[Emotion detection by motion analysis: a comparison of human and machine-based processing]

Introduction: The relationship between human emotions and their bodily expressions has been receiving increasing attention in psychology and cognitive sciences. Emotions are not only expressed through verbal and facial cues but also through posture, gestures, and movement patterns. Motion analysis is one of the possible methods for identifying emotions; however, there is currently no clear consensus on which method is the most suitable for this purpose. Objective: The aim of this study is to systematically review motion analysis methods used for emotion recognition and to compare human-based and artificial intelligence-based approaches. Method: This study follows a systematic literature review using the PRISMA protocol. The initial search identified 7699 scientific articles from various databases. After applying inclusion and exclusion criteria, 16 relevant studies were selected for detailed analysis. The research examines different motion analysis techniques, their applicability, and their accuracy. Results: A total of 9 different motion analysis methods were identified in the literature, of which 4 utilized artificial intelligence for emotion recognition. These methods differ in terms of data processing techniques, applied technologies, and accuracy in detecting emotions. Discussion: Motion analysis is a promising tool for studying emotions; however, no universally applicable technique currently exists. Artificial intelligence plays an increasing role in emotion recognition, yet challenges remain in algorithm development and handling individual differences in emotional expression. Conclusion: Currently, there is no single, universally accepted motion analysis technique for emotion identification. Artificial intelligence can be a useful tool, but it has limitations when used alone. The most reliable approach appears to be the application of multidisciplinary methods, which combine human observation, classical motion analysis, and machine learning techniques. Orv Hetil. 2025; 166(46): 1803–1809.