A computational model of emotion assessment influenced by cognition in autonomous agents

Abstract In this paper, we present a computational model of emotions based on the context of an Integrative Framework designed to model the interaction of cognition and emotion. In particular, we devise mechanisms for assigning an emotional value to events perceived by autonomous agents using a set of appraisal variables. Defined as fuzzy sets, these appraisal variables model the influence of cognition on emotion assessment. We do this by changing the limits of fuzzy membership functions associated to each appraisal variable. In doing so, we aim to provide agents with a degree of emotional intelligence. We also defined a case study involving three agents, two with different personalities (as a cognitive component) and another one without a personality to explore their reactions to the same stimulus, obtaining as a result, a different emotion for each agent. We noticed that emotions are biased by the interaction of cognitive and affective information suggesting the elicitation of more precise emotions.