When an organ is unable to meet the demands placed on it, illness can arise. As the main functions of the brain are to compute and learn, an understanding of mental illnesses will benefit from an understanding of the computational and learning functions the brain performs, and how these are affected in states of ill-health.
We focus on developing computational tools with real potential for clinical applications.
Ongoing studies
Latest Publications
Berndt, Lioba C. S.; Crawley, Daisy; Tymchyk, Ruslana; Hall, Anna; Onysk, Jakub; Erdmann, Tore; Wimmer, Elliott; Berwian, Isabel M.; Norbury, Agnes; Huys, Quentin J. M.
Early reduction in aversive Pavlovian bias as a mediator of anhedonia improvement during Behavioural Activation in realistic treatment settings. Journal Article
In: PLoS Computational Biology, vol. 22, iss. 7, pp. e1014439, 2026, ISSN: 1553-7358.
@article{BerndtHuys26,
title = {Early reduction in aversive Pavlovian bias as a mediator of anhedonia improvement during Behavioural Activation in realistic treatment settings.},
author = {Lioba C. S. Berndt and Daisy Crawley and Ruslana Tymchyk and Anna Hall and Jakub Onysk and Tore Erdmann and Elliott Wimmer and Isabel M. Berwian and Agnes Norbury and Quentin J. M. Huys},
url = {https://acplab.org/wp-content/uploads/publications/BerndtHuys26.pdf},
doi = {10.1371/journal.pcbi.1014439},
issn = {1553-7358},
year = {2026},
date = {2026-07-14},
journal = {PLoS Computational Biology},
volume = {22},
issue = {7},
pages = {e1014439},
abstract = {Behavioral Activation therapy is an effective treatment for major depressive disorder. Conceptually, the mechanisms through which it acts are thought to involve alterations to reinforcement learning, and several recent studies have supported this in laboratory settings. However, it remains unclear whether reinforcement learning mechanisms are involved in a realistic treatment setting. In a randomized controlled observational study in the UK NHS talking Therapy service, 152 participants with low mood received reinforcement learning assessments using an affective Go/No-Go task. The assessment timepoints were randomized between subjects, with assessments occurring either before (control; n = 78) or during BA therapy (active; n = 74). Changes in Pavlovian biases were quantified using computational modeling. Anhedonia improved during treatment, but not before (p = 0.047; d = 0.53). The active treatment group showed significant changes in Pavlovian parameters compared to controls (pFDR = 0.012; d = 0.65). Pavlovian biases became more positive in the active group (M = 0.44},
keywords = {},
pubstate = {epublish},
tppubtype = {article}
}
Behavioral Activation therapy is an effective treatment for major depressive disorder. Conceptually, the mechanisms through which it acts are thought to involve alterations to reinforcement learning, and several recent studies have supported this in laboratory settings. However, it remains unclear whether reinforcement learning mechanisms are involved in a realistic treatment setting. In a randomized controlled observational study in the UK NHS talking Therapy service, 152 participants with low mood received reinforcement learning assessments using an affective Go/No-Go task. The assessment timepoints were randomized between subjects, with assessments occurring either before (control; n = 78) or during BA therapy (active; n = 74). Changes in Pavlovian biases were quantified using computational modeling. Anhedonia improved during treatment, but not before (p = 0.047; d = 0.53). The active treatment group showed significant changes in Pavlovian parameters compared to controls (pFDR = 0.012; d = 0.65). Pavlovian biases became more positive in the active group (M = 0.44
Cheng, Annie; Konova, Anna; Powers, Albert; Corlett, Philip; Levy, Ifat; Gu, Xiaosi; Huys, Quentin; Pushkarskya, Helen; Fineberg, Sarah; Hauser, Tobias; Bzdok, Danilo; Harpaz-Rotem, Ilan; Babuscio, Theresa; Nichols, Lisa; Zhao, Yize; Sharma, Manu; Meeker, Daniella; Xu, Hua; Rutledge, Robb B.; Pearlson, Godfrey D.; Pittenger, Christopher; Yip, Sarah W.
