Agents , Emergence , Emotion and Representation Darryl

This paper presents an analysis of issues pertaining to the relation between computational emergence and emotion in cognitive agent systems. We consider how a developing computational theory of cognition can be used to monitor and manage interactions with and within complex systems. Goal-based agent systems sometimes need to postpone or abandon important goals in favor of more immediate concerns. In more sophisticated goal-based agent systems goals may need to be abandoned due to conflict of motivations. This can lead to the emergence of perturbant behavior analogous in type to (negative) emotion-like states. Agent systems need mechanisms to recognize this type of scenario or otherwise risk compromising their design rationale. Through the development and use of a dynamic representation of emotion, a computational agent can harness unwanted and emergent states and behaviors before the agent system becomes dysfunctional. A comparison is made between emergent computational states and emotional states. This paper weaves these threads together in a consideration of the nature of sophisticated dynamic representations (e.g. goals and motivations), emotion and philosophical issues related to the modeling of cognitive agent systems and the relevance in future computational systems responsible for complex human-machine interactions.