Learning about Learning in Games through Experimental Control of Strategic Interdependence
Jason Shachat, J. Todd Swarthout
Journal of Economics Dynamics & Control
#002151 20131014 (published) Views:138
We report results from an experiment in which humans repeatedly play one of two games against a computer program that follows either a reinforcement or an experience weighted attraction learning algorithm. Our experiment shows these learning algo- rithms detect exploitable opportunities more sensitively than humans. Also, learning algorithms respond to detected payoff-increasing opportunities systematically; how- ever, the responses are too weak to improve the algorithms’ payoffs. Human play against various decision maker types does not vary significantly. These factors lead to a strong linear relationship between the humans’ and algorithms’ action choice propor- tions that is suggestive of the algorithms’ best response correspondences.