Although occasional mild social anxiety is a common human experience, social anxiety disorder has a devastating impact on patients' lives, leaving them vulnerable to medical, psychiatric, and socioeconomic complications. A key feature of social anxiety disorder is avoidance of social and/or performance situations in which judgment and evaluation from others might occur. Reducing avoidance is therefore an important treatment goal.

Despite the importance of avoidance, however, it is very difficult to assess a patient's tendency to avoid. Many prior assessments of avoidance measure adaptive (i.e., helpful) avoidance, in which an individual learns to avoid a truly noxious stimulus. However, anxiety disorders are characterized by maladaptive avoidance, in which a relatively safe stimulus is avoided resulting in interference with the individual's goals. In this study, the first aim is to validate a virtual reality paradigm to measure maladaptive avoidance behavior in adults with social anxiety disorder. The second aim is to test whether maladaptive avoidance behavior relates to safety learning (measured by a fear extinction task).