In: Biological Psychiatry. Cognitive neuroscience and neuroimaging, 2026, ISSN: 2451-9030.
@article{ChengYip26,
title = {Threading the needle: Practical considerations for merging theory-driven computational psychiatry with data-driven analytics to enhance precision health at scale.},
author = {Annie Cheng and Anna Konova and Albert Powers and Philip Corlett and Ifat Levy and Xiaosi Gu and Quentin Huys and Helen Pushkarskya and Sarah Fineberg and Tobias Hauser and Danilo Bzdok and Ilan Harpaz-Rotem and Theresa Babuscio and Lisa Nichols and Yize Zhao and Manu Sharma and Daniella Meeker and Hua Xu and Robb B. Rutledge and Godfrey D. Pearlson and Christopher Pittenger and Sarah W. Yip},
url = {https://acplab.org/wp-content/uploads/publications/ChengYip26.pdf},
doi = {10.1016/j.bpsc.2026.02.009},
issn = {2451-9030},
year = {2026},
date = {2026-02-26},
journal = {Biological Psychiatry. Cognitive neuroscience and neuroimaging},
abstract = {The rapidly evolving field of computational psychiatry enables quantification of specific cognitive processes, and their underlying mechanisms, in a translational and potentially scalable manner, using a combination of data collection via mechanistically informed behavioral tasks and theory-driven mathematical modeling. In parallel, transdiagnostic, dimensional approaches to psychiatric diagnostics, such as RDoC and HiTOP, seek to facilitate links between clinical research and real-world clinical reality, which rarely respects traditional diagnostic boundaries. These two approaches are seldom combined. In addition, while most psychiatric disorders are defined by their longitudinal course, our ability to predict symptom trajectories and tailor treatments to the individual remains limited, in part due to a dearth of longitudinal data collected using assessments sensitive to individual change over time. To address these gaps, the recently launched 'Individually Measured Phenotypes to Advance Computational Translation at Yale' (IMPACT-Y) study is collecting longitudinal data from a transdiagnostic cohort of 2400 individuals, using a combination of 'traditional' clinical research methods (e.g., health records, standardized assessments) and more novel computational approaches (e.g., behavioral tasks with demonstrated sensitivity to latent constructs and to within-person change, spoken narrative data). Here, we discuss unique challenges and opportunities in study design and analysis considerations of IMPACT-Y. Incorporating both theory- and data-driven analytics, we hope that IMPACT-Y will provide an unprecedented resource for characterizing longitudinal trajectories of core computational psychiatry constructs (e.g., reward learning) within and between individuals, for parsing heterogeneity beyond traditional diagnostic categories, and for linking inter- and intra-individual clinical variability to underlying mechanisms.},
keywords = {},
pubstate = {aheadofprint},
tppubtype = {article}
}
The rapidly evolving field of computational psychiatry enables quantification of specific cognitive processes, and their underlying mechanisms, in a translational and potentially scalable manner, using a combination of data collection via mechanistically informed behavioral tasks and theory-driven mathematical modeling. In parallel, transdiagnostic, dimensional approaches to psychiatric diagnostics, such as RDoC and HiTOP, seek to facilitate links between clinical research and real-world clinical reality, which rarely respects traditional diagnostic boundaries. These two approaches are seldom combined. In addition, while most psychiatric disorders are defined by their longitudinal course, our ability to predict symptom trajectories and tailor treatments to the individual remains limited, in part due to a dearth of longitudinal data collected using assessments sensitive to individual change over time. To address these gaps, the recently launched 'Individually Measured Phenotypes to Advance Computational Translation at Yale' (IMPACT-Y) study is collecting longitudinal data from a transdiagnostic cohort of 2400 individuals, using a combination of 'traditional' clinical research methods (e.g., health records, standardized assessments) and more novel computational approaches (e.g., behavioral tasks with demonstrated sensitivity to latent constructs and to within-person change, spoken narrative data). Here, we discuss unique challenges and opportunities in study design and analysis considerations of IMPACT-Y. Incorporating both theory- and data-driven analytics, we hope that IMPACT-Y will provide an unprecedented resource for characterizing longitudinal trajectories of core computational psychiatry constructs (e.g., reward learning) within and between individuals, for parsing heterogeneity beyond traditional diagnostic categories, and for linking inter- and intra-individual clinical variability to underlying mechanisms.
