During the performance of the IGT, the subject is repeatedly asked to choose a card from any of 4 decks (marked A, B, C, and D). After turning each card, participants receive some money; after turning certain cards, they are both given money and asked to pay a penalty. Cards in decks A and B bring higher instantaneous gains but there are also larger future losses. These decks are designated as disadvantageous, as their total balance is negative. Decks C and D bring lower instantaneous gains but the losses are slightly smaller; these decks are – in view of the long-term outcome – considered advantageous. Moreover, the decks differ with respect to the distribution of penalties. A penalty of 1250 and 250 CZK (Czech crowns; 1 Euro = approximately 26 CZK) is assigned for one card in decks B and D, respectively; in decks A and C, respectively, this amount is divided among 5 cards (for a detailed schedule, see Figure 1). The task finishes after 100 trials. The subjects were not informed as to the total number of the trials or the distribution of gains and penalties [7].
We used a computer version of IGT (software E-PRIME 1.1). The time scheme was adopted from an fMRI study by Fukui et al [25]. There was a 3.5 second time constraint for the card selection. The total score in the beginning of the performance was 0 CZK. In order to increase the motivation of the participants, the winnings were paid to those who earned it in real CZK (no penalties were enforced in the case of negative total scores).
Evaluation of the IGT performance
A successful IGT performance has been analysed traditionally in terms of the final outcome of the task: IGT score [7-10,12,26] and/ or total gain. The IGT score is the difference between the sums of selections from the advantageous and disadvantageous decks (C+D) – (A+B). A positive number means that the participant preferred the long-term winning decks.
Partial IGT scores (the task divided into fifths) during the task performance are included in majority of the studies as well [1,2,27,28]. Preference for individual decks is changing during the game according to experience from preceding selections. Development of IGT score during the game thus reflects ability to utilize previous experience and therefore characterises learning process of the participants.
A total gain (CZK) is used only as additional information about the quality of the IGT performance; its validity is lower from perspective of strategy. (For example, one selection from deck B can cause a difference of more than 1000 CZK. Therefore, the last selection at the end of the game can theoretically lead to a considerable change in total score despite the usage of the same strategy - in this case repeated selections of the deck B.)
Analysis of the shifts
We consider the objective assessment of the decision-making process in IGT a complex issue, as the traditionally applied scores (as described above) have certain limitations. An analysis of IGT performance based on shifts and their ‘direction’ has been introduced recently [15]. We decided to employ a similar analysis based on counting frequencies of all possible shifts between individual decks. For instance, ‘sAB’ means that a participant shifted his preference from deck A to deck B in next trial, ‘sCC’ means that a participant selected from deck C twice in a row, and so forth. The sum of these shifts during the whole IGT is 99 (there are 100 selections in the task).
These scores were also summed up (aggregated) in terms of the shifts between advantageous (decks C, D) and disadvantageous (A, B) decks. For example, ‘sAC’, ‘sAD’, ‘sBC’ and ‘sBD’ are aggregated in ‘sDaAd’ - e.g. shift from disadvantageous to advantageous.
Since various researchers [27,29] have emphasized the importance of the experience of loss (or penalty) during the task, we also calculated the shifts between decks following a penalty (another 20 scores; their total is equal to the sum of the penalties acquired in all shifts except for the last one).
Finally, we differentiated between the decks with more frequent (decks A, C) and less frequent (B, D) losses (another 8 indexes) [26-28].
Subjective evaluation of the decks
After completing the task, the subjects were questioned about their opinion on the advantageousness and disadvantageousness of the decks; specifically, which decks (or deck) they would have chosen if they could have started the game again and which decks they would have avoided next time. The study groups were compared as to positive or negative evaluation of individual decks. In order to capture the accuracy of the subjective evaluation, the following scoring system was used: correct estimation for each individual deck scored 1 point, incorrect scored -1 point. Scores for all subject thus ranged from -4 to +4.
1