Music-Enhanced Brain Connectivity and Memory Performance: Comparative Analysis of Traditional and Digital Learning Mediums Using EEG

Extensive research has demonstrated that music enhances cognitive functions, such as learning, reasoning, and memory. This study investigated the influence of music on short- and long-term memory while learning through traditional and digital mediums in healthy adult participants. Twenty-seven participants engaged in memory tasks under both learning conditions while electroencephalogram (EEG) data were collected. Functional and effective brain connectivity features in both the time and frequency domains were extracted and analyzed using SHAP-based feature interpretation. A hybrid ensemble learning architecture consisting of two ProtoNet-MLP (prototype-based embedding) models and a calibrated SVM as base classifiers, with logistic regression as the meta classifier was employed. The proposed framework integrates FDR-derived (Flow Degree Ratio) node statistics and off-diagonal edge-level connectivity representation as complementary feature inputs. These feature sets were learned independently using two prototype-based embedding learning, enabling the model to capture both node and edge-level characteristics within the EEG connectivity. Stronger significant theta band connectivity was observed during memory tasks with music particularly under long-term memory conditions, suggesting a potential association between music exposure and memory consolidation. Long-term memory exhibited a significantly broader network involving frontal, temporal, parietal, and occipital regions compared with short-term memory. In addition, differences in connectivity organization were observed between learning modalities, where traditional learning showed relatively synchronized connectivity patterns while digital learning exhibited more complex unsynchronized interaction patterns. The proposed model achieved mean accuracies of 65.17% and 59.5% under subject-independent protocol for digital and traditional learning conditions, whereas 69.06% and 63.5% under within-subject protocol, reflecting inter-subject variability impact on performance. These findings suggest that music may influence memory-related brain connectivity and provide insights into the brain connectivity patterns across different learning mediums.