Three Essays on Cooperation and Reciprocity

This dissertation aims to contribute to the literature of cooperation and social preferences. We use experimental and computational methods to understand the role and extent of reciprocity on cooperation. The first paper is a methodological contribution to the large literature on conditional preferences of cooperation. Cooperation generated by this type of preferences is notoriously unstable, as individuals reduce their contributions to the public good in reaction to other subjects' free-riding. This has led to the widely-shared conclusion that cooperation observed in experiments (and its collapse) is mostly driven by imperfect reciprocity. In this study, we explore the possibility that reciprocally cooperative preferences may themselves be unstable. We do so by observing the evolution of subjects' preferences in an anonymously repeated social dilemma. Our unsettling result is that, in the course of the experiment, a significant fraction of reciprocally cooperative subjects become egoistic, while the reverse is rarely observed. The non-selfish preferences that appear to be more stable are those most easily attributed to confusion. We are thus driven to the conclusion that egoism is more resistant to exposure to social dilemmas than reciprocity. The second paper the evolutionary success of conditional preferences by using simulations. We use an agent-based model in which agents play a variation of the iterated Prisoner's Dilemma game. We estimate the likelihood of cooperation levels as well as the likelihood of the existence of conditional types for different continuation probabilities. We show that an all-or-none type of conditional cooperation strategy together with the perfect conditional cooperation strategy are most likely to emerge when the continuation probability is sufficiently high. Our most surprising finding is related to the so-called hump-shaped strategy, a conditional type that is commonly observed in experiments. Our simulations show that those types are likely to thrive for intermediate levels of the continuation probability due to their relative advantage when probability of interaction is not enough to sustain a full-cooperation, but instead merely sustains mid-level cooperation. The third paper aims to understand the underlying reciprocal motives in altruistic behavior. We argue that the altruism that is revealed in dictator games can be explained by what we call presumptive reciprocity. Subjects may display non-selfish preferences because they presume that the other subjects would have revealed similar, non-selfish preferences if the roles had been reversed. This kind of intuitive reasoning, although partially captured by indirect reciprocity, is overlooked in the literature on social preferences, especially when it comes to explaining the behavior that appears to be purely altruistic. The experimental evidence we provide shows that people's choices reveal mostly presumptive reciprocity, while purely altruistic preferences play a much smaller role.