Are avoidance biases in social anxiety due to biases in stimulus coding or in post-coding behavioral tendencies?

On the Approach-Avoidance Task (AAT) high socially anxious (HSA) individuals exhibit increased behavioral avoidance of faces displaying emotional expressions. Two accounts could explain these findings: 1) HSA individuals have a heightened readiness to code faces in terms of emotionality, and all individuals behaviorally avoid faces coded as emotional (pre-behavior coding bias hypothesis), or 2) everyone is equally ready to code faces in terms of emotionality, but when faces are coded as emotional HSA have a heightened tendency to behaviorally avoid such stimuli (post coding behavioral bias hypothesis). To test these hypotheses, we created the Stimulus-Coding AAT. Participants categorized emotional and non-emotional faces in terms of either gender or emotional expression, before making a standard AAT response. Time to make each type of categorization, and to execute AAT responses following categorization, was assessed in 32 HSA and 32 LSA participants. Groups did not differ in their relative speed to categorize face stimuli on the two dimensions. When participants coded faces in terms of their emotionality, HSA relative to LSA participants demonstrated increased behavioral avoidance of emotional faces. We conclude that these findings are inconsistent with the pre-behavior coding bias hypothesis and support the post-coding behavioral bias hypothesis.