EEG Spectral Analysis in Valence and Arousal Dimensions of Emotion

This paper presents the initial studies of spectral analysis of the electroencephalogram signals for emotion recognition. Two-dimensional emotion model of valence and arousal is used to represent emotion, which is advantageous in quantifying the emotion. Ninety photographs, categorized into high valence-high arousal, medium valence-low arousal, and low valence-high arousal groups, were chosen from the International Affective Picture System (IAPS) for the visual stimuli, and the electroencephalogram was recorded from 12 electrode positions referenced to Cz and grounded to Fz. In addition to the 12 channel data, twenty virtual channels data were generated using the transverse and the longitudinal bipolar montage. Ten experiments with one male mature adult participant under the three categorized visual stimuli were performed. The results demonstrated the associations between emotional experience and four spectral band powers of theta, alpha, beta, and gamma at 32 channels, in order to find the combinations of the band power and the channel that are effective in discerning the valence and the arousal levels of emotion.