Paper Title
Collective Behavior in Spatial Prisoner’s Dilemma Game

Abstract
In this paper we study conditions of emerging collective behavior of agents acting in the 2-dimensional (2D) cellular automata (CA) space, where each agent plays locally the Prisoner’s Dilemma (PD) game. The system is modeled by 2D CA evolving in discrete moments of time, where each cell-agent changes its state according to a currently assigned to it rule. We assume existing of three rules: always cooperate (all-C), always defect (all-D), and k-D – defect only if more than k neighbors defect. Rules are initially assigned randomly to cells-agents, but during iterated game agents may replace their current rules by rules used by their neighbors. While each agent is oriented on maximization of its own profit in the game, we are interested in answering the question if and when a phenomenon of a global cooperation in a large set of agents is possible. We present results of experimental study showing conditions and degree of emerging such a cooperation. Index Terms - Collective behavior, multi-agent systems, spatial prisoner’s dilemma game, second order cellular automata.