Page 71 - Demo

Physiological Resonance and Interpretation of Emotional Expressions693In order to examine whether the perceived intensity of an emotional expression systematically varied depending on expressed emotion and/or the expression modality, we fitted a linear mixed-effects model on the intensity ratings of each participant with regard to the facial and bodily expressions. As in the analysis above, emotion category, expression modality and an interaction between the two of them were defined as fixed effects and we added a random intercept for each subject. Finally, we examined the ratings of subtle facial cues. Given that their nature was largely different from the other stimuli (i.e. artificially created and exclusively added to neutral facial expressions), we kept the analysis for this modality separate. Further, we focused on their perceived intensity since there is no past evidence to indicate that a specific emotion is associated with these cues, hence, they cannot be accurately labelled (see Table 2 in Online Resource 2 for an overview of the provided emotion labels). Thus, we used cue type (tear, blush, dilated pupils versus no cue/neutral) as the sole predictor in the LMM on the intensity scores and added a random intercept for the subject variable. All three models were fitted using the lme4 package (v1.1-23; Bates et al., 2015) in R 3.6.3 (R Core Team, 2020). After fitting a model, post-hoc pairwise comparisons between factor levels and their interactions were calculated by contrasting estimated mariginal means with the emmeans package (v1.4.8; Lenth, 2023). Reporting the test results of all pairwise comparisons would exceed the scope of this paper which is why they are listed in the Tables 3-8 in Online Resource 2. Online Resource 2 also contains the description and results of analyses in which we explored the effect of demographic and personality variables on emotion recognition performance and perceived intensity of emotional expressions. Analysis 2 (Physiological analysis)In the analysis of physiological data, we were specifically interested in identifying expression-specific changes in the shape of each physiological signal related to passive viewing of emotional expressions. Thus, we aimed to describe the entire time course in the response window of interest which differed in duration depending on the signals%u2019 temporal dynamics (see Data preprocessing). For modeling changes in pupil size, SKT and SCL, we extended the approach from studies looking at factors affecting pupil dilatation (Quesque et al., 2019; Wehebrink