A Multi-agent Based Migration Model for Evolving Cooperation in the Spatial N-Player Snowdrift Game

The Snowdrift game is a well-known social dilemma model frequently used in evolutionary game theory to investigate the emergence of cooperative behaviour under different biologically or socially plausible conditions. In this paper, we examine a multi-player version of the Snowdrift game where (i) the agents playing the game are mapped to the nodes of a regular two-dimensional lattice, (ii) the number of rounds of the game varies from a “one-shot” version to a fixed number of repeated interactions, and (iii) a genetic algorithm is used to evolve agent actions (strategy update) over a fixed number of generations. Comprehensive Monte Carlo simulation experiments show that cooperative behaviour is promoted in the multi-player iterated Snowdrift game. This emergent behaviour may be attributed to the combination of spatial reciprocity, based on the inherent capabilities of the genetic algorithm to explore the diverse sets of agents’ strategies, and repeated interactions. The simulation results also uncover some interesting findings regarding the effect of repeated interactions in the game.