Exploring the Impact of Color and Daytime/Nighttime on Visual Memory in Virtual Reality Teaching Environments Using Explainable Machine Learning

Virtual reality (VR) has proven to be effective in creating immersive learning experiences for both teachers and students. Notwithstanding the evident utility of VR teaching environments, there is a notable paucity of research investigating the impact of environmental factors on learning tasks. This study compares the effects of visual memory among learners in a VR classroom environment with different environments. A visual memory experiment was conducted with 46 participants, in which the influence of color and daytime/nighttime on memory was investigated. The results demonstrated that learners exhibited significantly enhanced visual memory reaction time under daytime environments compared to nighttime, particularly with yellow color ( t = 2.784, P = 0.011) backgrounds. The average reaction time reaches 1.223 s, and the accuracy reaches 98.4%. This highlights the pivotal role of color in optimizing memory effects. In contrast, the impact of color on memory tasks is comparatively diminished in nighttime, with no significant difference in learners’ memory capabilities observed under different color temperatures. Furthermore, we employ machine learning techniques to predict the average response time based on facial features. Results show that the XGBoost model has the best performance (mean absolute error: 0.14, mean squared error: 0.06, R 2 : 0.05). An explainable machine learning model, the SHapley Additive exPlanations (SHAP) algorithm, was employed to identify the importance and influence direction of each feature. The results demonstrated that lip movement and eye fixation direction exhibited the most indicative effect on reaction time. In light of these findings, recommendations are provided for enhancing the VR teaching environment and corresponding indicators for detecting learner focus.