Compound facial expressions of emotion.
Level 4 - case-series / case-control
Cross-sectional observational laboratory study with computational modeling in healthy participants
PubMed 24706770 · doi:10.1073/pnas.1322355111
What was done
Defined 21 distinct emotion categories by combining basic emotion categories (e.g., happily surprised, angrily surprised). Sample facial expression images were collected from 230 human participants. The authors performed Facial Action Coding System (FACS) analysis to evaluate facial muscle movements and applied a computational model of face perception to assess visual discriminability among the categories.
What was found
The abstract reports no numerical values, classification accuracies, or statistical test metrics. FACS analysis showed that the muscle movements used to produce the 21 compound categories were distinct from one another but consistent with their subordinate basic categories. The muscle movement differences were sufficient to distinguish all 21 categories, and computational modeling demonstrated that most categories were visually discriminable.
Why it matters
This work expands facial expression research beyond the traditional six basic categories, providing a broader framework for affective neuroscience, cognitive psychology, and computer vision interfaces.
Limits
The abstract does not provide quantitative data, classification accuracy percentages, or statistical confidence bounds. Details regarding participant demographics, cultural background, whether expressions were posed or spontaneous, and real-world human-to-human recognition accuracy are not reported.