, Azuraliza Abu Bakar

The framework of the Affective Decision Making Engine outlined here provides a blueprint for creating software agents that emulate psychological affect when making decisions in complex and dynamic problem environments. The influence of affect on the agent’s decisions is mimicked by measuring the correlation of feature values, possessed by objects and/or events in the environment, against the outcome of goals that are set for measuring the agent’s overall performance. The use of correlation in the Affective Decision Making Engine provides a statistical justification for preference when prioritizing goals, particularly when it is not possible to realize all agent goals. The simplification of the agent algorithm retains the function of affect for summarizing feature-rich dynamic environments during decision making