Page 188 - Demo

Chapter 6186found when specifying a narrow prior (BF = 4.15), and not medium (BF = 2.47) or wide (BF = 1.27) priors. Exploratory: Altered Interoception as Potential Explanation to Alterations in the Link between Physiological Resonance and Perceived Emotional IntensityOur analyses revealed that group differences only existed in the link between physiological measures and Perceived Emotional Intensity, and not Emotion recognition accuracy or the Confidence therein. Therefore, we focused on investigating whether self-reported interoception measures would account for these observed effects in our exploratory analyses. More specifically, both the Bayesian and Frequentist approaches indicated that stronger increases in skin conductance were more strongly (and even exclusively) associated with higher perceived emotional intensity ratings of sad facial expressions in individuals with social anxiety. Based on past findings as well as the increased interoceptive attention reports in the SA group in our study, we were interested to see whether the link in the SA group would change when accounting for differences in interoceptive attention in the link between Skin conductance and Perceived emotional intensity for the different expression categories. Hence, we re-fitted the model on Perceived emotional intensity of all expressions with the interactions including Skin conductance as predictor, and included another three-way interaction term between Emotion category, Skin conductance and Interoceptive attention, as well as the two associated two-way interactions and the main effect of Interoceptive attention (see Table S12 in the Supplemental Material for the model fit). The addition of Interoceptive attention (including its interactions with other predictors) did not change the effect of interest, namely that the association between Perceived emotional intensity of sad expressions and increases in skin conductance was positive (i.e., larger than 0) in the SA group, estimate = 0.53, 95%CI [0.10, 0.96], t(3836) = 3.19, p = 0.007, and significantly stronger compared to the NC group, estimate = 0.60, 95% CI [0.10, 1.11], t(3867) = 2.68, p = .015. These observations were confirmed by Bayes factors for the SA-NC group comparison (BF = 2.47) as well as the slope comparison against zero (BF = 7.99) which were similar in size to the model without Interoceptive attention for at a medium prior specification.