Prior research shows that humans adjust their learning rate based on environmental volatility. This adaptation has been shown to be reduced in individuals with high levels of anxious-depression, but past studies have been cross-sectional, making it unclear whether this adaptation is trait-like or responsive to treatment. There are also questions about the basic replicability of this effect with recent studies failing to find robust associations.

In the Precision in Psychiatry study, we examined changes in learning rate adaptation in a points-loss task, following four weeks of treatment in two patient groups receiving either internet-based cognitive behavioral therapy (n = 677) or antidepressants (n = 95), compared to a group of healthy participants (n = 95).

All groups showed increased learning rates in volatile environments at baseline, but against our prediction, learning rate adaptation was not cross-sectionally associated with individual differences in anxious-depression. There was also no evidence that learning rate adaptation changed post-treatment, despite symptom improvements. However, in the antidepressant group, symptom reductions correlated with learning rate adaptation. This finding should be interpreted cautiously as we found that the key outcome measure had poor test–retest reliability.

Overall, our results suggest that associations between learning rate adaptation and anxious-depression, if present, are likely modest, depend on task characteristics, may not be present in treatment-seeking adults, and do not improve following treatment.