Preliminary research implicates threat-related attention biases in paediatric anxiety disorders. However, major questions exist concerning diagnostic specificity, effects of symptom-severity levels, and threat-stimulus exposure durations in attention paradigms. This study examines these issues in a large, community school-based sample.

A total of 2046 children (ages 6–12 years) were assessed using the Development and Well Being Assessment (DAWBA), Childhood Behavior Checklist (CBCL) and dot-probe tasks. Children were classified based on presence or absence of ‘fear-related’ disorders, ‘distress-related’ disorders, and behavioural disorders. Two dot-probe tasks, which differed in stimulus exposure, assessed attention biases for happy-face and threat-face cues. The main analysis included 1774 children.

For attention bias scores, a three-way interaction emerged among face-cue emotional valence, diagnostic group, and internalizing symptom severity (F = 2.87, p < 0.05). This interaction reflected different associations between internalizing symptom severity and threat-related attention bias across diagnostic groups. In children with no diagnosis (n = 1411, mean difference = 11.03, s.e. = 3.47, df = 1, p < 0.001) and those with distress-related disorders (n = 66, mean difference = 10.63, s.e. = 5.24, df = 1, p < 0.05), high internalizing symptoms predicted vigilance towards threat. However, in children with fear-related disorders (n = 86, mean difference = −11.90, s.e. = 5.94, df = 1, p < 0.05), high internalizing symptoms predicted an opposite tendency, manifesting as greater bias away from threat. These associations did not emerge in the behaviour-disorder group (n = 211).

The association between internalizing symptoms and biased orienting varies with the nature of developmental psychopathology. Both the form and severity of psychopathology moderates threat-related attention biases in children.