3 - (OP18) Associations Among Affective Sentiment, Expectancy Updating, and Overall Outcomes in Exposure Therapy

Exposure therapy is the gold-standard treatment for anxiety disorders, derived from extinction learning principles. Accumulating evidence suggests that reward processes, including experiences of positive affect, may influence extinction learning. However, few studies have explored the role of reward processes and positive affect in exposure therapy. Such exploration could be used to guide therapists’ approach to exposure and improve response. To address these gaps, this study examined whether affect expressed during exposure predicted degree of learning and anxiety symptoms throughout exposure.

Participants (N=47) were adults with social anxiety or panic disorder who completed 9 weekly sessions of individual exposure therapy. Before each exposure, participants rated their expectancy that their feared outcome would occur (pre-expectancy). After each exposure, they rated their expectancy that their feared outcome would occur if they repeated the same exposure in the future (future-expectancy). Expectancy change (the difference between pre- and future-expectancy) was calculated as a measure of learning during exposure. After each exposure, participants engaged in post-exposure rehearsal of outcomes by responding to the following: (1) “If your feared outcome did not occur, how did you know?” (2) “Did anything surprise you?” (3) “What did you learn?” We used sentiment analysis to extract participants’ written and verbal affective expression during post-exposure rehearsal and calculated a difference score that reflected tendency towards positive over negative sentiment (affective sentiment). Anxiety symptoms were assessed before each exposure session and at the end of treatment. Study aims were tested using multilevel models and linear regression.

For written sentiment, higher between-person positive sentiment predicted greater expectancy change (p=.03). There was a main effect of written sentiment predicting anxiety symptoms, such that higher within-person positive sentiment predicted lower anxiety symptoms at the next session (p=.02). Results for verbal sentiment were in a similar direction. Higher within-person positive sentiment predicted greater expectancy change (p=.01). There was a marginal main effect of verbal sentiment predicting anxiety symptoms, such that higher within-person positive sentiment marginally predicted lower anxiety symptoms at the next session (p=.09). Written and verbal average sentiment across exposures did not predict anxiety symptoms at the end of treatment (ps>.22).

This is the first study to examine the role of affective sentiment in exposure learning mechanisms and overall exposure outcomes. More positive sentiment during exposure was associated with greater expectancy change and therefore greater learning, as well as reduced subsequent anxiety symptoms. However, results varied across the between- and within-person level, and overall sentiment did not predict outcomes at the end of treatment. Results lend further support to the importance of reward and positive affect during exposure. It is possible that therapists’ may be able to maximize exposure learning by directing attention to more positive exposure elements, though further study is needed to test this.