Related Experiment Videos
Detecting hemifacial asymmetries in emotional expression with three-dimensional computerized image analysis
Michael E R Nicholls1, Brooke E Ellis, John G Clement
1Department of Psychology, University of Melbourne, Parkville, VIC 3010, Australia. m.nicholls@psych.unimelb.edu.au
Proceedings. Biological Sciences
|June 24, 2004
Summary
Facial emotion expression shows left-sided asymmetry, supporting right hemisphere dominance. However, observer ratings indicate left hemifaces convey negative emotions and right hemifaces convey positive emotions, supporting the valence model.
Area of Science:
- Neuroscience
- Psychology
- Human Behavior
Background:
- Facial asymmetry in emotion expression is often attributed to right hemisphere dominance for emotional processing.
- The valence model proposes that this asymmetry differs for positive and negative emotions.
Purpose of the Study:
- To investigate the right hemisphere and valence models of facial emotional expression.
- To measure facial movement and observer ratings of emotionality to assess hemispheric asymmetries.
Main Methods:
- Utilized a 3D imaging technique to capture and manipulate digital facial models (n=16) with neutral, happy, and sad expressions.
- Measured asymmetries in facial movement and analyzed observer ratings (n=357) of hemiface portraits.
Main Results:
- Facial movement analysis supported the right hemisphere model, showing greater left hemiface movement for all emotions.
- Observer ratings supported the valence model, with left hemifaces rated more expressive for negative emotions and right hemifaces for positive emotions.
- The valence effect was moderated by image manipulation (mirror-reversal).
Conclusions:
- The study provides evidence for both right hemisphere dominance and valence-specific asymmetries in facial emotion expression.
- Subtle head rotations significantly impact the perceived valence of emotional expressions.
- Discrepancies between movement and perception highlight the complexity of facial emotion signaling.