Emotion-based learning on a simplified card game: The Iowa and Bangor gambling tasks

The Iowa Gambling Task (IGT) has been widely used in the assessment of neurological patients with frontal lesions. Emphasis has been placed on the complexity of the task (i.e., four decks of varying contingency pattern) with the suggestion that the participant must use emotion-based learning to deal with a complex decision-making process. The present study used a single deck card game (the Bangor Gambling Task, BGT), matched in many respects with the Iowa Gambling Task, in which the contingencies varied over time (gradually becoming worse for the participant) rather than across deck (as in the IGT). Forty participants performed both tasks. Performance on the tasks showed many similarities, with participants showing a comparable pattern of incremental learning on both tasks, reaching an analogous final level of performance. More importantly, there was a high correlation (r(2) = .93) in performance between the two tasks, the most salient feature of which was that virtually every participant who fell below Bechara et al.'s (2001) categorisation of impaired IGT performance, also did very poorly on the BGT. These findings bear on the question of whether arguments about the 'complexity' of the Iowa Gambling Task necessarily explain why it appears to require emotion-based learning. The Bangor Gambling Task might also be a useful tool for clinical neuropsychologists, in the assessment of patients with executive dysfunction-given that the task is easier and quicker to administer than the Iowa Gambling Task, but appears to share the same performance features. (C) 2004 Elsevier Inc. All rights reserved.