The Evaluation of the Effects of Music Characteristics on Driving Performance and Physiological Arousal in a Simulator Environment

This study quantifies the effects of the structural components of in-cabin music on driving performance and physiological arousal in a within-subject design with 50 licensed drivers. Fifteen music scenarios, derived from a fractional factorial plan manipulating three structural variables—emotion conveyed by the music, tempo, and playback volume—were implemented in a driving simulator; scenario order was counterbalanced with 5-min inter-scenario breaks. The physiological index was recorded from a GSR sensor as skin resistance (SR) and mapped to a unified metric Arousal Index = −z(log SR); performance measures included driving errors, mean/variability of speed, lateral lane deviation, and reaction time. Music-attitude profiling via PCA confirmed data adequacy and, using k-means, yielded an optimal K = 5 attitude typology. Clustering on performance–physiology features revealed four stable driver response types (K = 4; Silhouette = 0.463): (1) low errors/low arousal, (2) medium errors/medium arousal, (3) high errors/high arousal, and (4) low errors/high arousal. Within-subject tests showed significant differences across musical conditions (Friedman for errors: χ² = 24.33, df = 2, p < 0.001; for SR/Arousal: χ² = 22.11, df = 2, p < 0.001); Wilcoxon pairwise comparisons with Holm correction indicated robust contrasts for happy ↔ anxiety-inducing and sad ↔ anxiety-inducing music across both domains (p ≤ 0.003), while happy ↔ sad differed only in SR/Arousal (p = 0.021). Findings indicate that music-induced arousal modulates both behavioral safety and physiological state, and that inter-individual heterogeneity is decomposable into stable response types.