Applied Computational Psychiatry Group

Head: Quentin Huys

The Applied Computational Psychiatry group focuses on developing computational tools with potential for clinical applications. While we are fascinated by the brain and by computational methods in general, our research invests most in those aspects which we think are most likely to result in treatments.

The research directions over the past two years have been focused on four strands:

To understand what happens when patients stop taking antidepressants and how this leads to relapse. We then attempt to use this knowledge to predict who will relapse and hence aid clinical decision making. In early 2020 we described how effort/reward tradeoffs remain altered in the remitted, medicated depressed state, and how aspect of these changes are predictive of relapse after antidepressant. Since then, we have continued to study this and established further correlates and predictors of discontinuation and relapse after discontinuation (including EEG markers of affective reactivity) and candidate mechanisms (including changes in amygdala reactivity, which is sensitive to discontinuation only those individuals who later go on to relapse; resting-state connectivity with the dorsolateral prefrontal cortex; and dorsolateral prefrontal cortex engagement in recall) while excluding several others.

To understand which cognitive processes are engaged by which psychotherapies and how they contribute to treatment response and treatment response variability. We completed and published a small pilot study suggesting there may be specific changes in affective Pavlovian decision-making during specific phases of psychotherapy. We are currently running a larger-scale replication of this finding in a controlled study within the NHS IAPT services. With Ray Dolan, Tobias Hauser and Steve Fleming we are collaborating on developing novel task-based interventions, and on the use of tasks to examine specific changes in behavioural activation. With other collaborators, we are probing the cognitive structure of dialectic behaviour therapy and metacognitive therapy for anxiety. We are also partnering with industry to develop related approaches to anxiety disorders.

To develop MEG-decoding approaches and cognitive probes to understand automatic negative thoughts. We have developed and preprinted one novel approach to using MEG to relate how internal metareasoning decisions relate to risky decisions and how this in turn relates to individual differences in impulsivity. We are expanding this work to examine automatic thoughts more broadly in depression, and their sensitivity to antidepressant discontinuation.

To develop computational methods to understand affect dynamics. Based on established engineering approaches, we have developed a latent dynamical analysis of experience-sampling data and are currently using this to test novel predictions.

Key publications

1. Berwian, I. M., Wenzel, J. G., Collins, A. G. E., Seifritz, E., Stephan, K. E., Walter, H., & Huys, Q. J. M. (2020). Computational mechanisms of effort and reward decisions in patients with depression and their association with relapse after antidepressant discontinuation. JAMA Psychiatry, 77(5), 513–522. https://doi.org/10.1001/jamapsychiatry.2019.4971
2. Berwian, I. M., Wenzel, J. G., Kuehn, L., Schnuerer, I., Kasper, L., Veer, I. M., Seifritz, E., Stephan, K. E., Walter, H., & Huys, Q. J. M. (2020). The relationship between resting-state functional connectivity, antidepressant discontinuation and depression relapse. Scientific Reports, 10(22346). https://doi.org/10.1038/s41598-020-79170-9
3. Huys, Q. J. M., Russek, E. M., Abitante, G., Kahnt, T., & Gollan, J. K. (2022). Components of behavioral activation therapy for depression engage specific reinforcement learning mechanisms in a pilot study. Computational Psychiatry.