Summary: New research shows that our bodily movements can change how we interpret facial expressions. In virtual reality experiments, participants were more likely to judge a face as angry when they actively moved away from it than when the face moved away from them.
The findings reveal a bidirectional link between movement and emotion recognition: avoidance behavior increases the perceived threat in others’ expressions. These results have implications for designing more natural social interactions in VR, telepresence, and emotion-aware AI systems.
Key Facts:
- Behavior Shapes Perception: Actively avoiding a face made participants more likely to perceive anger, indicating that our actions influence how we recognize emotions.
- Bidirectional Relationship: The results support a reciprocal connection between bodily movement and emotional perception.
- Practical Applications: Insights can inform improvements in virtual communication, social VR, and emotionally intelligent interfaces.
Source: TUT (Toyohashi University of Technology)
Summary of the study
Researchers from the Cognitive Neurotechnology Unit and the Visual Perception and Cognition Laboratory at Toyohashi University of Technology investigated whether approach–avoidance behavior can influence how people recognize facial expressions. Using virtual reality, the team tested whether actively moving toward or away from a face, or passively being approached or avoided by a face, alters emotion judgments.
Published online on July 31, 2025 in the International Journal of Affective Engineering, the study used three psychophysical VR experiments with head-mounted displays and 3D face models. Faces were morphed across seven levels between emotional expressions (happy–angry in Experiments 1 and 3; happy–fearful in Experiment 2). Participants judged each face as “happy” or “angry” (or “happy” or “fearful”) while experiencing one of four approach–avoidance conditions:
- Active approach: participant moves toward the avatar.
- Active avoidance: participant moves away from the avatar.
- Passive approach: avatar approaches the participant.
- Passive avoidance: avatar moves away from the participant.
Across experiments, movement context influenced emotion recognition. In Experiment 1, participants were likelier to label faces as angry when they themselves moved away from the face versus when the avatar moved away. Experiment 2 found that approach generally biased judgments toward happiness, while avoidance favored fearful perceptions regardless of who initiated the movement. In Experiment 3, when both parties were physically close, faces were judged as angrier when the avatar approached rather than when the participant approached.
Taken together, the results suggest that approach–avoidance behavior modulates how we interpret facial expressions. The authors propose that this link may arise from unconscious learning embedded in biological instincts, where bodily actions and perceived social signals form an interactive loop.
Implications
The study highlights that our own movements can bias emotion perception, which matters for contexts where body movement is limited—such as video conferencing or remote interaction. Incorporating body-based cues in virtual environments or emotional AI could enhance the realism and accuracy of social signals, improving empathy and communication quality in digital settings.
Author comment
Yugo Kobayashi, first author and a doctoral student in the Department of Computer Science and Engineering, noted that face-to-face communication involving bodily action may support more natural recognition of expressions, a quality often constrained in current remote communication tools.
Next steps
Future work will explore which elements of approach–avoidance behavior drive these effects—motor intention, visual motion, proprioceptive feedback, or their combination—to better understand the mechanisms linking action and perception.
Funding:
This research was supported by JSPS KAKENHI (Grant Numbers JP21K21315, JP22K17987, JP20H05956, JP20H04273), the Nitto Foundation, and doctoral research funding from Toyohashi University of Technology (FY2024).
Key Questions Answered
A: Participants were more likely to perceive a face as angry when they actively moved away from it, compared with when the face moved away from them.
A: The study demonstrates that physical behavior—approaching or avoiding—directly shapes how we interpret others’ emotions, indicating a feedback loop between motion and perception.
A: Findings can inform the design of emotional AI, telepresence systems, and virtual environments by incorporating body-based perception cues to improve social realism and empathy.
About this social neuroscience research news
Author: Shino Okazaki
Source: TUT (Toyohashi University of Technology)
Contact: Shino Okazaki – TUT
Image: The image is credited to Neuroscience News
Original Research: Open access. “Facial Expression Recognition is Modulated by Approach-Avoidance Behavior” by Yugo Kobayashi et al., International Journal of Affective Engineering.
Abstract
Facial Expression Recognition is Modulated by Approach-Avoidance Behavior
Facial expression recognition influences approach–avoidance behaviors, but can these behaviors in turn alter facial expression recognition? To examine this reverse relationship, we conducted psychophysical experiments in virtual reality.
Participants judged static 3D faces morphed between happy and angry (Experiments 1 and 3) and between happy and fearful (Experiment 2). In different trials, participants either approached or avoided the face, or the face approached or avoided the participant.
Results indicated: in Experiment 1, faces were judged as angrier when participants avoided them rather than when they were avoided; in Experiment 2, approaching biased judgments toward happiness while avoiding biased them toward fear regardless of who moved; and in Experiment 3, when both parties were close, faces were judged as angrier when the avatar approached than when the participant approached.
These findings suggest that approach–avoidance behavior modulates facial expression recognition, likely reflecting learned, biologically rooted associations between bodily actions and perceived social signals.