Reanalyzing the Maia and McClelland (2004) Empirical Data: How Do Participants Really Behave in the Iowa Gambling Task?

Abstract

Background: Since 2007, the Iowa Gambling Task (IGT) has been a standardized clinical assessment tool for assessing decision behavior in 13 psychiatric/neurological conditions. After the publication of Maia and McClelland's (1) article, there were two responses in 2005 from Bechara et al. and Maia and McClelland, respectively, discussing whether implicit emotion or explicit knowledge influences the development of foresighted decision strategies under uncertain circumstances (e.g., as simulated in the IGT).

Methods and results: We reanalyze and verify the data obtained by Maia and McClelland (1) in their study "What participants really know in the Iowa Gambling Task" and find that decision-makers were lured into shortsighted decisions by the prospect of immediate gains and losses.

Conclusion: Although the findings of this reanalysis cannot support any arguments concerning the effect of either implicit emotion or explicit knowledge, we find evidence that, based on the gain-loss frequency in the IGT, participants behave myopically. This is consistent with most IGT-related articles (58 out of 86) in Lee et al.'s (2) cross-cultural review. Alternatively, under uncertain circumstances, there is probably no such thing as foresighted decision strategy irrespective of the proposed mechanisms of implicit emotion or explicit knowledge.

Keywords: Iowa Gambling Task; explicit knowledge; foresight; gain–loss frequency; implicit emotion; myopic; somatic marker hypothesis.

Figure 1

The average deck selection frequency for the four IGT decks. Left-hand chart: data from the original experiment, as generated from Bechara et al. (10). Right-hand chart: the orange bars represent the data from Maia and McClelland (1), while the blue bars represent the results obtained by Maia and McClelland when they replicated the Bechara et al. (11) approach. This chart was generated from Maia and McClelland's (1) original data. The right-hand chart presents the average deck selection frequency in Bechara et al. (11) and the study of Maia and McClelland (1), taking their different methodologies (see also Supplementary Figures 1, 2) into account. The analysis shows that, following Bechara et al.'s (11) procedures, there was no difference in terms of participants' preferences for Decks B, C, and D. However, participants showed a lower preference for Deck B than for Deck D, indicating that Maia and McClelland's questionnaire had influenced participants by alerting them to the negative properties of Deck B, thereby reducing the frequency with which Deck B was selected. It should be noted that participants in both studies selected Deck B more often than Deck A, a result that counters the original hypothesis proposed by Damasio and by Bechara et al. (–11), as well as the view held by Maia and McClelland that participants possessed explicit knowledge relating to gains. The color bars represent the mean number of card selections in each deck, and the error bars mark the 1 positive/negative standard deviation from the mean selection number of each deck. Due to the limited number of participants in Maia and McClelland's study, the error bars are only for presentation purposes and not for data correction.