Importance of punishment frequency in the Iowa gambling task: an fMRI study

It has been widely found that in the Iowa Gambling Task (IGT; Bechara et al. Cognition, 50(1), 7-15 1994) normal subjects would gradually learn to prefer obtaining rewards for long-term benefits than seeking immediate rewards to maximize the overall profit. The current study aimed to gain an understanding of how punishment frequency in the IGT would be processed and its association with subjects' reward preferences. In this study, we employed the clinical version of the IGT, in which response options are not only different in the long-term outcome, but also associated with different punishment frequencies. Event-related functional Magnetic Resonance Imaging (fMRI) was used to capture the subjects' brain activity when performing the IGT. A total of 24 male subjects (mean age = 21.7 years, SD = 1.8 years), who were university students, participated in the experiment. It is found that subjects learned to select more from the decks that were advantageous in the long-term, but they were more sensitive to the effect of long-term outcome under the condition of high punishment frequency. The corresponding brain activation showed that the Anterior Cingulate Cortex (ACC) had significantly higher activation during the disadvantageous choices than the advantageous choices. Such activity difference between the two conditions of long-term outcome was more prominent with high punishment frequency than low punishment frequency; and this brain activity difference was significantly correlated with the behavioral performance under the condition of high punishment frequency. The results suggested that only in the context with high punishment frequency, there would be increased neural activity in ACC when subjects intended to select from the disadvantageous choices so that these choices would be inhibited and advantageous choices would be selected.