Fielder, Jennifer C.; Shi, Jinyu; McGlade, Daniel; Huys, Quentin J. M.; Steinbeis, Nikolaus
Sense of control buffers against stress. Journal Article
In: eLife, vol. 14, 2026, ISSN: 2050-084X.
@article{FielderSteinbeis26,
title = {Sense of control buffers against stress.},
author = {Jennifer C. Fielder and Jinyu Shi and Daniel McGlade and Quentin J. M. Huys and Nikolaus Steinbeis},
url = {https://acplab.org/wp-content/uploads/publications/FielderSteinbeis26.pdf},
doi = {10.7554/eLife.105025},
issn = {2050-084X},
year = {2026},
date = {2026-02-26},
journal = {eLife},
volume = {14},
abstract = {Stress is one of the most pervasive causes of mental ill health across the lifespan. Subjective dimensions of stress perception, such as perceived control, are especially potent in shaping stress responses. While the impact of reduced or no control over stress is well understood, much less is known about whether heightened feelings of control buffer against the negative impact of later stress. We designed a novel paradigm with excellent psychometric properties to sensitively capture and induce different states of subjective control. Across two studies with a non-clinical sample of 768 adults, we show a robust association between sense of control and stress as well as symptoms of mental ill health. More importantly, in a subsample of 295 participants, we show that compared to a neutral control group, inducing a heightened state of subjective control buffers against the impact of later stress. These findings demonstrate a causal role for a heightened sense of control in mitigating the negative impact of stressful experiences and spell out important directions for future preventative interventions.},
keywords = {},
pubstate = {epublish},
tppubtype = {article}
}
Stress is one of the most pervasive causes of mental ill health across the lifespan. Subjective dimensions of stress perception, such as perceived control, are especially potent in shaping stress responses. While the impact of reduced or no control over stress is well understood, much less is known about whether heightened feelings of control buffer against the negative impact of later stress. We designed a novel paradigm with excellent psychometric properties to sensitively capture and induce different states of subjective control. Across two studies with a non-clinical sample of 768 adults, we show a robust association between sense of control and stress as well as symptoms of mental ill health. More importantly, in a subsample of 295 participants, we show that compared to a neutral control group, inducing a heightened state of subjective control buffers against the impact of later stress. These findings demonstrate a causal role for a heightened sense of control in mitigating the negative impact of stressful experiences and spell out important directions for future preventative interventions.
Laszewska, Agata; Helter, Timea; Baldwin, Ashley; Cleare, Anthony J.; Cowen, Philip J.; Evans, Jonathan; Huys, Quentin J. M.; Kurkar, Micheal; Lewis, Alexander C.; Nixon, Neil; Rastogi, Abhinav; Watson, Stuart; Geddes, John R.; Browning, Michael; Simon, Judit
In: The Lancet Regional Health. Europe, vol. 61, pp. 101533, 2026, ISSN: 2666-7762.
