Evolutionary Ethics in Agent Societies

Abstract

To investigate whether the theory of evolutionary ethics is a valid explanation for the existence of ethical behaviour, we approach the ethical issues of morality from a computational point-of-view. We define a model of multi-agent societies, in which agents are able to evolve a moral sense. In the model, moral sense is defined by every agent’s personal set of rules, which determines its interaction with other agents. By performing simulations of the model we investigate under what circumstances the agents in the societies develop a moral sense which allows the society to thrive. We use four conceptually different agent designs: agents with a minimum of attributes, agents with family relations, agents with a memory, and agents with reputation in the society. From our results we conclude that there are circumstances under which agents evolve ethical behaviour, those circumstances being specific settings of family relations and reputation.