Can we manage first impressions in cooperation problems? An experimental study on "Broken (and Fixed) Windows"

Cooperation problems are at the heart of many everyday situations. In this paper, we propose a very simple and light-handed mechanism to sustain cooperation and test its performance in a rich laboratory environment. The mechanism moderates cooperation by controlling experiences, more specifically, it manipulates subjects' initial beliefs by providing them with selective information about (un)cooperative behavior in other, unrelated, groups. We observe that contributions are considerably sensitive to such selective information. First impressions participants happen to make predict subsequent behavior. Our results, however, suggest an asymmetry in the strength of the reaction - which might pose a limit on the effectiveness of the mechanism in natural settings.