@article{LaszewskaSimon26,
title = {Cost-effectiveness of pramipexole augmentation for acute phase and maintenance therapy of treatment-resistant depression compared to placebo augmentation: economic evaluation of the PAX-D randomised controlled trial.},
author = {Agata Laszewska and Timea Helter and Ashley Baldwin and Anthony J. Cleare and Philip J. Cowen and Jonathan Evans and Quentin J. M. Huys and Micheal Kurkar and Alexander C. Lewis and Neil Nixon and Abhinav Rastogi and Stuart Watson and John R. Geddes and Michael Browning and Judit Simon},
url = {https://acplab.org/wp-content/uploads/publications/LaszewskaSimon26.pdf},
doi = {10.1016/j.lanepe.2025.101533},
issn = {2666-7762},
year = {2026},
date = {2026-02-01},
journal = {The Lancet Regional Health. Europe},
volume = {61},
pages = {101533},
abstract = {Pramipexole augmentation of antidepressant treatment for treatment-resistant depression (TRD) has been shown to improve symptom burden over 12 weeks but with some adverse effects compared to placebo augmentation. We aimed to evaluate the cost-effectiveness of pramipexole augmentation for TRD. We conducted an economic evaluation as part of the PAX-D trial over 12 and 48 weeks. Two costing perspectives, National Health Service and Personal Social Services (NHS + PSS) and societal, were adopted. The primary outcome was quality-adjusted life year (QALY) based on the EQ-5D-5L. Secondary outcomes included year of full capability (YFC) based on the ICECAP-A, and capability-weighted life year (CWLY) based on the OxCAP-MH. Incremental cost-effectiveness ratios (ICERs), cost-effectiveness planes and cost-effectiveness acceptability curves were reported alongside sensitivity analyses. The trial was registered with ISCTRN (ISRCTN84666271) and EudraCT (2019-001023-13) and is complete. From the NHS + PSS perspective, mean incremental cost of pramipexole was £60 (95% CI: -£55, £176) over 12 weeks and £811 (95% CI: £110, £1513) over 48 weeks. The difference in QALY gained was 0.012 (95% CI: 0.003, 0.021) over 12 weeks and 0.090 (95% CI: 0.036, 0.144) over 48 weeks, equivalent to 4 (95% CI: 1, 8) and 33 (95% CI: 13, 52) days in perfect health. The ICER was £5069/QALY (95% CI: -£3642, £35,608) over 12 weeks and £9007/QALY (95% CI: £2,219, £27,258) over 48 weeks, representing over 90% probability of cost-effectiveness at £20,000/QALY threshold. From the societal perspective, pramipexole was on average cost saving and more effective over 48 weeks. Alternative analyses provided consistent conclusions. Pramipexole augmentation for TRD has demonstrated both clinical and cost-effectiveness. Further trials, directly comparing pramipexole to other augmentation strategies, will be useful in determining the position of this repurposed medication in the treatment pathway of depression. National Institute for Health and Care Research, Efficacy and Mechanism Evaluation Programme.},
keywords = {},
pubstate = {epublish},
tppubtype = {article}
}
Pramipexole augmentation of antidepressant treatment for treatment-resistant depression (TRD) has been shown to improve symptom burden over 12 weeks but with some adverse effects compared to placebo augmentation. We aimed to evaluate the cost-effectiveness of pramipexole augmentation for TRD. We conducted an economic evaluation as part of the PAX-D trial over 12 and 48 weeks. Two costing perspectives, National Health Service and Personal Social Services (NHS + PSS) and societal, were adopted. The primary outcome was quality-adjusted life year (QALY) based on the EQ-5D-5L. Secondary outcomes included year of full capability (YFC) based on the ICECAP-A, and capability-weighted life year (CWLY) based on the OxCAP-MH. Incremental cost-effectiveness ratios (ICERs), cost-effectiveness planes and cost-effectiveness acceptability curves were reported alongside sensitivity analyses. The trial was registered with ISCTRN (ISRCTN84666271) and EudraCT (2019-001023-13) and is complete. From the NHS + PSS perspective, mean incremental cost of pramipexole was £60 (95% CI: -£55, £176) over 12 weeks and £811 (95% CI: £110, £1513) over 48 weeks. The difference in QALY gained was 0.012 (95% CI: 0.003, 0.021) over 12 weeks and 0.090 (95% CI: 0.036, 0.144) over 48 weeks, equivalent to 4 (95% CI: 1, 8) and 33 (95% CI: 13, 52) days in perfect health. The ICER was £5069/QALY (95% CI: -£3642, £35,608) over 12 weeks and £9007/QALY (95% CI: £2,219, £27,258) over 48 weeks, representing over 90% probability of cost-effectiveness at £20,000/QALY threshold. From the societal perspective, pramipexole was on average cost saving and more effective over 48 weeks. Alternative analyses provided consistent conclusions. Pramipexole augmentation for TRD has demonstrated both clinical and cost-effectiveness. Further trials, directly comparing pramipexole to other augmentation strategies, will be useful in determining the position of this repurposed medication in the treatment pathway of depression. National Institute for Health and Care Research, Efficacy and Mechanism Evaluation Programme.
