Abstract
Understanding the neurophysiological mechanisms underlying driving behavior in young drivers is essential for improving cognitive-aware driver assistance and vehicle-human interaction systems. This study systematically examines EEG dynamics and functional brain network reconfigurations across both manual and video-based car-following observation, providing a neurophysiological framework for differentiating driving modes among young adult drivers. EEG characteristics were analyzed under three car-following strategies-aggressive, conservative, and personalized-implemented within a simulated driving environment, to capture the variability of cognitive engagement during distinct control demands. Key findings reveal that power spectral density (PSD) in the θ, β, and γ bands, combined with brain functional connectivity (BFN) measures, effectively characterizes workload-related modulation and attentional resources across driving conditions. A novel computational framework integrating Time-Frequency Common Mutual Information (TFCMI) features with a Parallel Compact Convolutional Neural Network (PCNet) achieved an average classification accuracy of 85.26%, surpassing traditional single-modality approaches. Neurotopographic results further indicate context-dependent functional specialization: frontal regions showed stronger activation and connectivity during manual control, while occipital regions exhibited enhanced synchronization during video-based car-following observation tasks. Collectively, these findings advance the understanding of driving-related cognitive processes in young drivers and provide neuroergonomic insights for designing adaptive human-machine interfaces in future intelligent transportation systems. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s11571-026-10442-2.