From Expectation to Learning: A Mediational Model of Exposure Therapy Outcome in Anxiety Disorders
Benjamin Straube, Lukas Kirchner, Gina L. Haccou, Isabelle C. Ridderbusch, Ingmar Heinig, Jan
Richter, Ulrike Lueken, Paul Pauli, Jürgen Deckert, Katharina Domschke, Peter Neudeck,
Katja Koelkebeck, Udo Dannlowski, Volker Arolt, Thomas Fydrich, Andreas Ströhle, Jürgen
Margraf, Silvia Schneider, Jürgen Hoyer, Alfons O. Hamm, Hans-Ulrich Wittchen, Tilo Kircher,
Winfried Rief, Andre Pittig
Introduction: Negative expectations about treatment outcomes are associated with poorer
recovery in anxiety disorders. However, it remains unclear whether (1) patients’ expectations
about how effective the treatment will be (treatment‐specific expectations) mediate the link
between generalized negative expectations (i.e. unspecific pessimistic beliefs) and treatment
outcomes, and (2) these expectations affect outcomes via the learning rate during exposure
therapy, that is, the extent to which patients update their fear‐related beliefs based on
corrective experiences. Since treatment‐specific expectations may be more modifiable,
clarifying these mechanisms could improve interventions. This study investigated whether
treatment‐specific expectations and learning rate mediate the effect of generalized negative
expectations on treatment outcomes.
Methods: Data from 605 patients with various anxiety disorders undergoing 12 sessions of
manualized exposure therapy were analyzed. Generalized negative expectations at baseline
(session 0) were hypothesized to predict treatment outcomes (i.e., symptom severity at post
treatment) via expectations about treatment success measured at session 4 and learning rate
during exposure. Depression and anxiety at baseline were covariates. Mediation analyses were
conducted using the PROCESS macro with bootstrap confidence intervals.
Results: Generalized negative expectations predicted poorer treatment outcomes (higher
symptom severity), mediated by lower expectations about treatment success (β = 0.0375; 95%
CI: 0.0088–0.0730). A sequential mediation via these treatment‐specific expectations and
learning rate was also significant (β = 0.0059; 95% CI: 0.0008–0.0134), even after accounting for
baseline symptom levels (R² = 0.3375).
Conclusion: Targeting both generalized and treatment‐specific expectations early in treatment
may enhance exposure‐based learning and improve clinical outcomes in anxiety disorders.
Keywords: Anxiety Disorders, Negative Expectations, Treatment Outcome, Mediation, Learnin
Translational Neuroimaging Marburg 