A computational cognitive model of the Iowa Gambling Task for finely characterizing decision making in methamphetamine users

Abstract

BackgroundConventional scores from the Iowa Gambling Task (IGT) have been used to assess decision making in a variety of disorders, including substance use. This study extends that work by examining a novel computational cognitive model of the IGT that was hypothesized to capture fine-grained differences in decision-making styles in persons with recent methamphetamine use disorders. MethodsA total of 64 participants were evaluated in this study, which included individuals with lifetime methamphetamine dependence diagnoses, individuals with diagnoses of both methamphetamine and other substance dependence, and control participants with no histories of substance use disorders. Participants were individually modeled performing the IGT by adjusting the parameters of the computational model MAIDEN-IGT with an evolutionary strategy. ResultsThe resultant models explained more than 80% of the participants’ behavior. The common IGT net score alone was not able to detect any inter-group differences. However, the modeling process yielded differentiated inter-group patterns of behavior extracted from the computational models. Relative to control participants, methamphetamine users exhibited a lower estimation of the possible losses as well as the risk associated to the losses. In addition, multi-substance users showed a behavior pattern that affects the evaluation of losses and the risk associated with gains. ConclusionsMAIDEN-IGT computational models were more sensitive than traditional metrics to the risky decision making of persons with methamphetamine use disorders. These novel models raise plausible hypotheses about the nature of motivated behaviors in the context of substance use disorders.