Figure 1.

The bandit task paradigm. A, Participants selected among three virtual slot machines (shown as blue squares) whose payout values drifted independently and randomly across trials. The time-varying monetary rewards required participants to learn continuously about the slot machines to maximize their monetary payoffs. At the start of each trial, participants saw three bonuses (numbers displayed within the squares in first screenshot) that had to be added to the slot machine's underlying payout value to determine the total reward. After participants made their choice (circled option), the total reward was displayed. B, One example payoff sequence for all three slot machines over the course of 284 trials. Each colored line indicates a different slot machine.