An Efficient EEG Microstate Analysis Method for Emotion Study

Emotion plays an essential role in human health and daily life. Estimating emotions in a brain-level dynamic approach helps to understand the underlying neural mechanism, deepen emotion interpretation, and boost the development of affective computing technology for practical application. EEG microstate analysis is a powerful neurophysiological tool for dynamic EEG characterization, covering both temporal and spatial information of brain activities. In this paper, EEG microstate analysis is introduced for the dynamic analysis of video-evoked emotions. A sequential clustering process is proposed for validated and representative microstates detection for emotion-related EEG dynamics characterization, and the underlying neural activation patterns under different emotion states are explored. A study of emotion-related electrophysiological mechanisms is conducted for investigating the emotional perception and processing in the brain responses. The results demonstrate that EEG microstates extracted from the proposed sequential clustering are discriminative for dynamic emotion analysis. Besides, the dynamically evoked emotions can be effectively described by the activation patterns of EEG microstates, where an increased activation of MS2 and MS4 but decrease activation of MS3 are found after emotion induction. Furthermore, distinct emotional-level effects for valence and arousal are observed, where MS4 activities are negatively associated with valence level, and MS3 activities are positively associated with arousal level. In all, our work validates the possibility of applying EEG microstate analysis for emotion-related neural mechanism investigation. It has also proved EEG microstate analysis is a powerful tool for exploring spatial-temporal brain changes through emotion perception.