Analysing the role of conditional dissociation in the evolution of cooperation, using agent-based modeling and evolutionary game theory

Abstract

In this seminar, we will analyse the role of conditional dissociation in the evolution of cooperation. Conditional dissociation, i.e. the option to leave your current partner in response to his behaviour, is a mechanism that has been shown to be able to promote the emergence and sustainability of cooperation in many social interactions. Nevertheless, this mechanism has always been studied in situations where conditional dissociation was combined with other factors that are also known to promote cooperation by themselves. In the seminar, we will study the isolated effect of conditional dissociation in the evolution of cooperation. To this end, we will study a finite population of individuals involved in a series of iterated Prisoner’s Dilemmas. Individuals in the model are subject to evolutionary pressures, which can be interpreted in strict biological terms, or as the result of a learning process by imitation. We consider the simplest possible strategy space that allows for conditional dissociation: Individuals either always cooperate or always defect, but they may choose to stay with their current partner, or leave him for a new random one, after having observed his action. Individuals without a partner are immediately re-matched at random. We study the evolution of cooperation in this evolutionary setup using both computer simulation and mathematical analysis, and show how this simple mechanism of conditional dissociation can lead to endogenous positive assortment of cooperators, within a polymorphic partially cooperative regime.