Emotion recognition in virtual reality: EEG and EDA-based analysis of stress in high-risk scenarios

Abstract

Virtual Reality (VR) is a powerful tool for analyzing human emotions in simulated crisis scenarios. This study integrates electroencephalography (EEG) and electrodermal activity (EDA) in VR environments to improve emotion recognition in dynamic, high-stress situations that are difficult to replicate in real-world settings. The experiment involved a simulated elevator sabotage scenario, during which participants’ neurophysiological and physiological responses were recorded as the elevator descended. The collected data were compared with self-reported assessments. The primary objective was to evaluate whether combining VR and biosignal-based measurements enables objective analysis of emotional experiences. A hybrid emotion recognition system combining EEG and EDA was developed to quantify stress levels in dynamic crisis scenarios. Statistical analyses and SVM classification confirmed significant phase differences, with 91.67% accuracy and higher reliability of the combined EEG-EDA approach compared to standalone methods. The results indicate that this approach enhances the understanding of psychophysiological stress responses and provides a reliable tool for identifying emotional states during simulated security incidents. This system has potential applications in emotion research, security training, crisis simulations, and environmental psychology. These findings highlight the potential of VR and biosignal analysis for realistic crisis simulations, offering valuable insights for investigative methodologies and security personnel training.