Learning in a black box

We study behavior in repeated interactions when agents have no information about the structure of the underlying game and they cannot observe other agents' actions or payoffs. Theory shows that even when players have no such information, there are simple payoff-based learning rules that lead to Nash equilibrium in many types of games. A key feature of these rules is that subjects search differently depending on whether their payoffs increase, stay constant or decrease. This paper analyzes learning behavior in a laboratory setting and finds strong confirmation for these asymmetric search behaviors in the context of voluntary contribution games. By varying the amount of information we show that these behaviors are also present even when subjects have full information about the game.