Equilibria}, url = {http://dx.doi.org/10.2307/117009}, volume = {88}, year = {1998} }

RIS record

TY - JOUR ID - citeulike:404918 L3 - citeulike-article-id:404918 N2 - We examine learning in all experiments we could locate involving 100 periods or more of games with a unique equilibrium in mixed strategies, and in a new experiment. We study both the ex post ("best fit") descriptive power of learning models, and their ex ante predictive power, by simulating each experiment using parameters estimated from the other experiments. Even a one-parameter reinforcement learning model robustly outperforms the equilibrium predictions. Predictive power is improved by adding "forgetting" and "experimentation," or by allowing greater rationality as in probabilistic fictitious play. Implications for developing a low-rationality, cognitive game theory are discussed. IS - 4 JF - The American Economic Review EP - 881 TI - Predicting How People Play Games: Reinforcement Learning in Experimental Games with Unique, Mixed Strategy Equilibria VL - 88 SP - 848 KW - economics KW - gametheory KW - learning KW - reinforcement AU - Erev, Ido AU - Roth, Alvin PY - 1998/// UR - http://dx.doi.org/10.2307/117009 ER -

RIS BibTeX RTF/PDF