Problem DefinitionThe purpose of this simulation is to explore the dynamics and effectiveness of social motives in an iterated "Prisoner's Dilemma" game. How will social motives evolve in a given Prisoner's Dilemma environment, what motives are most effective under what conditions, and are there any motives and conditions that result in a near optimal equilibrium? The simulation has both ecological and evolutionary components. Ecologically, agents coded for strategies interact with each other and then are allowed to reproduce based on how close to optimal their score in the game is. Because of the way the agents interact with each other, and due to the nature of the Prisoner's Dilemma in which the score of a player depends on the actions of its opponent, the interacting strategies are dependent on each other. The ecological aspect of the simulation shows the dynamics of dependance among strategies, and also provides the environment for social motives to evolve within. On an evolutionary level, agents have a simple genome coded which represents the values for calculating their individual social motive. When an agent reproduces, its offspring gets the same genome. Because reproduction is based on how close an agent is getting to the optimal score, agents with more successful social motives will reproduce more than agents with less successful social motives. The genomes for agents may mutate over time, enabling the potential for both better and worse social motives. The evolutionary aspect of the simulation attempts to locate the social motives which provide the fittest (most optimal) players for the given Prisoner's Dilemma environment. |