Modeling Emotions and Reason in Agent-Based Systems

We analyze how to develop an agent-based system in which agents evolve co-evolutionary endogenous rules of behavior by using best response and emotions. We show that best response is not sufficient to define complete and consistent rules of behavior and we prove that the use of emotions, which complement reason, is necessary to learn rules of behavior. We model four different emotions (apathy, patience, anger and confidence) which enable the agent to deal with the rewards and with others. We propose an algorithm to model automata-based systems incorporating rationality and emotions.