Punishment, defined as the behavior that inhibits instigators and prevents them from repeating cheating, is widely observed as a means of resolving conflicts among selfish individuals. Due to punishment, cheaters suffer the cost that exceeds the benefit gained by defection. As a result they lose the incentive to cheat, thus cooperation is stably maintained in a population. In this seminar I will talk about two separate topics on evolution of punishment; in humans and in social insects.
[1] Cooperation between non-relatives in humans is often characterized by indirect reciprocity. In order to realize discriminate altruism, donors of help pay attention to social information of recipients, such as reputation or rumor, to judge whether he is worth cooperating or not. Recent theoretical studies have shown that the first-order information of recipients, that is whether he had cooperated or not in their previous interaction, is not sufficient to maintain indirect reciprocity, suggesting the limit of image-scoring criterion (Nowak and Sigmund. 1998). Here I explore moral criteria which make the discriminating strategy evolutionarily stable. I formulate moral criteria mathematically as “reputation dynamics”, and have exhaustively searched ESSs from all the possible combinations of moral and strategy (4096 combinations). As a result I found the best eight combinations called “reading eight”. The reading eight tells us second-order information of recipients, especially its ability to discriminate justified defection from unjustified ones, is critical in indirect reciprocity.
[2] In many eusocial hymenoptera, workers retain their functional ovaries and are able to lay unfertilized haploid eggs (male). The theory of kin selection predicts that workers disfavor male-production by other workers thus police their eggs (worker-policing) only when their queen mates more than two times. However, even in monandrous colonies worker policing is widely observed, suggesting other benefits than from relatedness. Here I study workers' optimal reproductive scheduling as dynamic game and investigate its Nash equilibria. Results show that male-reproduction at an early (ergonomic) stage is always mal-adaptive irrespective of relatedness patterns in the colony, hence worker policing is favored there. This is because investment in sexual individuals at an early stage greatly reduces initial colony productivity and incurs much loss in inclusive fitness later.