Abstract
Green walls are a common biophilic design element in indoor environments, contributing to the improvement of individuals' psychophysiological health. This study, utilizing virtual reality technology, constructed three different types of green walls: cool-colored flower, warm-colored flower, and non-flower combined green walls, with no green wall serving as the control. The visual preference vote (VPV), subjective evaluations, and electroencephalogram (EEG) of 26 young adults were measured to investigate how varying levels of preference for green walls influence restoration. The study found that green walls reduced psychophysiological stress levels; however, significant individual differences were observed in visual preferences. High-preference green walls were associated with more positive emotional responses and more stable patterns of brain activity. Compared to medium- and low-preference conditions, the changes in brain oscillatory power were 1.39-2.96 times greater, and the effective delay time of alpha rhythms was 1.49-1.68 times longer, suggesting enhanced neural stability. Exposure to high-preference green walls induced smaller and faster neural avalanche activities. The avalanche criticality index (ACI), an indicator of how close brain activity is to a critical and balanced state, decreased by up to 30.31%, reflecting enhanced stability and comfort of neural dynamics. VPV was closely related to psychophysiological indicators (p < 0.05). A prediction model of green wall preference was constructed based on four EEG features, with the random forest classifier achieving an accuracy of 0.88. Among these, ΔACI was the most important predictor of VPV (weight: 0.48). This study provides a method for predicting individual preferences for green walls, offering strong evidence for indoor green wall design.