Bidirectional regulatory effects of exercise on emotional eating in depression: an ERP-based narrative review

Depression is frequently accompanied by emotional eating (EE). However, existing research has largely reduced EE to excessive food intake, overlooking the two opposing phenotypes within depressive populations: emotional overeating (EOE) and emotional undereating (EUE). Furthermore, self-report measures are limited in capturing the dynamic neural processes triggered by negative emotions. To address these issues, this narrative review integrated event-related potential (ERP) evidence from studies on depression, EE, and exercise interventions. The neural characteristics of the two EE phenotypes were compared across multiple processing stages, including attentional allocation, conflict monitoring, inhibitory control, and reward processing. The synthesized evidence suggests that EOE may be characterized by exaggerated reward responsiveness coupled with deficient inhibitory control, whereas EUE is marked by attenuated reward processing accompanied by excessive conflict monitoring. To mitigate inferential risks from cross-population and cross-task integration, the reviewed evidence was hierarchically stratified according to inferential distance, and model conclusions were explicitly bounded. Based on this framework, we propose a bidirectional regulatory model in which exercise is hypothesized to modulate neural homeostasis in a baseline-dependent manner. Within this hypothesis-generating framework, exercise may primarily affect putative EOE-related profiles by supporting cognitive control processes and attenuating the motivational salience of food cues, whereas it may affect putative EUE-related profiles by potentially supporting reward responsiveness and attenuating excessive monitoring. These proposed pathways remain to be tested in phenotype-stratified ERP intervention studies. Future research should implement phenotype-stratified randomized controlled exercise trials combined with emotion-induced food-related ERP tasks to test the model predictions and advance personalized intervention strategies.