AIMS: This study for the first time proposed a novel prefrontal internal event-driven analytic framework for electroencepalography (EEG) data, which aim to dynamically resolve neural processes during natural emotional auditory tasks. METHODS: The framework employed a novel unsupervised time-series clustering model for internal prefrontal event extraction, which supports event-related analyses with the absence of external event labeling. The framework was validated using a 64-channel EEG data obtained from 110 (55 depressed) subjects in a three-polar (positive, neutral, and negative) emotional-auditory task. RESULTS: Our results suggest that anhedonia in depressed patients are associated with high activation levels in multiple brain regions during specific internal events, and we found that cross-frequency modulation of the bilateral prefrontal lobe with other relevant regions revealed completely different unidirectional patterns for the positive and negative tasks. CONCLUSION: Our study confirmed the effectiveness of the framework in resolving fine-grained internal event-driven neural processes without relying on traditional precise event-related data acquisision paradigms that often require high attention on the task events and causes high cognitive load. Our study present new insights for identifying dynamical electroencephalographic biomarkers in depression, which potentially provide EEG signal decoding solutions for EEG feedback-based closed-loop intervention of depression.