(PS5-47) Predicting Social Anxiety in Behaviorally Inhibited Children: Testing Error-related Negativity and Sex as Potential Third Variables

As social anxiety is often treatment resistant, it is important to study early risk factors to identify potential prevention targets (Bui & Pollack, 2012). A leading risk factor appearing in infancy is behavioral inhibition (BI), a temperament characterized by wariness and avoidance when faced with novelty and uncertainty (Kagan et al., 1984). While BI is a prominent risk factor for social anxiety, a large portion of variance is unaccounted for (Fox & Pine, 2012). The Detection and Dual Control framework suggests that salience detection, such as heightened attention toward making mistakes, is one possible mechanism that links BI to social anxiety (Fox et al., 2023). This hypervigilance to mistakes can be measured with error-related negativity (ERN) though EEG recordings (Meyer et al., 2015). ERN was found to mediate the relation between infant BI and age 12 anxiety, but this relation has not been measured in younger ages (Buzzell et al., 2017). Additionally, as there are sex differences in the prevalence of social anxiety, research has found that the relation between ERN and anxiety is not the same in adult men and women; this has yet to be studied in younger children (Moser et al., 2016; Schneier & Goldmark, 2015). Thus, in this study we hypothesized that BI would predict social anxiety symptoms through ERN, such that high BI in infancy would predict more negative ERN in early childhood, and that more negative ERN would then predict more social anxiety symptoms. Additionally, we predicted that sex would moderate this mediation, with the relation between ERN and social anxiety symptoms being stronger among female children than among male children.  

197 child-mother dyads (children 45% female, 82% White, 93% not Hispanic; mothers 91% White, 95% not Hispanic) completed lab visits when children were 1, 2, and 5-6 years old. At ages 1 and 2, children completed an observational measure of BI (Fox et al., 2001), and mothers completed a demographics form. At age 5-6, children completed a modified flanker EEG task (Brooker & Buss, 2014), and mothers reported their child’s social anxiety (PAS social anxiety subscale, Spence & Rapee, 1999).

Analyses were conducted using SPSS and PROCESS macro (Hayes, 2013). Bivariate correlations between all variables were tested, yielding a significant correlation between BI and social anxiety (r= .233, p < .05). A moderation model was assessed to see if sex moderated the path from ERN to social anxiety. Sex did not significantly moderate; thus it was removed from the model; a mediation model was tested next. ERN did not significantly mediate the relation from BI to social anxiety. Thus, there may be a different mechanism that accounts for the remaining variance between BI and social anxiety. One possibility for these non-significant models is that this sample is too young to exhibit typical ERN patterns associated with anxiety (Meyer et al., 2012). Another possibility is the limited EEG participant data. At present 34 participants had sufficient trials needed for analysis; data collection for an additional 39 participants is complete and will be analyzed for this conference, which may increase statistical power. Thus reducing inhibition in children may remain the best intervention target to prevent the development of social anxiety.