Abstract
Social dilemmas, and in particular situations where individuals interact with others without having ex-ante information on the length of their interaction horizon, are pervasive in our society.
Although wide, the literature studying the determinants of cooperation in infinitely repeated Prisoner Dilemmas hasn’t yet succeeded in fully explaining the variation in cooperation levels observed empirically. This paper - relying on both a meta-analysis and the collection of new experimental data – seeks to answer the following questions: can we distinguish individuals who cooperate only for strategic motives (profitability of future interactions based on game parameters) from individuals who cooperate for non-strategic motives? Can non-strategic motives be measured through social preferences? The meta-analysis shows game parameters perform disproportionally better as cooperation predictors in contexts where cooperation is an equilibrium. The experimental results confirm that social preferences do matter in shaping cooperation across a variety of strategically different contexts, highlighting the key role of beliefs, which serve as a transmission channel for the effect of social preferences on behavior.