The relationship between animated and realistic facial geometry in eliciting audience emotions

Facial features significantly impact the emotions elicited in perceivers. Different races possess unique facial geometric characteristics, which may influence how animated or realistic faces evoke emotional responses. However, it remains unclear to what extent facial geometric features regulate perceivers' emotional experiences, such as valence and arousal. The current study explores how animated and realistic facial geometry elicits self-reported emotional responses (valence and arousal) from perceivers. This study collected four types of facial stimuli, including real and animated faces from the U.S. and Japan, with geometric measurements derived from 41 facial geometric features defined by 18 lengths, 17 angles, and 6 areas within 26 points around the frontal face. Furthermore, after viewing the facial stimuli, 60 participants used a five-point Likert scale to assess their subjective emotional valence and arousal levels. Our results show that ANOVAs (2 races × 2 face types) for each measure (valence and arousal) revealed significant differences (p < 0.001). A multiple regression model revealed a significant relationship between facial category geometric features and valence/arousal, with about 28% of geometric features in American animations having the greatest impact on valence predictions. Following this, about 24% of predictions in Japanese animations. Additionally, approximately 8.8% of geometric features in Japanese animations had the greatest impact on arousal predictions, followed by about 7.6% in real Japanese faces. Therefore, we conclude that facial geometric features in U.S. and Japanese realistic and animated faces significantly influence the emotions elicited in perceivers, particularly with regard to valence. We also discuss the impact of feature similarity between animated and realistic faces on emotional regulation.