Elad, Doron; Story, Giles W.; Berwian, Isabel M.; Stephan, Klaas E.; Walter, Henrik; Huys, Quentin J. M.
Delay discounting correlates with depression but does not predict relapse after antidepressant discontinuation Journal Article
In: Molecular Psychiatry, 2026, ISSN: 1476-5578.
@article{EladHuys26,
title = {Delay discounting correlates with depression but does not predict relapse after antidepressant discontinuation},
author = {Doron Elad and Giles W. Story and Isabel M. Berwian and Klaas E. Stephan and Henrik Walter and Quentin J. M. Huys},
url = {https://acplab.org/wp-content/uploads/publications/EladHuys26.pdf},
doi = {10.1038/s41380-025-03402-5},
issn = {1476-5578},
year = {2026},
date = {2026-01-08},
urldate = {2026-01-08},
journal = {Molecular Psychiatry},
abstract = {Approximately one third of people with Major Depressive Disorder (MDD) experience a relapse within six months of discontinuing antidepressant medication (ADM), however, reliable predictors of relapse following ADM discontinuation are currently lacking. A putative behavioural predictor is delay discounting, which measures a person's impatience to receive reward. Previous studies have linked delay discounting to both MDD and reduced serotonergic function, rendering it a plausible candidate predictor. In this multi-site study we measured delay discounting in participants with remitted MDD (N = 97), before and within six months after discontinuation of ADM, and in matched controls without a lifetime history of MDD (N = 54). Using predictive models, we tested whether either baseline discounting, or an early change in discounting following ADM discontinuation, predicted depressive relapse over a six month follow up period. We also tested differences between remitted MDD and control groups in delay discounting at baseline, and associations between discounting and depressive symptoms. We found that the remitted MDD group, compared to the control group, showed significantly higher (p < 0.05; Cohen's d = 0.34) discounting at baseline. In addition, baseline discounting was positively correlated with depression rating scores (Spearman ρ = 0.24). However, delay discounting did not increase following ADM discontinuation. Neither baseline discounting, nor a change in discounting following ADM discontinuation, predicted subsequent depressive relapse. We conclude that delay discounting is elevated in remitted MDD treated with antidepressant medication. However, delay discounting neither increases following ADM discontinuation, nor does it prospectively predict depressive relapse. These results suggest that delay discounting in Major Depressive Disorder has little relationship with illness trajectory following ADM discontinuation.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Approximately one third of people with Major Depressive Disorder (MDD) experience a relapse within six months of discontinuing antidepressant medication (ADM), however, reliable predictors of relapse following ADM discontinuation are currently lacking. A putative behavioural predictor is delay discounting, which measures a person's impatience to receive reward. Previous studies have linked delay discounting to both MDD and reduced serotonergic function, rendering it a plausible candidate predictor. In this multi-site study we measured delay discounting in participants with remitted MDD (N = 97), before and within six months after discontinuation of ADM, and in matched controls without a lifetime history of MDD (N = 54). Using predictive models, we tested whether either baseline discounting, or an early change in discounting following ADM discontinuation, predicted depressive relapse over a six month follow up period. We also tested differences between remitted MDD and control groups in delay discounting at baseline, and associations between discounting and depressive symptoms. We found that the remitted MDD group, compared to the control group, showed significantly higher (p < 0.05; Cohen's d = 0.34) discounting at baseline. In addition, baseline discounting was positively correlated with depression rating scores (Spearman ρ = 0.24). However, delay discounting did not increase following ADM discontinuation. Neither baseline discounting, nor a change in discounting following ADM discontinuation, predicted subsequent depressive relapse. We conclude that delay discounting is elevated in remitted MDD treated with antidepressant medication. However, delay discounting neither increases following ADM discontinuation, nor does it prospectively predict depressive relapse. These results suggest that delay discounting in Major Depressive Disorder has little relationship with illness trajectory following ADM discontinuation.
