The Iowa Gambling Task (IGT) is sensitive to decision-making impairments in several clinical groups with frontal impairment. However the complexity of the IGT, particularly in terms of its learning requirements, makes it difficult to know whether disadvantageous (risky) selections in this task reflect deliberate risk taking or a failure to recognise risk. To determine whether propensity for risk taking contributes to IGT performance, we correlated IGT selections with a measure of propensity for risk taking from the Balloon Analogue Risk Task (BART), taking into account potential moderating effects of IGT learning requirements, and trait impulsivity, which is associated with learning difficulties. We found that IGT and BART performance were related, but only in the later stages of the IGT, and only in participants with low trait impulsivity. This finding suggests that IGT performance may reflect different underlying processes in individuals with low and high trait impulsivity. In individuals with low trait impulsivity, it appears that risky selections in the IGT reflect in part, propensity for risk seeking, but only after the development of explicit knowledge of IGT risks after a period of learning.
The Iowa Gambling Task (IGT) is widely used to study decision-making under risk and uncertainty and is a sensitive tool for detecting frontal dysfunction in several psychiatric populations (e.g. substance dependence, ADHD, pathological gambling) (e.g. Bechara and Damasio, 2002, Goudriaan et al., 2006, Malloy-Diniz et al., 2007, Stout et al., 2004 and Stout et al., 2005). Although the IGT’s sensitivity for detecting decision-making impairment is well established, recent studies have highlighted the complexity of this task and the challenges this poses for understanding what functions (or dysfunctions) it measures (Brand et al., 2007, Buelow and Suhr, 2009 and Dunn et al., 2006).
For example, the results of a recent study suggest that risk taking in the early and later stages of the IGT need to be considered separately (Brand et al., 2007). This study found that a person’s propensity for risk taking, as measured by the Game of Dice Task (where risks are explicit) (Brand et al., 2005), was related to later IGT selections, but not earlier IGT selections. This suggests that in the earlier stages of the IGT, when players have little explicit knowledge about IGT alternatives, risk taking is not a deliberate act, but rather, reflects a failure to recognise risk. As the task progresses however, players presumably develop explicit knowledge of the risk profile across IGT alternatives. At this stage of the task, the player is able to express their propensity for risk taking, either by continued ‘risky’ choices (despite knowing risks), or by safer choice behaviour, which reveals their avoidance of risk.
In contrast to the IGT, where players cannot express their risk propensity until they have learned the risks, the Balloon Analogue Risk Task (BART) (Lejuez et al., 2002) is designed so that players are able to express their risk propensity from the beginning of the task. This may account for why three previous studies have found no association between the tasks (Aklin et al., 2005, Bishara et al., 2009 and Lejuez, Aklin, Jones et al., 2003), despite the fact that each task is separately related to drug abuse and other risk taking behaviours (Lejuez, Aklin, Zvolensky et al., 2003, Lejuez, Aklin, Jones et al., 2003, Stout et al., 2004 and Stout et al., 2005). Perhaps if these studies had separated IGT selections into early (pre-learning) and later (post-learning) stages of the task, later IGT selections would have been associated with BART performance. For such a relationship to emerge however, a player’s ability to learn IGT risks should be taken into account, especially since there is evidence of heterogeneity in the ability of some groups to learn about risk from experience (Stout et al., 2005).
Trait impulsivity, measured in its narrow form using questionnaires such as the Eysenck I7 and Barratt Impulsiveness Scale1, is associated with learning difficulties in problem solving situations (McMurran, Blair, & Egan, 2002) and increased risk taking in situations where learning is required (e.g. IGT) (Franken et al., 2008 and Sweitzer et al., 2008). This is important, because past studies that have attempted to correlate IGT and BART performance have focussed on samples with high trait impulsivity such as substance abusers, who are known to have problems learning from experience about risk (Stout et al., 2005). In an impulsive sample, inefficient learning in the IGT may mean that IGT performance, even in the later stages of the task, reflects unintentional risk taking rather than deliberate risk taking. Thus, it may be premature to conclude that IGT and BART are generally unrelated until the BART is compared to the most relevant stage of the IGT, and relevant individual differences such as impulsivity have been taken into account.
Thus, the aim of this study was to re-examine the association between IGT and BART, by correlating IGT and BART performance in early and later stages of the IGT separately. We also correlated IGT and BART performance from early and late stages of the IGT in groups with low and high trait impulsivity separately. We hypothesised that IGT and BART performance would be associated in the later stages of the IGT, but only in individuals with low trait impulsivity, reflecting their ability to learn IGT risks and therefore express their propensity for risk taking following a period of learning.
