Arianne Richer¹, Jeanine Ohene-Agyei², Pénélope Pelland-Goulet¹, Alexis Bellerose¹, Zohair Mharchat¹, Michèle D. Berthaud¹, Francis Gingras^1,3, Émilie St-Pierre¹, Camille Saumure⁴, Roberto Caldara⁴, Daniel Fiset¹, Caroline Blais¹; ¹University of Quebec in Outaouais, Canada, ²University of California Davis, USA, ³University of Quebec in Montreal, Canada, ⁴University of Fribourg, Switzerland

Facial expressions are used as cues to infer others' affective states. However, observers often make mistakes when identifying facial expressions, especially for pain facial expressions (PFE). Moreover, these difficulties are exacerbated when PFE are expressed by women. Drawing on evidence that imprecise mental representations (MR) of emotion can induce perceptual errors, we hypothesized that difficulties in adequately identifying PFE, especially on female faces, may emerge during MR formation. To investigate potential gender differences in the MR of PFE, 49 participants (27 women) completed a genetic-algorithm-based reverse correlation task. This procedure generated individual MR of pain and six basic emotions (anger, disgust, joy, sadness, surprise and fear) on both male and female faces (using MakeHuman software). We analyzed differences in the facial movements captured in these MR proxies and collected independent observer (20 observers, 10 women) ratings of perceived intensity of pain and six basic emotions. Four machine learning models (Decision tree, Support Vector Machine, K-Nearest Neighbors & Naive Bayes) trained on facial movements of the MR collected revealed no significant distinctions between male and female PFE (performing at chance level), although they clearly distinguished between affective states. Furthermore, a MANOVA (7 affective states x 2 genders) revealed that independent observers perceived significantly more sadness and pain in MR of PFE of female compared to male faces. These findings suggest that people hold largely similar mental representations of what a pain expression looks like for male and female targets. The perception task, however, revealed differences in the perceived intensity of affective states for these generated proxies between the two genders.

Acknowledgements: The present study is supported by the Canada Research Chair in cognitive and social vision to Caroline Blais (#CRC-2023-00019) and by the Canada Graduate Scholarship - Doctoral program of the Natural Sciences and Engineering Research Council of Canada to Arianne Richer (#CGS D - 589787).