Measuring reversal learning: Introducing the Variable Iowa Gambling Task in a study of young and old normals

Abstract

We developed a modification of the Iowa Gambling Task (IGT) to test whether it is primarily a measure of reversal learning. Named the Variable IGT (VIGT), the design involves a contingency reversal midway through the task. Two versions of the task enabled us to study the effect of a stronger prepotent response on the ability to identify and adapt to contingency reversal. A significant reversal delay was observed among normal young players with a more dominating reward response. Although transitory, this delay is comparable to the characteristic behavioural impairment observed in patients with damage to the ventromedial prefrontal cortex (VM), addicts, psychopaths and individuals with other self-destructive disorders: They persist in a previously rewarding behaviour despite long-term heavy costs. We also conducted the VIGT in a sample of healthy elderly adults. Results from this sample do not support VM-like or risk-aversive theories of ageing but are inconclusive regarding the frontal ageing hypothesis. Overall, our findings indicate that the VIGT is a sensitive and versatile measure of reversal learning and will serve as a useful instrument in future studies of affective decision making, addiction, and other self-destructive behaviour.