Abstract
Previous studies found that ensemble judgments about faces (e.g., judgments about average emotional expression) were less influenced by expertise effects such as the face inversion effect. For instance, people could report the average emotion from a crowd of inverted faces relatively well compared to their performance with single inverted faces (Sun & Chong, 2020), suggesting that ensemble judgments are different from face recognition. The present study sought to find evidence of face recognition (i.e., shared variance in judging face-specific features) in people’s performance for ensemble judgments. Specifically, we investigated the correlation between ensemble judgment abilities for facial expression and facial identity after controlling for ensemble judgment ability for novel objects. Participants completed three tasks: facial expression mode (i.e., the most frequent instance in a group) task, facial identity mode task, and object shape mode task. Random silhouette shapes were used for the object shape mode task. We used the object shape mode task to measure and control for participants’ ensemble judgment ability for novel object categories. In each task, nine face/object images were briefly presented on a virtual 3-by-3 grid, with a random jitter in their positions. After a short interval, three face/object images were presented as multiple-choice options and participants chose the image that was the most frequent in the previous display. Non-guided feedback was given in all trials. We found that facial emotion mode task performance and facial identity mode task performance were positively correlated after controlling the participants’ ensemble judgment performance. This result suggests shared behavioral variance between ensemble judgments for different facial features (expression and identity) that originates not from ensemble judgment ability, but from face recognition ability.