People with depression may be less likely to misclassify ambiguous facial expressions as happy, according to preliminary results from a facial expression recognition study published in the Journal of Affective Disorders.

To further test the theory that people with depression interpret self-referential social information negatively, investigators conducted a prospective cohort study of individuals who presented with depressive symptoms at participating general surgery clinics in the United Kingdom. Enrolled patients attended 4 study visits, each 2 weeks apart, either at home or at their primary care clinic. Depressive symptoms were assessed at each visit per the Patient Health Questionnaire (PHQ-9). Participants performed a computerized facial expression recognition task at the first 3 study visits. During the task, participants were presented with a facial stimulus and then asked to select the perceived emotion from a multiple-choice list. Each emotion was presented in 15 intensities, beginning with an ambiguous facial expression and intensifying to clearer expressions over subsequent trials. An “unbiased hit rate” was calculated for happy and sad facial expressions only, although 6 total emotions were represented across the computerized task. The unbiased hit rate captured patient ability to correctly identify an emotion while adjusting for any tendency to select that emotion when it was not present.

Related Articles

A total of 509 patients were recruited for the original sample, among whom 505 completed the first facial recognition task. At baseline, 45% of patients met diagnostic criteria for depression, and the mean PHQ-9 score in the total sample was 10.38 (standard deviation, 6.62). Overall, participants were more likely to correctly identify happy facial expressions compared with sad facial expressions (P <.001). Compared with happy faces, participants more often misidentified sad faces as another emotion (P =.001). No association was identified between the happy or sad unbiased hit rates and concurrent depressive symptoms. In addition, emotion recognition did not appear to be longitudinally associated with depressive symptoms, as unbiased hit rates were not associated with PHQ-9 scores over time. Weak evidence was present for an association between happy “false alarms” and fewer depressive symptoms; for each nonhappy facial expression incorrectly classified as happy, PHQ-9 scores reduced by 0.06 points (95% CI, −0.11 to −0.004; P =.04). This association was most pronounced for ambiguous facial expressions (P <.001).

These data suggest that although people with depression may be less likely to misclassify ambiguous faces as happy, interpretation of ambiguous facial expressions is not associated with depressive symptoms over time. The relationship between depressive symptoms and interpretation of social cues merits further research.

See source for complete disclosure information.


Bone JK, Lewis G, Button KS, et al. Variation in recognition of happy and sad facial expressions and self-reported depressive symptom severity: a prospective cohort study. J Affect Disord. 2019;257:461-469