AABSS: Affective Agent-Based Social Simulation

This research thesis considers the role of affect in the domain of social simulation. Focus is given to primate-like social organisation and potential internalisation mechanisms. Autonomous (bottom-up) adaptation of hierarchical social organisation and interaction is argued as feasible, by the introduction of affective and simple normative mechanisms into an individual agent architecture. This includes procedures for individual acceptance, or adherence to, approved behaviours and sanctions when considered appropriate by the individual agent. Synthetic agents are addressed as autonomous software entities capable of managing primate-like social dominance in small-scale societies, consisting of a maximum of sixteen agents. The critique involves an explanation of fundamental agent-based social simulation (ABSS) concepts, a comparison of the relevant individual actionselection mechanisms; plus a demonstration and discussion of the potential of affective roles in an agent architecture. The diversity of social and emotional accounts within the design of ABSS systems is discussed by considering behavioural control strategies that may be not only reactive, but also simple deliberative agent operations. Other pertinent models are compared in terms of architectural design, agent behavioural features and the limitations involved in the attempt to integrate emotional concepts into a system designed to simulate social interaction and group organisation.