Resting-State Electroencephalogram and Heart Rate Variability Jointly Predicts Intentional Episodic Memory Control Using Machine Learning Fusion

Deficits in intentional control over episodic memory constitute a risk factor for multiple psychiatric disorders. Guided by a body-brain dynamic systems perspective, we reasoned that intrinsic cortical dynamics and autonomic regulatory capacity may jointly shape individual differences in memory control. Accordingly, we tested whether resting-state electroencephalogram (EEG) and heart rate variability (HRV) each predict intentional episodic memory control and whether adding HRV-derived autonomic signatures can enhance EEG-based discrimination of individual differences in memory control. Forty-two healthy adults were recruited and completed an in-house laboratory 5-min eyes-closed resting-state recording with concurrent EEG and electrocardiogram (ECG), followed by an item-method directed forgetting task indexing memory control. Machine learning algorithms were separately applied to multiple EEG feature sets and on HRV features to predict memory control ability. To test the incremental value of autonomic information, we constructed EEG-HRV fusion models using weighted score-level fusion. Results demonstrated that all unimodal models (EEG and HRV) performed above chance. Crucially, five of six fusion models outperformed the corresponding EEG model. These findings indicate that resting-state EEG and HRV each contain predictive information about intentional episodic memory control, and that incorporating HRV can enhance EEG-based prediction, supporting the value of combining central and autonomic physiology when modeling individual differences in intentional episodic memory control.