Modeling Expectations for Affective Agents

Expectations and emotions require pervasive control over cognitive processes of artificial agents. In this work we discuss an enhanced architecture for cognitive agents guiding and orienting reasoning processes according to different kinds of expectations and affective states. They are assumed to involve cognitive processes including deliberation and practical reasoning (where emotional biases are assumed to affect decision processes), as well as situated adaptation to contexts and environment dynamism (where external events are appraised as opportunities or determent and recruits additional resources to pro-actively respond). We further provide experiments in a complex scenario, showing improvements in agent pro-activeness, opportunism and anticipatory effectiveness when goal selection is influenced by affective biases.