Poster #4 - Measuring Attention Bias: Predicting Anxiety in Adolescents Using Eye Tracking, Reaction Time, and EEG Measures

• In Event: Poster Session 08

Abstract

Research on Social Anxiety Disorders (SAD) suggests a relatively robust association with attentional bias towards threatening stimuli (threat bias; Abend, et al., 2018). One measure of widely-used measure of threat bias in the developmental literature is the facial Dot Probe Task (DPT). Historically, threat bias in the DPT has been measured behaviorally using differences in reaction time between threat congruent and threat incongruent trials (Bantin, et al., 2016), however other research using the DPT has operationalized threat bias through EEG-derived event-related potentials (ERPs; Torrence & Troup 2017) or eye tracking-based measures (Burris, et al., 2017). The current study sought to explore the association between reaction time-based, ERP-based, and eye tracking-based measures of threat bias, and the extent to which these various measures of attention bias predict SAD in adolescents.
Participants were recruited as part of a larger longitudinal study on anxiety development during adolescence (of which T1 has been completed and T2 data one year later are currently being collected). Participants included 78 adolescents, 56% Female, M Age = 13.75, Age Range 12.75 – 15.83, 85% White, 11% Mixed Race/Ethnicity, 3% Black/African-American, 1% Hispanic/Latino, with a median family income of $71,000 - $80,000 per year. Participants completed a vertically aligned DPT, in which angry and neutral face pairs were displayed for 500ms followed by a 500ms arrow probe and 1000ms response period in which participants indicated the direction the arrow pointed using a handheld response box. Reaction time-based threat bias was calculated as the mean response time to probe (angry – neutral trials). Eye tracking-based threat bias was calculated in two ways: proportion time fixating angry faces compared to total trial length and latency to fixate on the probe following angry trials. ERP-based threat bias was calculated as differences in P1, N170, and N2 ERPs between angry and neutral trials. SAD symptoms were calculated as the total number of symptoms adolescents self-reported on the SAD subscale of the Anxiety Disorders Interview Schedule (ADIS; Silverman & Albano, 1996).
Initial correlations indicated somewhat low congruence between measures of threat bias. Specifically, dwell time proportion to angry faces was significantly correlated with ΔN170 (r=.23, p=.04) and was negatively correlated with reaction time-based threat bias (r=-.26, p=.03). No other measures of threat bias were significantly correlated with one another. Results also indicated a significant sex difference in SAD such that females reported higher SAD than males, t(59) = 3.30, p = .002. As related to SAD, regression analyses controlling for sex differences show ΔN170, reflecting blunted attentional selection and discrimination in the presence of threat, marginally predicted more SAD symptoms (β = 1.00, p = .07), while latency to probe fixation marginally predicted fewer SAD symptoms (β = -.01, p = .06).
Together, these preliminary results show some measures of threat bias may predict adolescent self-reported anxiety. Further analyses to be presented at this conference will explore other potential measures of threat bias (i.e., trial-level bias scores, Zvielli, 2015) as well as how these measures of threat bias predict anxiety longitudinally.

Authors

• Austen Trainer, Pennsylvania State University, University Park
  Presenting Author

• Sarah Myruski, Pennsylvania State University, University Park

• Koraly Elisa Perez-Edgar, Pennsylvania State University, University Park

• Kristin A. Buss, Pennsylvania State University, University Park