Modulating Aesthetic Furniture Preferences through Music: A Multimodal Neurophysiological Study Using Neural Ordinary Differential Equations Networks

Cross-modal perceptual integration plays a crucial role in aesthetic cognition, yet the neural regulatory mechanisms underlying music's influence on visual design preferences remain insufficiently elucidated. This study aims to explore how musical environments modulate aesthetic evaluation of traditional furniture through neural plasticity mechanisms and establish a preference prediction model based on multimodal neurophysiological signals. Four core aesthetic dimensions—"tactile density, lubricity, minimalist, and exquisite"—were extracted through sensory engineering methodology. We collected 32-channel electroencephalography (EEG), eye-tracking trajectories, and subjective aesthetic ratings from participants viewing 238 traditional square stool images under conditions with and without natural sound effects. A novel NeuralODE-Transformer multimodal fusion architecture was innovatively constructed: neural ordinary differential equations were employed for EEG temporal dynamics modeling, combined with Transformer attention mechanisms to achieve cross-modal feature fusion, with aesthetic preferences predicted through regularized regression.The findings revealed that musical intervention significantly modulated neural activity in specific frequency bands: prediction error for tactile density in the Alpha band decreased by 26.9%, while prediction accuracy for exquisiteness in the Beta band improved significantly. The NeuralODE-Transformer model demonstrated superior predictive performance across all four aesthetic dimensions compared to conventional methods. Design scheme validation through AHP-entropy weight method confirmed alignment with the proposed NeuralODE-Transformer prediction model. This study represents the first quantification of cross-sensory regulatory effects of music on furniture visual aesthetics through multimodal neurophysiological signals, establishing a complete chain from neural mechanisms to design applications. It provides novel technical insights for multimodal aesthetic cognition modeling.