Penny Baillie and Dickson Lukose (Univ. of Southern Queensland)
Interacting intelligently within a temporally dynamic environment calls for adaptive performances from artificial beings. One mechanism currently being explored to produce an intuitive-like behaviour in AI applications is that of emotion. Several popular psychological theories of emotion have been the basis for a number of agent models that are capable of synthesizing emotions. However, while these models have demonstrated limited success, they lack the dynamic scalability to form complex emotional dimensions that are necessary for producing adaptive behaviour. This paper examines how an affective appraisal made with discrete evaluations of an event can overcomes contemporary limitations. An artificial agent that has had this theory integrated into its architecture is demonstrated. A formal model of the agent's Ontology and its Multidimensional Affective Decision Making process is outlined, with examples demonstrating an affective decision making process.