Inter-subject correlation of audience facial expressions during theatrical performances

During theatrical performances individuals within the audience experience various emotional states, and this is often reflected in their concurrent facial expressions. Information from these facial expressions has been used to understand the subjective experiences of individual audience members but could also provide information on the dynamic experiences of the audience as a whole. We explored this possibility in this study by filming the faces of multiple audience members at several different theatrical and cabaret performances. We analyzed each of the audience faces to determine the likelihood of the presence of 8 different facial expressions at 20 ms time intervals for the duration of the performances. By measuring the inter-subject correlation (ISC) between facial expression likelihood functions of audience members, we assessed how the expression synchrony was related to predictions of audience engagement. There was strong evidence that neutral, happy, anger and disgust expression ISCs were positively correlated with predictions of audience engagement across performances, and expression ISCs accounted for between 20% and 24% of the variance in predicted audience engagement. Facial expression synchrony was greater between individuals physically closer to each other within the auditoriums, suggesting physical proximity may result in emotional contagion during performances. Neutral, happy and angry expressions of older audience members were more idiosyncratic, whilst these expressions were more synchronized in female individuals and those individuals with higher measures of empathy. Overall, our findings suggest that the moments of theatrical performances that are predicted to be most engaging, are those where the emotional state of the audience is most synchronized. Consequently, facial expression synchronization could be used as a real-time and non-invasive indicator of audience engagement.