Prisoner's dilemma on graphs with heterogeneous agents - Robotics Institute Carnegie Mellon University

Prisoner’s dilemma on graphs with heterogeneous agents

Lingzhi Luo, Nilanjan Chakraborty, and Katia Sycara
Conference Paper, Proceedings of 11th Annual Conference Companion on Genetic and Evolutionary Computation Conference (GECCO '09), pp. 2097 - 2102, July, 2009

Abstract

The prisoner's dilemma (PD) game has been used as a prototypical model for studying social choice situations with self-interested agents. Although in a single shot PD game, both players playing defect is a Nash equilibrium, in social settings, cooperation is usually observed among self-interested agents. The emergence of cooperation has been shown in the setting of iterated PD games and PD games on graphs. In this paper, motivated by modeling of conflict scenarios in multi-cultural societies, we study the PD game on a graph with multiple types of agents. We assume that there are two types of agents forming the nodes of the graph and the agents play the PD game with neighbors of the other type. The strategy update neighborhood of the agents can consist of either (a) neighbors of its own type only or (b) neighbors of its own type and the other type. We show by simulation that in both the above cases the fraction of players playing defect in the final solution is much more than the conventional case where there is no distinction between the game playing and strategy update neighborhoods (i.e., the agents are of the same type).

BibTeX

@conference{Luo-2009-10286,
author = {Lingzhi Luo and Nilanjan Chakraborty and Katia Sycara},
title = {Prisoner's dilemma on graphs with heterogeneous agents},
booktitle = {Proceedings of 11th Annual Conference Companion on Genetic and Evolutionary Computation Conference (GECCO '09)},
year = {2009},
month = {July},
pages = {2097 - 2102},
}