Item

Abstract

We present the results of an experiment on learning with minimal information. Particularly, subjects are only provided with feedback about their own payoff from the last period of the game being played, but not with information about the structure of the game. We compare the empirical structure of the decision algorithm for this setting with the empirical structure of algorithms for subjects who receive sufficient information to learn the game. The laboratory data show that, depending on the information setting, players adjust their strategy choice differently. The structure of the decision algorithm for subjects operating with minimal information indicates myopic responses to success, while the structure for sufficiently informed players is more complex. As a consequence, sufficiently informed players outperform players who have minimal information in a simple coordination game. Yet, if the structure of the game changes, readjustment is more successful for the players operating with minimal information.