Evaluating multimodal affective fusion using physiological signals

In this paper we present an evaluation of an affective multimodal fusion approach utilizing dimensional representations of emotion. The evaluation uses physiological signals as a reference measure of users' emotional states. Surface electromyography (EMG) and galvanic skin response (GSR) signals are known to be correlated with specific dimensions of emotion (Pleasure and Arousal) and are compared here to real time continuous values of these dimensions obtained from affective multimodal fusion. The results (both qualitative and quantitative) suggest that the particular multimodal fusion approach described is consistent with physiological indicators of emotion, constituting a first positive evaluation of the approach.