Investigating the genetic and environmental bases of biases in threat recognition and avoidance in children with anxiety problems.

Background: Adults with anxiety show biased categorization and avoidance of threats. Such biases may emerge through complex interplay between genetics and environments, occurring early in life. Research on threat biases in children has focuses on a restricted range of biases, with insufficient focus on genetic and environmental origins. Here, we explore differences between children with and without anxiety problems in under-studied areas of threat bias. We focused both on associations with anxious phenotype and the underlying gene-environmental correlates for two specific processes: the categorisation of threat faces and avoidance learning. Method: Two-hundred and fifty 10-year old MZ and DZ twin pairs (500 individuals) completed tasks assessing accuracy in the labelling of threatening facial expressions and in the acquisition of avoidant responses to a card associated with a masked threatening face. To assess whether participants met criteria for an anxiety disorder, parents of twins completed a self-guided computerized version of the Development and Well-being Assessment (DAWBA). Comparison of MZ and DZ twin correlations using model-fitting were used to compute estimates of genetic, shared and non-shared environmental effects. Results: Of the 500 twins assessed, 25 (5%) met diagnostic criteria for a current anxiety disorder. Children with anxiety disorders were more accurate in their ability to recognize disgust faces than those without anxiety disorders, but were commensurate on identifying other threatening face emotions (angry, fearful, sad). Children with anxiety disorders but also more strongly avoided selecting a conditioned stimulus than non-anxious children. While recognition of socially threatening faces was moderately heritable, avoidant responses were heavily influenced by the non-shared environment. Conclusion: These data add to other findings on threat biases in anxious children. Specifically, we found biases in the labelling of some negative-valence faces and in the acquisition of avoidant responses. While non-shared environmental effects explained all of the variance on threat avoidance, some of this may be due to measurement error