Why and How to Build Emotion-Based Agent Architectures

94 Motivation There are many reasons for researchers and developers to be interested in creating computational models of, and agent architectures inspired from, affective phenomena. Affective phenomena include core affect, mood, emotion, as well as personality. This chapter discusses useful terminology and specific theories of affective phenomena and introduces some of the main motivations for this topic. Building computational models of the roles of affective phenomena in human cognition is of interest to the cognitive science community. The main objective of cognitive science is to understand the human mind by developing theories of mind, creating computational models of these theories, and testing whether the input/output and timing behaviors of the resulting systems correspond to human behaviors (Thagard, 2008). Computational models of emerging cognitive and affective science theories of human emotion and affect will enable us to shed new light on the complexity of human affective phenomena. Building emotion—or affect-based agent architectures—is also useful in subfields of computer science, such as artificial intelligence (AI), human-computer interaction (HCI), among others. AI, which focuses on developing algorithms to make rational intelligent decisions, can simulate and emulate the functional roles of affect and rational emotions in human decision making (Johnson-Laird & Oatley, 1992; Picard, 1997; Lisetti & Gmytrasiewicz, 2002). HCI on the other hand, is concerned with creating artificial agents that can adapt to users’ emotion or personality to enhance adaptive human-computer interaction (Hudlicka, 2003; Picard, 1997). The interest in building emotion-based agent architectures revolves around the notion that Abstract