Deconstructing delay discounting in human cocaine addiction using computational modelling and neuroimaging

Michal M Graczyk; Rudolf N Cardinal; Tsen Vei Lim; Salvatore Nigro; Elijah Mak; Karen D Ersche

doi:10.1016/j.bpsc.2024.12.010

Deconstructing delay discounting in human cocaine addiction using computational modelling and neuroimaging

Biol Psychiatry Cogn Neurosci Neuroimaging. 2024 Dec 26:S2451-9022(24)00385-9. doi: 10.1016/j.bpsc.2024.12.010. Online ahead of print.

Authors

Michal M Graczyk¹, Rudolf N Cardinal², Tsen Vei Lim¹, Salvatore Nigro³, Elijah Mak¹, Karen D Ersche⁴

Affiliations

¹ Department of Psychiatry, University of Cambridge, Cambridge, UK.
² Department of Psychiatry, University of Cambridge, Cambridge, UK; Cambridgeshire & Peterborough NHS Foundation Trust, Fulbourn Hospital, Cambridge, UK.
³ Department of Psychiatry, University of Cambridge, Cambridge, UK; Institute of Nanotechnology, National Research Council (CNR-NANOTEC), Campus Ecotekne, Lecce, Italy.
⁴ Department of Psychiatry, University of Cambridge, Cambridge, UK; Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Department of Addictive Behaviour and Addiction Medicine, Central Institute of Mental Health, University of Heidelberg, Mannheim, Germany. Electronic address: ke220@cam.ac.uk.

PMID: 39732337
DOI: 10.1016/j.bpsc.2024.12.010

Abstract

Background: A preference for sooner-smaller over later-larger rewards, known as delay discounting, is a candidate transdiagnostic marker of waiting impulsivity and a research domain criterion. While abnormal discounting rates have been associated with many psychiatric diagnoses and abnormal brain structure, the underlying neuropsychological processes remain largely unknown. Here, we deconstruct delay discounting into choice and rate processes by testing different computational models and investigate their associations with white matter tracts.

Methods: Patients with cocaine use disorder (CUD, n=107) and healthy participants (n=81) completed the Monetary Choice Questionnaire. We computed their discounting rate using the well-known Kirby method, plus logistic regression, single-subject and full hierarchical Bayesian models. In Bayesian models, we additionally included a choice sharpness parameter. Seventy CUD patients and 69 healthy participants also underwent diffusion tensor imaging tractography to quantify streamlines connecting the executive control and valuation brain networks.

Results: CUD patients showed significantly higher discounting rates, and lower choice sharpness, suggesting greater indifference in their choices. Importantly, the full Bayesian model had the greatest reliability for parameter recovery compared with Kirby and logistic regression methods. Using Bayesian estimates, we found that white matter streamlines connecting executive control network with the nucleus accumbens predicted discounting rate in healthy participants, but not in CUD patients.

Conclusions: We demonstrate that measuring delay discounting and choice sharpness directly with a novel computational model explains impulsive choices in CUD patients better than standard hyperbolic discounting. Our findings highlight a distinct neuropsychological phenotype of impulsive discounting, which may be generalizable to other patient groups.

Keywords: Bayesian model; brain networks; decision making; impulsivity; intertemporal choice; research domain criteria.