EEG-Based Emotion Analysis: Characterizing Emotion-Related Modulations of Brain Rhythms

This thesis examined whether different emotional conditions are associated with distinct modulations of electroencephalographic (EEG) spectral power and hemispheric asymmetry, and whether these modulations change over the course of emotional stimulation. EEG (128 chans) were acquired in 9 participants while viewing validated videos (lasting 2-min) taken from the eMOVIE database and belonging to six categories: Fear, Compassion, Erotic, Sadness, Scenery, and Neutral. Subjective ratings via the Self-Assessment Manikin (SAM) generally reflected the intended affective content of the stimuli. After EEG preprocessing, Power Spectral Density was estimated via Welch’s method across distinct temporal windows: baseline (before video presentation), First Minute, and Last (i.e. second) Minute. Power in each frequency band (delta, theta, alpha, beta, gamma) was computed for each category and each time window. Emotion conditions (Compassion, Erotic, Fear, Sadness, Scenery) were compared with Neutral across the scalp, at each band and time window, using non-parametric cluster-based permutation tests. Hemispheric asymmetry was assessed by comparing homologous right and left electrodes. Alpha power decreased during Emotion relative to Neutral, with broader effects for arousing stimuli (Erotic, Compassion, Fear), and more localized posterior effect for visually-driven states (Scenery). Theta power was higher during Neutral than Emotion, possibly shifting from working-memory demands of documentary narration (Neutral) toward viscerally driven affective perception (Emotion). The Erotic condition showed a broad and sustained reduction in Beta power relative to Neutral, possibly indicating greater involvement of sensorimotor-related cortical processes. No emotion-specific hemispheric asymmetry emerged. Future research should extend sample size, leverage acquired physiological signals (GSR, PPG) for a joint analysis, and apply effective connectivity to map emotion regulation networks.