Exploring neural reorganization and modifiable connectivity patterns in sadness: a multi-band EEG study

Negative emotions have been recognized to influence cognitive and emotional functioning, which could contribute to the pathogenesis of mental disorders. While prior research has elucidated broad neural correlates of emotion, the specific mechanisms underlying the induction and spontaneous regulation of them remain poorly characterized. In this study, we introduced an electroencephalographic (EEG) based network analysis to investigate sadness-related alterations in neural activities. Specifically, sadness emotion was firstly induced via audiovisual stimuli, followed by a natural recovery to assess spontaneous emotion regulation. Functional connectivity (FC) networks were constructed from multi-band EEG signals, which were quantitatively assessed across the baseline, sadness, and regulation states. Furthermore, a machine learning framework integrating feature selection and classification was applied to identify discriminative connectivity features associated with emotional transitions. Graph-theoretical analysis revealed significant emotion-related alterations in functional networks from α, β, and θ band. Additionally, the machine learning pipeline produced consistent results, which identified modifiable FC features predominantly in the occipitotemporal-prefrontal networks within α band. These findings provided quantitative evidence of frequency-specific network dynamics during sadness induction and regulation, offering promising targets for developing EEG-based biomarkers to assess and regulate emotional dysfunction in clinical applications.