Integrating Electroencephalography (EEG) and Machine Learning to Reveal Nonlinear Effects of Streetscape Features on Perception in Traditional Villages

Public perception of traditional villages’ streetscape is a crucial link for unlocking their benefits in promoting physical and mental health and realizing environmental value transformation. Current studies on the influence mechanisms of rural streetscape characteristics on perception largely rely on subjective ratings and mostly depend on linear models. To address this, this study takes a traditional village in eastern China, which is rich in natural and cultural conditions, as an example and constructs an evaluation framework comprising 29 streetscape feature indicators. Based on multimodal data including electroencephalography (EEG), image segmentation, color, and spatial depth computation, XGBoost-SHAP was employed to reveal the nonlinear influence mechanisms of streetscape features on neurophysiological indicators (alpha-band power spectral density, α PSD) in the traditional rural context, which differs from the blue–green spaces and residential, campus, and urban environments in previous studies. The results indicate that (1) the dominant factors affecting α PSD in traditional villages are tree, color consistency, architectural aesthetics, spatial enclosure index, P_EBG, and road, in descending order. (2) Threshold effects and interaction effects that differ from previous studies on campuses, window views, and other contexts were identified. The positive effect of tree view index on α activity peaks at the threshold of 0.09, beyond which diminishing returns occur. Color complexity, including high color difference from the primary village scheme (i.e., low color consistency, color diversity, and visual entropy), inhibits α activity. The effect of spatial enclosure index (SEI) on α activity exhibits an inverted U-shape, peaking at 0.35. Tree–VE_nats, road–SEI, and building–SEI show antagonistic effects. Road–sky and SEI–P_FG display conditional interaction effects. (3) Based on k-means clustering analysis, the “key factor identification—threshold effect management—multi-factor synergy optimization” design can directionally regulate α PSD, promoting relaxed and calm streetscape schemes. This approach can be applied to urban and rural environment assessment and design, providing theoretical and technical support for scientific decision-making.