Abstract:We study the coupling dynamics between the evolution of cooperation and the evolution of partnership network on a geographical plane. While agents play networked prisoner's dilemma games, they can dynamically adjust their partnerships based on local information about reputation. We incorporate geographical features into the process of the agent's partner switching and investigate the corresponding effects. At each time step of the coevolution, a random agent can either update his strategy by imitation or adjust his partnership by switching from the lowest reputation partner to the highest reputation one among his neighbors. We differentiate two types of neighbors: geographical neighbors (i.e., the set of agents who are close to the focal agent in terms of geographical distance) and connectivity neighbors (i.e., the set of agents who are close to the focal agent in the partnership network in terms of geodesic distance). We find that switching to either geographical neighbors or connectivity neighbors enhances cooperation greatly in a wide parameter range. Cooperation can be favored in a much stricter condition when agents switch to connectivity neighbors more frequently. However, an increasing tendency of reconnecting to geographical neighbors shortens the geographical distance between a pair of partners on average. When agents consider the cost of geographical distance in adjusting the partnership, they are prone to reconnect to geographical neighbors.

Partner Switching Promotes Cooperation among Myopic Agents on a Geographical Plane.