Detecting emotion from motion : head movement as a predictor for emotion

Abstract

One of the possible visual cues for emotional recognition is head movement, because it is assumed that the emotion is organized by two motivational system: appetitive and defensive system and drives the behavior. This study examines the relationship between the head movement and emotional experience. We recruited forty participants (20 males, 20 females) and asked then to watch the pictorial stimuli with the online experimental platform. Their head movements index were captured through the web camera of their laptops. Results indicated approach behavior and behavioral activation performed in pleasant stimuli, and avoidance and behavioral freezing in unpleasant stimuli. We tested the feasibility of predicting emotional valence with the head movement data with four machine learning algorithms. We found that head movement could be the predictor of emotional valence slowly and Support Vector Machine (SVM) with radial basis function kernel was the best classifier, with the horizontal head movement as the most importance features. To our best knowledge, it is one of the minority studies reveals the head movement convey sufficient information of emotional experience to be detected. It extends our knowledge on emotional recognition and informs the application of the methodology in psychological experiment and facilitates the development of affective analysis technology.