Correlation between brain activity and comfort at different illuminances based on electroencephalogram signals during reading

The impact of light comfort on user health and productivity is profound, and brain activity serves as a sensitive indicator of how light environments influence individuals. This study establishes the correlation between light comfort and brain activity by utilizing time-frequency domain features and nonlinear dynamic features extracted from electroencephalogram (EEG) signals during reading. The experiment was conducted at three illuminance levels (200 lx, 500 lx, and 1000 lx) and involved collecting EEG signals and subjective evaluations from 28 young college students. Analysis of key EEG features under varied illuminances included average power, power distribution across different brain regions and frequency bands, fuzzy entropy, and avalanche parameters. The results demonstrate that at 500 lx, significant activation occurred in oscillatory activity within the prefrontal lobe and the β band. This illuminance level was identified as the most comfortable, with the lowest power observed at 2.34 μv2. Additionally, the avalanche criticality index was lowest (1.68) at 500 lx, indicating a stable and healthy brain state. Light comfort was significantly correlated with the relative power of δ and θ in parietal lobe and β in occipital lobe. The multilayer perceptron (MLP) model, trained on EEG features, achieves an 83.3% accuracy in predicting light comfort, with a notable 12.7% improvement when using only the 10 significantly correlated features compared to the full set. This study provides a method for using EEG signals to predict user light comfort under different illuminances and offers insights for the design of personalized indoor lighting environments.