CoChat

Professional actors demonstrate variability, not stereotypical expressions, when portraying emotional states in photographs

Open access

The DOI matches its Crossref record; 93 of the 93 references it cites were found in a registry or an index.

Abstract

It is long hypothesized that there is a reliable, specific mapping between certain emotional states and the facial movements that express those states. This hypothesis is often tested by asking untrained participants to pose the facial movements they believe they use to express emotions during generic scenarios. Here, we test this hypothesis using, as stimuli, photographs of facial configurations posed by professional actors in response to contextually-rich scenarios. The scenarios portrayed in the photographs were rated by a convenience sample of participants for the extent to which they evoked an instance of 13 emotion categories, and actors’ facial poses were coded for their specific movements. Both unsupervised and supervised machine learning find that in these photographs, the actors portrayed emotional states with variable facial configurations; instances of only three emotion categories (fear, happiness, and surprise) were portrayed with moderate reliability and specificity. The photographs were separately rated by another sample of participants for the extent to which they portrayed an instance of the 13 emotion categories; they were rated when presented alone and when presented with their associated scenarios, revealing that emotion inferences by participants also vary in a context-sensitive manner. Together, these findings suggest that facial movements and perceptions of emotion vary by situation and transcend stereotypes of emotional expressions. Future research may build on these findings by incorporating dynamic stimuli rather than photographs and studying a broader range of cultural contexts.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-021-25352-6

