Cognitive approaches to non-invasive brain stimulation in the treatment of depression

Thesis

This thesis investigated two cognitive approaches for improving the application of non-invasive brain stimulation in the treatment of depression. The goal of the first project was to test whether combining transcranial direct current stimulation (tDCS) with a reinforcement learning task can reduce negative biases in depression. Recent research suggests that negative biases might arise from overestimating the informativeness of negative vs. positive information which might lead to increased learning from negative information. Two samples were recruited to test whether bifrontal tDCS applied during reinforcement learning increases reward vs. punishment learning in healthy participants and individuals with low mood. For the experimental paradigm, we used the Information Bias Learning Task which manipulates the relative informativeness of reward and punishments. As hypothesised, real compared to sham tDCS selectively increased reward learning in healthy individuals. To investigate whether this effect might be helpful to normalise reinforcement learning in depression, we compared performance between individuals with low mood and healthy participants. We found that individuals with low mood adjusted their punishment learning rate less, and their reward learning rate more to changes in informativeness than healthy individuals. Bifrontal tDCS normalised learning rate adjustment in individuals with low mood, which might be beneficial in depression treatment. The reported findings were only observed in response to bifrontal tDCS applied during (as opposed to before) task performance. This suggests the hypothesis that tDCS combined with reinforcement learning might have larger antidepressant effects than tDCS alone, which can be tested in future clinical trials. The second project was motivated by the idea of investigating different cognitive phenotypes of depression to guide treatment selection for individual patients. Previous research indicates that the relative severity of anxiety vs. anhedonia might differentiate between phenotypes that differ in the TMS target region they best respond to. We therefore recruited participants with varying symptom levels on an online platform to test which parameters extracted from five cognitive tasks differentially relate to symptoms of anxiety and anhedonia. The cognitive process which seemed to best differentiate between anxiety and anhedonia was punishment learning with volatile and stable associations. Future clinical trials are needed to test whether these cognitive task parameters have any clinical value, for example for selecting the optimal TMS target region.

Authors: Sarrazin, V

• DOI: 10.5287/ora-zrgdqzxjq
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: Phenotyping; Negative bias; Depression; Non-invasive brain stimulation; Reinforcement learning
• Subjects: Cognitive neuroscience; Brain stimulation; Computational neuroscience; Psychiatry