People with mood and anxiety disorders learn better from mistakes and negative experiences than they do from rewards, a recent study shows. The researchers published their findings from a meta-analysis in JAMA Psychiatry.
Previous studies show anxiety and depression tied with increased punishment learning, while others show reduced reward sensitivity. The researchers of the current study wanted to obtain more specific information to inform how doctors treat mood and anxiety disorders.
For the meta-analysis, the researchers selected studies that reported choice data from an association between a cognitive task and points-based or monetary rewards or punishments. The search produced 27 studies involving 3085 total patients. Of them, 1242 were diagnosed with mood or anxiety disorders. The researchers conducted both a conventional and simulation analysis.
The researchers found a slight decrease in reward learning rates (standardized mean difference [SMD], −0.021 [95% CI, −0.022 to −0.020]) in anxiety and mood disorder patients compared with controls. Single learning rates showed a slight increase (SMD, 0.041 [95% CI, 0.040 to 0.042]).
From the simulation meta-analysis, the researchers found that individuals with mood and anxiety disorders exhibited a different balance between reward and punishment learning rates compared with control participants. Those patients updated their learned values more than control individuals after receiving a punishment and slightly less than control individuals after receiving a reward outcome. This association was not apparent in the researchers’ conventional meta-analysis. The researchers did not find meaningful differences in inverse temperature or outcome sensitivity.
The researchers assumed findings related to one specific task would translate to another task, which may affect the strength of the study when considering different tasks. A wider selection of benchmarking tasks may also be useful in future, similar studies they reported.
Overall, the meta-analysis supports the researchers’ hypothesis that reinforcement learning is different in patients with mood and anxiety disorders compared to those without. They found a greater benefit in punishment learning compared to reward learning. The results better inform the understanding of negative affective bias, and how physicians target cognitive interventions.
For physicians, the researchers suggest that “rather than encouraging individuals with depression or anxiety to downweigh the subjective experience of negative outcomes or to sit with and tolerate resulting distress…interventions should focus directly on modifying how an individual changes their behavior in response to that negative outcome.”
Pike AC, Robinson OJ. Reinforcement learning in patients with mood and anxiety disorders vs control individuals: a systematic review and meta-analysis. JAMA Psychiatry. 2022;e220051. doi:10.1001/jamapsychiatry.2022.0051