Lack of a Benign Interpretation Bias in Social Anxiety Disorder

Abstract

Cognitive models of social anxiety posit that recurrent interpretation of ambiguous information as threatening maintains symptoms (e.g. Clark & Wells, 1995, pp. 69–93, Social phobia: Diagnosis, assessment, and treatment. New York: Guilford Press; Rapee & Heimberg, 1997, pp. 741–756, Behavior Research and Therapy, 35). However, biased interpretation may also be represented as a failure to make a benign interpretation of the ambiguous event. Furthermore, interpretation bias can be characterized by both an online (automatic) component and an offline (effortful) component (Hirsch & Clark, 2004, pp. 799–825, Clinical Psychology Review, 24). To measure both benign and threat biases, as well as examine the effect of social anxiety on offline versus online interpretations, Beard and Amir (2009, pp. 1135–1141, Behaviour Research and Therapy, 46) developed the Word Sentence Association Paradigm (WSAP). In the current study, we administered the WSAP to a group of participants diagnosed with social anxiety disorder (SAD) as well as to a group of non-anxious control (NAC) participants. We found that participants with SAD demonstrated a lack of benign online bias, but not an online threat bias when compared to NACs. However, when examining offline biases, SAD patients endorsed social threat interpretations and rejected benign social interpretations to a greater degree than non-anxious individuals. Our results, when taken together, clearly implicate the role of reduced bias toward benign information in SAD.

Keywords:

• interpretation bias
• social anxiety
• information processing

Notes

^1. Studies differ in their terms to describe bias. Some studies refer to positive and negative interpretations, whereas others refer to non-threat and threat interpretations. To remain consistent, we will refer to benign (includes neutral and positive) and threat interpretations from this point.

^2. Four individuals in the NAC group did not complete the LSAS-SR.

^3. Due to the nature of the WSAP paradigm, participants had varying number of response latencies for each trial type. The number of response latencies per trial type ranged from 0 to 30 (endorsement of threat), 1 to 35 (rejection of threat), 1 to 32, (endorsement of benign), and 2 to 30 (rejection of benign).