Latest Preprints
Qiu, Zeguo; Wang, Muzhi; Lu, Haoyang; Abir, Yaniv; Zharmakhan, Raziya; Singh, Nikki; Poppe, Michaela; Degni, Luigi; Huys, Quentin JM
EEG biomarkers of reinforcement learning and motivation: A multi-task battery Journal Article
In: 2026.
@article{QiuHuys26,
title = {EEG biomarkers of reinforcement learning and motivation: A multi-task battery},
author = {Zeguo Qiu and Muzhi Wang and Haoyang Lu and Yaniv Abir and Raziya Zharmakhan and Nikki Singh and Michaela Poppe and Luigi Degni and Quentin JM Huys},
url = {https://acplab.org/wp-content/uploads/publications/QiuHuys26.pdf},
doi = {10.64898/2026.06.28.735051},
year = {2026},
date = {2026-07-04},
publisher = {openRxiv},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Onysk, Jakub; Chen, Jiazhou; Huys, Quentin JM
On the computational nature of emotions: insights from metareasoning and transformers Journal Article
In: psyArxiv, 2026.
@article{OnyskHuys26,
title = {On the computational nature of emotions: insights from metareasoning and transformers},
author = {Jakub Onysk and Jiazhou Chen and Quentin JM Huys},
url = {https://acplab.org/wp-content/uploads/publications/OnyskHuys26.pdf},
doi = {10.31234/osf.io/m478p_v1},
year = {2026},
date = {2026-05-19},
urldate = {2026-05-18},
journal = {psyArxiv},
publisher = {Center for Open Science},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Serfaty, Jade; Huys, Quentin J. M.
Subjective emotion judgements adhere to principles of Bayesian inference and efficient representation Journal Article
In: psyArxiv, 2026.
@article{SerfatyHuys26,
title = {Subjective emotion judgements adhere to principles of Bayesian inference and efficient representation},
author = {Jade Serfaty and Quentin J. M. Huys},
url = {https://acplab.org/wp-content/uploads/publications/SerfatyHuys26.pdf},
doi = {https://doi.org/10.31234/osf.io/ufmqy_v1},
year = {2026},
date = {2026-05-18},
journal = {psyArxiv},
publisher = {Center for Open Science},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Kim, Taekwan; Viding, Essi; Huys, Quentin J M
Reliable detection of longitudinal change incomputational models Journal Article
In: psyArxiv, 2026.
@article{KimHuys26,
title = {Reliable detection of longitudinal change incomputational models},
author = {Taekwan Kim and Essi Viding and Quentin J M Huys},
url = {https://acplab.org/wp-content/uploads/publications/KimHuys26.pdf},
doi = {https://osf.io/s9dhk},
year = {2026},
date = {2026-04-26},
journal = {psyArxiv},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Hall, Anna F; Huys, Quentin J. M.
Goal progress shapes hedonic experience Journal Article
In: psyArxiv, 2026.
@article{HallHuys26,
title = {Goal progress shapes hedonic experience},
author = {Anna F Hall and Quentin J. M. Huys},
url = {https://acplab.org/wp-content/uploads/publications/HallHuys26.pdf},
doi = {https://doi.org/10.31234/osf.io/4xftg_v1},
year = {2026},
date = {2026-03-11},
journal = {psyArxiv},
keywords = {},
pubstate = {published},
tppubtype = {article}
}