Figure 1. Reinforcement learning task.

Participants completed 200 total trials of a reinforcement learning task, which was divided into two blocks. In each block, participants drew a card from one of four colored decks on every trial by pressing the ‘2’, ‘4’, ‘6’, and ‘8’ keys at the top of the keyboard. After selecting a deck, the top card flipped over to reveal its token outcome. Each deck included three cards with positive outcomes and three cards with negative outcomes. Across task contexts, the distribution of cards within each deck varied (see ‘Table 1’) such that riskier choices were advantageous in one block but disadvantageous in the other block. Participants completed the two blocks in separate ‘casino rooms’ with different colored backgrounds and different colored decks.