Abstract
<p>Virtual reality exposure (VRE) is thought to rely on the same mechanisms as in vivo exposure, although research investigating these processes is still scarce. According to inhibitory learning accounts, providing safety information prior to exposure may reduce expectancy violation, thereby weakening therapeutic learning. The present study examined this proposition in VRE for public speaking anxiety (PSA). Forty-eight participants with elevated PSA were randomized to either a virtual-emphasis condition, in which the virtual and safe nature of the audience was emphasized, or a control condition before completing a single-session VRE intervention. Primary PSA outcomes included the Personal Report of Public Speaking Anxiety (PRPSA) and Social Anxiety Thoughts Inventory (SATI). PSA decreased across both conditions. Contrary to predictions, emphasizing the virtual nature of the audience did not affect threat expectancies prior to exposure, expectancy change during exposure, or treatment outcomes. Across conditions, we explored potential working mechanisms of VRE. Reductions were found in both VR-specific and general threat expectancies over time. Reductions in the expected severity of feared outcomes were more strongly associated with changes in PSA-related cognitions (SATI), whereas reductions in anticipatory fear were more strongly associated with broader improvements in PSA symptoms (PRPSA). Self-efficacy also increased following exposure and was associated with short-term improvements across outcomes. These findings suggest that safety information prior to exposure does not necessarily attenuate outcomes and provide preliminary insight into potential change mechanisms during VRE.Key words: virtual reality exposure, expectancy violation, expectancy change, public-speaking anxiety, mechanisms, inhibitory learning</p>