An experimental study of the repeated trust game with incomplete information

Abstract

In the trust game first player 1 decides between non-cooperation or trust in reciprocity. Only in the latter case player 2 chooses between exploiting or rewarding player 1. In our computerized experiment, player 2 can be always rewarding (implemented as a robot strategy) or decide opportunistically. To allow for reputation formation, this base game is played repeatedly. Learning can be analyzed since participants play successively several repeated games with changing partners. Participants can explicitly rely on mixed strategies which allows testing the qualitative and quantitative aspects of reputation equilibria also at an individual level.