Emotion Representation and Neural Synchrony: Decoding Valence and Arousal with Wearable EEG

Abstract

Emotions are dynamic experiences that unfold over time, yet most affective neuroscience studies have relied on static stimuli and laboratory-based EEG systems. This study examined whether emotional valence and arousal can be reliably decoded using a consumer-grade wearable EEG device in naturalistic contexts. Forty-three participants viewed video clips designed to elicit four core affect categories including high-arousal positive, low-arousal positive, high-arousal negative, and low-arousal negative, while EEG signals were continuously recorded. Multivariate analyses, including classification, multidimensional scaling (MDS), and intersubject correlation (ISC), were employed to assess affective representation and neural synchrony. Behavioral data demonstrated robust classification of both valence and arousal, whereas EEG data yielded consistent above-chance classification of valence but less stable decoding of arousal, particularly in within-participant analyses. MDS revealed that both behavioral and EEG responses were primarily organized along the valence dimension, with weaker separation along arousal. ISC analyses further indicated frequency- and region-specific neural synchrony, with stronger alignment in left and temporal electrodes, though overall ISC values were modest, likely reflecting the brief duration of stimuli. Taken together, these findings suggest that valence is more stably represented in both subjective and neural domains, whereas arousal may require time-resolved or longer-duration approaches for reliable decoding. This work demonstrates the feasibility and limitations of employing wearable EEG for theory-driven affective neuroscience, underscoring its potential for scalable and ecologically valid emotion research beyond laboratory settings.