Neural characteristics of emotion regulation and derived machine learning classification in high negative affectivity: Based on event-related potentials and nonlinear analysis.

As a core component of negative affectivity, high trait anxiety (HTA) elevates the risk of the onset of depressive disorders through impaired emotion regulation processes. However, its underlying neural mechanisms remain unclear due to methodological constraints. This study addressed these gaps by integrating event-related potential (ERP), nonlinear analysis, and machine learning (ML), while controlling for emotion regulation strategy use habits. We recruited 33 college students with HTA and 30 with low trait anxiety (LTA), who performed cognitive reappraisal and expressive suppression during emotion regulation. ERP results showed that both groups reduced P2 amplitudes for negative stimuli via the two regulation strategies. However, HTA individuals only downregulated centro-parietal LPP in the early stage (500-1000 ms) and failed to maintain this effect in the late stage, indicating impaired sustained emotion regulation. Nonlinear and correlation analyses revealed that lower wavelet entropy (reflecting less flexible brain states) was associated with poorer sustained reappraisal in HTA individuals, implying that reduced neural complexity may underlie deficits in prolonged regulation. Group differences vanished after controlling for depressive symptoms, implying such neural abnormalities stem from a general negative affectivity predisposition rather than trait anxiety alone. ML further identified reappraisal-related LPP and wavelet entropy modulation as the most discriminative features for this negative affectivity phenotype. These results provide a theoretical foundation for translational research and may aid in the early identification of individuals at high risk for such disorders.