The role of sensitivity to reward and sensitivity to punishment in decision making on the Iowa Gambling Task

Abstract

Poor decision making, particularly in situations involving complexity or uncertainty, is associated with brain injury to ventromedial prefrontal cortex, along with a variety of self-control disorders (e.g., pathological gambling, eating disorders, and substance abuse). Decision-making deficits have been attributed to atypical sensitivity to reward and/or punishment, which may reflect abnormalities in the way impaired individuals scale the physical dimensions of rewards and punishers. Current cognitive neuroscience techniques for identifying decision-making deficits and measuring sensitivity are problematic. The Iowa Gambling Task (IGT) is widely used to identify deficits; however, methodological issues contribute to inconsistent cross-study results and high individual variability in non-clinical samples. Popular self-report measures of sensitivity yield inconsistent and weak correlations with the IGT, and may not correctly measure the scaling of rewards and punishers. A paradigm from behavioural psychology was adopted to address these issues. In Part 1, 50 healthy adults completed an extended IGT (200 trials) and two self-report measures of sensitivity (the BIS/BAS and SPSRQ). Analysis of stable decision making from IGT Trials 101-200 revealed two main behavioural patterns: 76% of participants strongly preferred one or both advantageous card decks (good decision makers), while 20% showed no strong deck preferences and hence obtained the lowest scores (poor decision makers). In Part 2, a sub-sample of 30 participants completed a novel operant card game, the Auckland Card Task (ACT). The generalised matching law was fitted to data, yielding estimates of sensitivity to reward frequency, reward magnitude, punishment frequency, and punishment magnitude. Poor decision makers did not differ from good decision makers in sensitivity to the frequency of rewards and punishers, but exhibited significantly lower sensitivity to the magnitude of rewards and punishers. Self-report scores were unrelated to decision making in the ACT or IGT. It was concluded that poor decision making in the IGT may be associated with a poor ability to track the average size of infrequent and unpredictable monetary penalties. More generally, the study supports the view that decision-making ability is related to the way in which individuals scale rewards and punishers, and demonstrates the viability of using operant procedures to investigate human decision-making deficits.