Page 108 - Demo

Chapter 4106Metacognitive sensitivity. To examine how well an individual%u2019s confidence ratings could distinguish between accurate and inaccurate trials in the emotion recognition task, we calculated the hit and false alarm rate pairs with increasing confidence levels (11, according to points on the Likert scale) for each subject and employed the area under the type 2 ROC curve (AUROC2) approach according to Fleming and Lau (Fleming & Lau, 2014). More specifically, each confidence level was taken as a criterion to distinguish between low and high confidence trials; starting with a criterion in which only zeroes were regarded as low confidence ratings and all higher values were regarded as high confidence ratings, up until a criterion in which all trials below the highest confidence rating (100) were regarded as low confidence trials and only the highest rating was regarded as high confidence. The resulting probabilities for hits, p(high confidence|correct), and false alarms, p(high confidence|incorrect), were plotted against each other for each confidence level. The resulting area under this ROC2 curve was taken as an index for the subject%u2019s metacognitive sensitivity, describing how well an individual%u2019s confidence ratings were scaled to actual emotion recognition accuracy. The link to each clinical trait was then tested with a correlational analysis. Facial EMG AnalysisFacial muscle activity (mimicry). By measuring facial muscle activity over the Corrugator Supercilii and Zygomaticus major regions, we could assess mimicry responses to angry, happy, sad and fearful expressions, with neutral expressions acting as a reference category. In order to examine whether social anxiety traits are associated to an enhanced mimicry of specifically angry (negative) expressions, we fitted a linear model on the category-averaged corrugator activity (i.e., taking the mean corrugator activity of all trials belonging to the same emotion category) with emotion category, social anxiety traits and their interaction as fixed effects. We also aimed to explore zygomaticus activity for mimicry of happy expressions and, therefore, used the same independent variables to predict categoryaveraged zygomaticus activity (i.e., taking the mean zygomaticus activity of all trials belonging to the same emotion category). By replacing social anxiety traits with autistic traits in the other two linear models on category-averaged corrugator and zygomaticus activity, we then tested whether typical mimicry patterns are indeed reduced with higher autistic traits (i.e., less corrugator activity for negative expressions (specifically anger), less zygomaticus activity for happy expressions and less decrease in corrugator activity for happy expressions). Coefficients for the emotion categories (main effects and interactions) were calculated by contrasting