How observers decode pain from facial reactions. Distributed patterns of facial muscle contractions drive pain judgments in both naturalistic and synthetic expressions
Abstract
Assessing others’ pain relies heavily on facial expressions, yet how individuals translate observable facial features into pain judgements remains poorly understood. We combined a retrospective analysis of naturalistic spontaneous pain expressions across three different cohorts (Study 1, N = 398) with a reverse correlation experiment using synthetic facial stimuli (Study 2, N = 63), to investigate which facial muscle contractions (Action Units, AUs) drive pain judgments. Across both studies, we consistently found brow lowering (AU 4), nose wrinkling (AU 9) and brow raising (AU 1/2) as key contributors, alongside a more variable role of mouth movements. Multivariate pattern analysis indicated that pain judgments rely on a distributed set of facial AUs, rather than on single diagnostic cues. Finally, tailored co-engagement analysis from Study 2 identified AU pairs that, when combined, facilitate (AUs 1 + 12, AUs 4 + 24) or inhibit (AUs 4 + 9) pain assessments. Neither participants’ own sex, the sex of the displayed face, nor participants’ empathy traits substantially influenced pain judgments. These findings provide a first systematic mapping of the observable facial muscle contractions that drive human pain decoding. The identified facial features are best interpretable in terms of multiple, partly independent patterns of facial response, rather than a single prototypical template of pain expression.
