PMC

(A) Session 1 proportion (± SEM) of draws from each deck (A, B, C, D) in 4 blocks of 50 trials for all participants combined over condition (N = 92). Solid lines represent low-frequency penalty decks and dashed lines represent high-frequency penalty decks; blue lines represent negative expected value (EV) decks and orange lines represent positive EV decks. In block 1 participants choose significantly more from deck B followed by deck D and least from decks A and C. In block 4 participants choose similarly from deck B, C, and D, each with significantly more choices than deck A. Supported by a significant block x deck interaction in a three-way analysis of variance (ANOVA) of block x deck x condition, followed by significant simple effects of decks at block 1 and block 4, followed by significant Tukey tests for pairwise comparisons. (B) Percent change (± SEM) in deck choice from session 1 to session 2 in wake (n = 26), sleep (n = 33) and sleep/wake control (n = 33) conditions. Percent change in Deck C was significantly different from percent change in Deck B in ANOVA of condition by deck; there were no significant effects of condition.

(A) Comparison of losses in deck A' and deck B'. In deck A', loss cards are drawn frequently, whereas in deck B', loss cards are drawn infrequently, but include 6 big loss cards. (B) Persistence in the face of big losses from deck B': Relationship between number of big losses as a result of selecting from deck B' and number of intervening trials before selecting from the same deck B' after a big loss. Some participants tended to repeatedly select from deck B' even as they experienced big losses.

(5a) In the simplified iIGT, decision-makers preferred deck iA over deck iC. However, good deck iA ($ 250) contains good final outcome and a better high-frequency gain (5 gains, 5 losses) than bad deck iC ($ −250; 5 draws, 5 losses). Therefore, the gain-loss frequency and final outcome can be used to interpret the experimental result for the iA-iA-iC-iC version. Distinguishing the guiding principle for choice is difficult in this version. (5b) The warm color represents high selection probability and the cold color represents low selection probability. The selection probability maps demonstrated that most subjects preferred to choose good deck iA and avoided bad deck iC across both stages.

Results from one subject in the complex learning task. (A) The history of card selection. Initially, the subject sampled from four decks, then discovered the good deck and stuck with this deck until the reshuffle (red vertical bars). Small red arrows indicate the trial where the subject started to choose from the good deck consistently, which we define as “the onset of good deck choice.” The red circles indicate the trials where the subject received a negative payoff. (B) The history of actual wagering (red) and the predicted probability of a high wager based on the explicit task results (blue). (C) The Bayesian probability for an objectively good deck [the good deck in the upper row in (A)] being a good deck. As the evidence accumulates, this probability updates from 0.5 (the Bayesian prior). It increases when this good deck is selected with a positive outcome or when other decks are selected with a negative outcome. (D) The aligned results of nine onsets of consecutive good deck selection from this subject. The probability of good deck selection (black) and wagering high (red) are aligned at the onsets (shown as the red arrow). The green line is the Bayesian probability of the chosen deck being a good deck.