How balance, contour, symmetry, and complexity predict the perceived emotions in Chinese traditional music

Understanding how musical structure shapes emotion perception remains a central question in music psychology, yet the role of mid-level structural features in Chinese traditional music is underexplored. This study investigated how balance, contour, symmetry, and complexity (BCSC framework) predict perceived emotions in Chinese traditional music using both linear (Lasso-selected multiple regression) and nonlinear (random forest regression) modeling approaches. Analyses included dimensional emotions (valence, arousal) and seven discrete emotions (anger, gentleness, happiness, peacefulness, sadness, solemnness, transcendence). Linear regressions showed that BCSC features explained 35.1% of variance in valence and 37.1% in arousal. For valence, Total Asymmetry (β = 0.266, p < 0.001), Average Local First-Order Entropy (β = 0.862, p < 0.001), Third-Order Duration Entropy (β = 0.511, p = 0.022), and Event Density (β = −0.890, p < 0.001) were among the strongest predictors. For arousal, Duration Abruptness, Total Asymmetry, Average Local First-Order Entropy, and Third-Order Duration Entropy were significant positive predictors (all ps < 0.01), while Rhythmic Abruptness, Event Density, and First-Order Duration Entropy showed significant negative effects (all ps < 0.001). For discrete emotions, predictive power ranged from adjusted R² = 0.157 (peacefulness) to 0.386 (happiness), with emotion-specific patterns: e.g., Bi-Unbalance (β = 0.316, p < 0.001) predicted anger, while First-Order Duration Entropy had opposing effects for anger (β = −0.252, p < 0.001) and peacefulness (β = 0.403, p < 0.001). Nonlinear models confirmed feature importance patterns, highlighting the roles of temporal irregularity, melodic interval size, and structural imbalance. Overall, these findings demonstrate that mid-level structural features provide systematic elements that serve as emotional cues for perceived emotions in Chinese traditional music.