Affective Computing

As the importance of emotion to rationality is becoming clearer in the realms of cognitive science and philosophy (see [12,5,13,3]), it is also of growing interest to practitioners in the field of artificial intelligence. Picard's book, Affective Computing [14], is a good example of the increase in interest in emotional computing. In experiments with humans whose brains are damaged, it has been shown that a lack of emotion damages decision making capability [4]. Picard notes that these people's reasoning malfunctions are eerily similar to computer malfunctions.

Picard splits emotional computing into three areas: emotion recognition, emotion expression and computers that actually `have' emotion [14, page 83]. Most computer scientists have concentrated on creating agents which act emotional and to some extent understand emotion. Unfortunately, there has been little work on creating artificial agents whose processing resembles emotional theories, ie. agents which actually are emotional. For example, Reilly [7] focussed on creating believable agents, freely sacrificing accuracy of emotion for simplicity and artistic applicability (see also [15]). While related, research into creating believable emotion agents is not the focus of this thesis.