Multi-componential Emotion Recognition in VR Using Physiological Signals

Abstract

AbstractEmotion recognition affords new approaches ranging from context-awareness to the efficiency of system interaction with the ability to perceive and express emotions. While most studies are dominated by discrete and dimensional theoretical models of emotion, neuroscience analysis aligns with the multi-component interpretation of emotional phenomena. One such componential theory is the Component Process Model (CPM), with five synchronized components: appraisal, motivation, physiology, expression and feeling. However, limited attention has been paid to the systematic investigation of emotions assuming a full CPM. Therefore, we induced various emotions in this preliminary analysis using 27 interactive Virtual Reality (VR) games. We measured the manifestation of 28 participants across CPM components, 20 discrete emotion terms, heart activity, skin conductance, and facial electromyography. Our work aims to analyze the relationship between discrete theory-based emotions and the theoretically defined components with physiological measures. Further, we analyze the correlation between subjective expression terms with objective facial expressions. Our Machine Learning (ML) analysis reveals a significant relationship between emotions and full componential features with physiological signals. Further, our study presents the role of each CPM component in emotion differentiation.KeywordsEmotionComponent process modelPhysiological responsesComputational modeling