Threat context distinctly influences multivariate brain-behavior correlations among adults and adolescents

• In Event: Biological embedding of early adversity: new insights on the effects on neurodevelopment and threat processing

Abstract

Aberrations in threat learning and generalization may contribute to anxiety disorders in childhood and adolescence (Glenn et al., 2020). Because several anxiety disorders are linked with selective, threat-specific biases, (LoBue & Pérez-Edgar, 2014; Ohman et al., 2001), biases in threat learning among non-clinical populations may also be context dependent. However, few studies have examined threat generalization across development and no developmental studies have directly compared different classes of generalization stimuli (e.g. social vs. nonsocial), making it difficult to identify factors that contribute to disruptions in threat learning. We used representational similarity analysis in conjunction with fMRI to assess developmental differences in threat representations across social vs non-social contexts.
Twenty adults (Mage = 25.8 ± 4.9) and 16 youth (Mage = 14.1 ± 1.7) completed two differential conditioning and extinction-recall paradigms: a social threat paradigm and a nonsocial threat paradigm. During conditioning, the nonsocial “bells paradigm” paired a colored bell with an aversive alarm and the social “screaming lady paradigm” paired a woman’s face with an aversive scream. Thus, even though paradigms differ on social versus nonsocial dimensions, both sets of stimuli co-occur in dangerous situations in real life. Three weeks after each conditioning paradigm, participants underwent an fMRI extinction recall task, during which they viewed the extinguished threat cue (CS+), safety cue (CS-), and generalization stimuli (GS) consisting of CS-/CS+ blends. Multivoxel BOLD signal patterns were measured in five anatomically-defined brain regions of interest (ROIs) (ventromedial prefrontal cortex (vmPFC), dorsomedial prefrontal cortex (dmPFC), anterior insular cortex (AIC), dorsal anterior cingulate cortex (dACC), amygdala) and representational dissimilarity matrices (RDMs) were created for each ROI. Neural RDMs were correlated with in-scanner ratings of participants’ memory of the CS and GS (e.g., ‘Did this face scream?’), averaged within each age group and task, to examine associations between brain activity and explicit threat memory.
A repeated measures ANOVA revealed a task x ROI x age interaction (F(4,136) = 4.47, p = .002). Follow-up t-tests found that both youth and adults’ AIC activity patterns were more correlated with memory ratings in the social task than nonsocial task (all ps < .002). Additionally, adults’ neural activity was more highly correlated with memory ratings for the social than nonsocial task in the dmPFC (t(19) = 4.57, p < .001) and vmPFC (t(19) = 3.32, p = .004). Finally, youth’s dACC activity was more correlated with memory ratings for the social than nonsocial task (t(15) = 3.61, p = .003).
These findings indicate that the association of neural and behavioral representations of threat is influenced by both age and context. Increased salience of social threats may contribute to greater correspondence between neural activity and behavior ratings across age groups, compared to nonsocial threats. Further, whereas adults may rely on PFC to make categorical distinctions between threatening and nonthreatening stimuli, adolescents may depend on salience network structures during recall of learned threat (Ganella et al., 2017). If replicated, this may have important implications when considering timelines for anxiety treatment.

Authors

• Dana Elizabeth Glenn, University of California - Riverside
  Presenting Author

• Megan A. K. Peters, University of California - Irvine
  Non-Presenting Author

• Daniel S. Pine, National Institutes of Health
  Non-Presenting Author

• Kalina J Michalska, University of California - Riverside
  Non-Presenting Author