Emotional interactive experiences in different environments and cultures: machine learning-based electrocardiogram analysis of the emotional elicitation effect in augmented reality
Abstract
Emotions are dynamic feedback mechanisms by which individuals respond to external stimuli, and the environment profoundly influences their changes. Therefore, it is necessary to incorporate healing and inclusive considerations into public environmental design. This study combines machine learning (backpropagation neural network) and augmented reality (AR) to evaluate the emotional interactive experience in different environments (natural forest, landscape garden, public square, and community street) through electrocardiogram signals (ECG) and psychological scales (I-PANAS-SF). Residents from different cultural backgrounds and experiencing psychological discomfort (N = 160) were invited to explore the intervention and maintenance effects of emotional elicitation materials (AR interactive installations). The overall precision of the BP neural network was 82.4%, high-arousal emotions (including happiness and anger) were more frequently misjudged, whereas low-arousal negative emotions (sadness) had a high precision rate. However, low-arousal positive emotions (peacefulness) were more easily confused with mild positive or negative emotions due to their blurred boundaries, resulting in low precision. The results of I-PANAS-SF indicate that AR interactive installations can significantly enhance positive emotions in short periods, but have limited effects on alleviating negative emotions. Landscape gardens have the most profound emotional awakening effect, followed by natural forests and public squares, which exhibit similar effects but are more effective than community streets. The AR evaluation matrix reveals participants' preferences for emotional resonance elements across different cultural backgrounds. These findings can provide empirical evidence and strategic guidance for promoting healing and inclusiveness in future public environmental design, thereby enhancing community emotional well-being.
// Source
Authors: Tanhao Gao, Mengshi Yang, Phillip Bernstein, Hongtao Zhou
Institutions: Yale University, Tongji University, University of Macau, Xiamen University