Page 44 - Demo

Chapter 242for all models was informed by model comparisons based on the first model (see Online Resource 2). Eventually, we included random intercepts for Subjectand Trial . The two three-way interactions Age*Congruency*Emotion category and Sex*Congruency*Emotion category, as well as all two-way interactions and main effects, were defined as predictors in all models. To test our first hypothesis, namely that there is an attentional bias to emotional expressions and that this bias is specifically pronounced for angry expressions, we looked at the interaction between Emotion category and Congruency (controlled for by Age and Sex) as hypothesis-specific predictor in our model. Both factors, Emotion category and Congruency, were sum-coded in all analyses. To test our second and third hypotheses, namely that the attentional bias to emotions and specifically to threat expressions is enhanced with higher social anxiety traits, a three-way interaction between Emotion category, Congruency and Social anxiety traits was added as hypothesis-specific predictor to the model (including all twoway interactions). Since we could not find a link between attentional biases and Social anxiety traitsfor any emotion, we did not explore a potential moderating effect of context in this interaction, as indicated in the preregistration2. Based on the same rationale, we did not conduct the second planned exploratory analysis, which aimed to disentangle whether an alteration in attentional bias would be due to heightened vigilance for emotional expressions or a stronger difficulty to disengage from emotional expressions with elevated social anxiety traits. In order to test the presence of a general reduction in attentional bias to emotions with higher autistic levels, with the exception of angry expressions (hypotheses 5), we added a three-way interaction between Emotion category, Congruency and Autistic traits as hypothesis-specific predictors to the general model (including all two-way interactions). Given that, there was indeed no link between the attentional bias towards angry expressions (as to all other expressions) and Autistic traits, we tested for a potential moderating effect of Social anxiety traits on Autistic traits in the prediction of the bias to angry expressions. More specifically, we fitted a model on the reaction times in trials with angry expressions with a three-way 2 To confirm that the attentional bias effect was comparable in all experimental settings, we tested for a modulation of the congruency effect by the specific location (primate park, science festival and lab) as well as by the context (public [primate park + science festival], lab). Neither interaction was significant.