Cites

93 of 93 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Examples of hypothesized facial configurations for emotion categories, based on the facial action units (AUs) listed in Supplementary Table 1. All facial configurations except “awe” and “contempt” based on those depicted in ref. 12; “awe” configuration based on ref. 9; “contempt” configuration based on ref. 96. Facial configurations were created in FaceGen (Singular Inversions, Inc., Toronto, ON, CA), with AUs set to 100% intensity, except where this would yield visible distortions—in those cases, AUs were set to 80% intensity.
Fig. 1 Examples of hypothesized facial configurations for emotion categories, based on the facial action units (AUs) listed in Supplementary Table 1. All facial configurations except “awe” and “contempt” based on those depicted in ref. 12; “awe” configuration based on ref. 9; “contempt” configuration based on ref. 96. Facial configurations were created in FaceGen (Singular Inversions, Inc., Toronto, ON, CA), with AUs set to 100% intensity, except where this would yield visible distortions—in those cases, AUs were set to 80% intensity.CC BY · PubMed Central
Fig. 2 Facial poses for inductive cluster 68, identified with unsupervised clustering of Sample 1 participants’ scenario ratings, with a model-selection procedure based on highest intra-cluster reliability in facial pose action units (AUs). Poses are presented along with their constituent AUs; consistent AUs for this cluster were 6, 7, and 12. These AUs occur in hypothesized facial configurations associated with happiness and amusement (Supplementary Table 1). However, the above facial poses depict configurations that were identified in photographs of actors portraying scenarios that evoked a variety of emotion categories. The associated emotion categories for the scenarios in this cluster (based on participants’ ratings) were happiness, interest, pride, and surprise. Facial configurations were created in FaceGen (Singular Inversions, Inc., Toronto, ON, CA), with AUs set to 100% intensity, except where this would yield visible distortions—in those cases, AUs were set to 80% intensity. Source data are provided in Supplementary Table 3.
Fig. 2 Facial poses for inductive cluster 68, identified with unsupervised clustering of Sample 1 participants’ scenario ratings, with a model-selection procedure based on highest intra-cluster reliability in facial pose action units (AUs). Poses are presented along with their constituent AUs; consistent AUs for this cluster were 6, 7, and 12. These AUs occur in hypothesized facial configurations associated with happiness and amusement (Supplementary Table 1). However, the above facial poses depict configurations that were identified in photographs of actors portraying scenarios that evoked a variety of emotion categories. The associated emotion categories for the scenarios in this cluster (based on participants’ ratings) were happiness, interest, pride, and surprise. Facial configurations were created in FaceGen (Singular Inversions, Inc., Toronto, ON, CA), with AUs set to 100% intensity, except where this would yield visible distortions—in those cases, AUs were set to 80% intensity. Source data are provided in Supplementary Table 3.CC BY · PubMed Central
Fig. 3 Reliability and specificity of hypothesized facial configurations in each emotion category. a Emotion categories for which all action units (AUs) were available to be coded. b Emotion categories for which AUs were simulated, given that dynamic AUs could not be coded from photographs. Reliability: The gray box-and-whisker plots show the distribution of match scores between the facial poses and hypothesized facial configurations for each emotion category. Each plot presents the maximum and minimum values as whiskers, the interquartile range as the vertical length of the box, and the median as the horizontal line within the box. The dotted horizontal lines denote the degree of reliability, based on criteria from prior research35: none (0 < 0.2), weak (0.2 < 0.4), moderate (0.4 < 0.7), high (0.7 ≤ 1). Specificity: The orange diamonds represent the proportion of facial poses (N = 604) matching the hypothesized facial configuration for each emotion category that were observed in response to scenarios classified as the same emotion category, calculated as the complement of the false positive rate (pe^), such that higher scores indicate greater specificity. Error bars represent estimated credibility intervals. We were not able to compute specificity for the emotion categories amusement, embarrassment, pride, or shame because we did not have information about dynamic AUs, which constitute part of the hypothesized facial configurations for these categories. Source data are provided in Supplementary Tables 7 and 10.
Fig. 3 Reliability and specificity of hypothesized facial configurations in each emotion category. a Emotion categories for which all action units (AUs) were available to be coded. b Emotion categories for which AUs were simulated, given that dynamic AUs could not be coded from photographs. Reliability: The gray box-and-whisker plots show the distribution of match scores between the facial poses and hypothesized facial configurations for each emotion category. Each plot presents the maximum and minimum values as whiskers, the interquartile range as the vertical length of the box, and the median as the horizontal line within the box. The dotted horizontal lines denote the degree of reliability, based on criteria from prior research35: none (0 < 0.2), weak (0.2 < 0.4), moderate (0.4 < 0.7), high (0.7 ≤ 1). Specificity: The orange diamonds represent the proportion of facial poses (N = 604) matching the hypothesized facial configuration for each emotion category that were observed in response to scenarios classified as the same emotion category, calculated as the complement of the false positive rate (pe^), such that higher scores indicate greater specificity. Error bars represent estimated credibility intervals. We were not able to compute specificity for the emotion categories amusement, embarrassment, pride, or shame because we did not have information about dynamic AUs, which constitute part of the hypothesized facial configurations for these categories. Source data are provided in Supplementary Tables 7 and 10.CC BY · PubMed Central
Fig. 4 Semi-partial regression coefficients of predictors for face + scenario ratings. Semi-partial regression coefficients for median scenario alone ratings (columns in the left matrix) and face alone ratings (columns in the right matrix) when regressed against face + scenario ratings (rows). As indicated by the color bar at the right of the figure, more positive coefficients appear in shades of green, whereas neutral and more negative coefficients appear in shades of blue. Source data are provided as Source Data file.
Fig. 4 Semi-partial regression coefficients of predictors for face + scenario ratings. Semi-partial regression coefficients for median scenario alone ratings (columns in the left matrix) and face alone ratings (columns in the right matrix) when regressed against face + scenario ratings (rows). As indicated by the color bar at the right of the figure, more positive coefficients appear in shades of green, whereas neutral and more negative coefficients appear in shades of blue. Source data are provided as Source Data file.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 93 references

