[Skip to Navigation]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 35.153.100.128. Please contact the publisher to request reinstatement.
Viewpoint
July 1, 2020

Translation of Computational Psychiatry in the Context of Addiction

Author Affiliations
  • 1Department of Psychiatry and Psychotherapy, Charité—Universitätsmedizin Berlin (Campus Charité Mitte), Berlin, Germany
  • 2Max Planck Centre for Computational Psychiatry and Ageing Research & Wellcome Centre for Human Neuroimaging, University College London, London, England
JAMA Psychiatry. 2020;77(11):1099-1100. doi:10.1001/jamapsychiatry.2020.1637
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    EXPAND ALL
    Modelling the Future of Treatment in Addiction
    Alex Robinson, PhD Candidate (Clin Psyc) | Turner Institute for Brain and Mental Health, Monash University, Melbourne, Australia
    In their Viewpoint, Liu et al [1] illustrate the promise of computational psychiatry in advancing addiction research. We agree that the model-based/model-free framework has great utility in bridging preclinical and clinical research. However, a priority will be to delineate the impact of each process on the vulnerability to, and recovery from, substance use disorders.

    Model-based and model-free decision-making are often used as proxies for habitual and goal-directed behaviour, respectively. This is because the model-free system solely learns from prior outcomes (through reward prediction errors), while the model-based system prospectively forecasts outcomes using state-transition probabilities [2]. Addiction researchers often
    pit model-based and model-free processes as diametrically opposed systems [3]. However, model-based algorithms also often incorporate elements of previous reward value [2].

    This overlap has important implications for disorders of addiction, in which reward learning is often dysfunctional. For example, studies in methamphetamine use disorder that assumed that model-based and model-free decision-making were orthogonal found a bias towards model-free learning. This result is often interpreted as evidence of “habitual” or “compulsive” behaviour amongst this population [3]. However, because these findings only described the relative weighting of model-based relative to model-free learning, it is difficult to establish whether the individual processes are themselves preserved or dysfunctional.

    In contrast, preclinical research that have not assumed an independence of model-based and model-free learning have shown deficits in both processes in models of methamphetamine use disorder [4]. Furthermore, such approaches have identified that lower model-free functioning (prior to drug administration) predicted subsequent methamphetamine seeking [4]. Such findings lead to a different interpretation to that above – that disruption to reward learning is at least as relevant as the propensity towards habit formation in the development of substance use disorders.

    The independence, or otherwise, of model-based and model-free learning particularly impacts how we translate this framework into novel treatment approaches. For example, clients with model-based deficits may benefit from strategies that strengthen state-related predictions and goal-orientated decisions [5]. In contrast, clients who have both model-based and model-free deficits may benefit from strategies that specifically account for their reduced reward learning to scaffold subsequent goal-orientated learning (e.g., Contingency Management) [5].

    Therefore, while we agree with Liu et al [1] on the potential utility of the model-based/model-free framework, it remains for addiction researchers to characterise the functional consequences of these systems. Specifically, we argue that assessing both model-based and model-free decision-making across the lifespan of substance use disorders will aid the development of more effective interventions.

    Authors: Alex H. Robinson, Trevor T.-J. Chong, Antonio Verdejo-Garcia

    References:

    1. Liu S, Dolan RJ, Heinz A. 2020. doi:10.1001/jamapsychiatry.2020.1637

    2. Daw ND, Gershman SJ, Seymour B, Dayan P, Dolan RJ. 2011. doi:10.1016/j.neuron.2011.02.027

    3. Voon V, Derbyshire K, Rück C, et al. 2015. doi:10.1038/mp.2014.44

    4. Groman SM, Massi B, Mathias SR, Lee D, Taylor JR. 2019. doi:10.1016/j.biopsych.2018.12.017

    5. Verdejo-García A, Alcázar-Córcoles MA, Albein-Urios N. 2019. doi:10.1007/s11065-018-9384-6
    CONFLICT OF INTEREST: AHR is supported by an Australian Government Research Training Program Scholarship; TC is supported by the Australian Research Council (DP 180102383, DE 180100389), the Judith Jane Mason and Harold Stannett Williams Memorial Foundation, the Brain Foundation, the Rebecca L Cooper Medical Research Foundation, and the Society for Mental Health Research; AVG is is funded by an Australian Medical Research Future Fund fellowship (MRF1141214). He has received honorarium from Servier for the preparation of a literature review, and from Elsevier for editing work. He is part of the scientific advisory board of Monclarity (Brainwell), but does not receive honorarium.
    READ MORE
    ×