Abstract
BACKGROUND: The Iowa Gambling Task (IGT) is a common paradigm used to study the interactions between emotions and decision making, yet little consensus exists on the cognitive process determining participants' decisions, what affects them, and how these processes interact with each other. A novel conceptual framework is proposed according to which behavior in the IGT reflects a balance between two dissociable processes; a cognitively demanding process that tracks each option's long-term payoff, and a lower-level, automatic process that is primarily sensitive to loss frequency and magnitude. METHODS: A behavioral experiment was carried out with a modified version of IGT. In this modified version, participants went through an additional phase of interaction, designed to measure performance without further learning, in which no feedback on individual decisions was given. A secondary distractor task was presented in either the first or the second phase of the experiment. Behavioral measures of performance tracking both payoff and frequency sensitivity in choices were collected throughout the experiment. RESULTS: Consistent with our framework, the results confirmed that: (a) the two competing cognitive processes can be dissociated; (b) that learning from decision outcomes requires central cognitive resources to estimate long-term payoff; and (c) that the decision phase itself can be carried out during an interfering task once learning has occurred. CONCLUSION: The experimental results support our novel description of the cognitive processes underlying performance in the Iowa Gambling Task. They also suggest that patients' impairments in this and other gambling paradigms can originate from a number of different causes, including a failure in allocating resources among cognitive strategies. This latter interpretation might be particularly useful in explaining the impairments of patients with ventromedial prefrontal cortex lesions and, by extension, the contribution of this brain region to human decision making.