Selfishness, Cooperation and Topology Dynamics in Networks

We study the dynamics of a network of agents (players) under two models settings. In the first model, players (agents) extend heterogeneous cooperation levels to their partners and in the second, perfectly rational and selfish agents engage in an iterated Snowdrift Game and the evolution of the system suggests that cooperation emerges and prevails even when no external rules or norms for punishing defectors are imposed. Agents base their decisions purely on local information and the network rewiring rules are such that agents pair-up with only recommended coplayers (this rules assumes the reason of perfect rationality). A cyclical behavior in the number of cooperators is observed in the second setting which is dependent upon its information update model parameter denoting the number of proactive agents in the population. The randomness (cyclic behavior in the number of cooperators) is lost when the number of agents who update proactively is small. Our results from the second model show that inspite of not imposing any societal norms, cooperation was prevalent in a rational population. We also present the simulation results of our first model (asymmetric cooperation system) and comment on some intersting observations where some agents are happy being exploited and some others exploit their co-players by leveraging on their degree of connectivity.