Agent-Based Simulation for Generalized Matching Pennies Games

Abstract

In this paper, to investigate the long-run behavior of players in the generalized matching pennies game, we employ an approach based on adaptive behavioral models and construct an agent-based simulation system in which artificial adaptive agents have mechanisms of decision making and learning based on neural networks and genetic algorithms. We examine the strategy choices of agents and the obtained payoffs in the simulations, and compare the predictions of the Nash equilibria, the experimental data, and the results of the simulations with artificial adaptive agents. Moreover, we also seek similarities between the behaviors of the human subjects in the experiments and those of the artificial adaptive agents in the simulations.