Dynamic neural trajectories of emotion bias processing in treatment-resistant depression: A novel perspective and computational approach.

Emotional bias processing may serve as a core underpinning of the onset and persistence of treatment-resistant depression (TRD). However, its neural mechanisms remain unclear. A comprehensive understanding of the mechanisms requires elucidation of the computational processes involved. We designed a double-blind, longitudinal, placebo-controlled trial in which 44 participants were randomly assigned to an intervention group or a control group. From a novel perspective, we employed an electroencephalography (EEG)-based multivariate pattern analysis (MVPA) framework, integrated with an emotion recognition task, to reveal for the first time the temporal trajectory of emotional bias processing in TRD. To enhance the robustness of MVPA analyses at individual level, we introduced temporal continuity constraints within the framework, effectively reducing false-positive interference. The results revealed significant temporal dynamics in emotional processing in TRD. Fluctuations in the decoding curve reveal that emotional bias processing involves dynamic processes spanning multiple cognitive stages rather than being confined to a specific phase. Following emotional stimulus onset, the stability of neural coding progressively increases. Remitted TRD patients exhibited longer decoding sustainability and an earlier reduction in neural coding stability. Further analyses showed that temporal parameters of emotional processing decoding demonstrated significant regression effects and were closely associated with depression severity. This study enhances the understanding of emotional bias processing mechanisms in TRD and uncovers novel temporal dynamics beyond event-related potential analyses, highlighting their potential as promising biomarkers for depression assessment, while also demonstrating the utility of MVPA in investigating emotional processing abnormalities.