Cooperation in an Adaptive Network

Abstract

We study the dynamics of a set of agents distributed in the nodes of an adaptive network. Each agent plays with all its neighbors a weak prisoner's dilemma collecting a total payoff. We study the case where the network adapts locally depending on the total payoff of the agents. In the parameter regime considered, a steady state is always reached (strategies and network configuration remain stationary), where co-operation is highly enhanced. However, when the adaptability of the network and the incentive for defection are high enough, we show that a slight perturbation of the steady state induces large oscillations (with cascades) in behavior between the nearly all-defectors state and the all-cooperators outcome.