Research works printed but not linked: 9

  1. Fridlund, A. J. In The Science of Facial Expression (eds Fernández-Dols, J.-M. & Russell, J. A.) 77–92 (Oxford University Press, 2017).
  2. Ekman, P. & Friesen, W. V. Pictures of Facial Affect (Consulting Psychologists Press, Palo Alto, CA, 1976).
  3. Schatz, H. & Ornstein, B. J. In Character: Actors Acting (Bulfinch Press, 2006).
  4. Gendron, M., Mesquita, B. & Barrett, L. F. In Oxford Handbook of Cognitive Psychology (ed. Reisberg, D.) 379–389 (Oxford University Press, 2013).
  5. Ekman, P. & Friesen, W. V. Facial Action Coding (FACS) Manual (Consulting Psychologists Press, San Francisco, 1984).
  6. Hess, U. & Hareli, S. In The Science of Facial Expression (eds Fernandez-Dols, J. M. & Russell, J. A.) 375–396 (Oxford University Press, 2017).
  7. Scarantino, A. In The Psychological Construction of Emotion (eds Russell, J. A. & Barrett, L. F.) 334–376 (Guilford Press, 2015).
  8. Matsumoto, D., Keltner, D., Shiota, M. N., O’Sullivan, M. & Frank, M. In Handbook of Emotions (eds Michael Lewis, M., Haviland-Jones, J. M., & Barrett, L. F.) 211–234 (Guildford Press, 2008).
  9. Keltner, D. & Cordaro, D. T. In The Science of Facial Expression (eds Fernández-Dols, J. M. & Russell, J. A.) 57–75 (Oxford University Press, 2015).

Cited by

See all 87

Metadata

Authors
Le Mau, T., Hoemann, K., Lyons, S. H., Fugate, J. M. B., Brown, E. N., Gendron, M., & Barrett, L. F.
Affiliations
  • KU Leuven (BE)
  • University of Pennsylvania (US)
  • University of Massachusetts Dartmouth (US)
  • Massachusetts Institute of Technology (US)
  • Yale University (US)
  • Northeastern University (US)
Corresponding author
Barrett, Lisa Feldman
Venue
Nature Communications
Published
2021-08-19
Volume
12
Issue
1
Pages
5037-5037
ISSN
2041-1723 (ISSN-L)
OpenAlex
W3193736813
Cited by
74
References
93
Citations by year
2026: 13 · 2025: 14 · 2024: 10 · 2023: 18 · 2022: 19
FWCI
6.18
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, other-oa)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Lirias (submittedVersion, cc-by)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc)
  • OSF Preprints (OSF Preprints) (submittedVersion)
Funders
  • U.S. Department of Health & Human Services | NIH | National Heart, Lung, and Blood Institute (F31 HL140943-01)
  • U.S. Department of Health & Human Services | NIH | National Institute of General Medical Sciences (GM118629)
  • JPB Foundation
  • U.S. Department of Health & Human Services | NIH | National Institute of Mental Health (5 F32 MH105052, R01-MH113234, R01-MH109464)
  • United States Department of Defense | United States Army | Army Research Institute for the Behavioral and Social Sciences (W911NF-16-1-019)
  • U.S. Department of Health & Human Services | NIH | National Cancer Institute (U01-CA193632)
  • National Science Foundation (Civil, Mechanical, and Manufacturing Innovation Grant 1638234)
  • U.S. Department of Health & Human Services | NIH | National Eye Institute (R01-EY020834)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Face Recognition and Perception
  • Physical Sciences › Computer Science › Artificial Intelligence › Emotion and Mood Recognition
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Evolutionary Psychology and Human Behavior
Keywords
emotional facial expressions, facial motion, emotion perception, emotion classification, fear, happiness, surprise
MeSH terms
Adult, Emotions, Face, Facial Expression, Fear, Female, Happiness, Humans, Male, Stereotyped Behavior, Visual Perception, Reproducibility of Results, Cluster Analysis

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Professional actors demonstrate variability, not stereotypical expressions, when portraying emotional states in photographs. https://app.cochat.ai/papers/10.1038/s41467-021-25352-6
The paper
Le Mau, T., Hoemann, K., Lyons, S. H., Fugate, J. M. B., Brown, E. N., Gendron, M., & Barrett, L. F. (2021). Professional actors demonstrate variability, not stereotypical expressions, when portraying emotional states in photographs. Nature Communications. https://doi.org/10.1038/s41467-021-25352-6

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.