Figure 1. Experimental design and trial structure.

(A) Trust Game and matched bandit task. Participants played 15 trials with each partner/bandit while in the scanner. On each trial, participants were paired with one of the four stimuli (partner/bandit) and given $10 to invest using a 5-button response box to indicate their investment in $2.50 increments. The monetary investment was then quadrupled, and partners/bandits returned anywhere from 0% to 50% of the money received, allowing for the possibility to double one’s earnings, lose the full investment, or any outcome in between. (B) Task reward structure. Stimuli were randomly assigned to respond with fixed reward rates generated from one of four outcome distributions. Each stimulus deterministically returned less than the participant initially invested (low), more (high), an amount close to the initial investment (neutral), or a random amount. (C) Task event sequence. Participants were given up to 3 s to indicate their choice, after which they experienced a jittered inter-stimulus delay. The returned investment was then displayed on the screen for a fixed 2 s duration. (D) Within-task stimulus presentation. Trials were randomly interleaved such that interactions with each stimulus could occur anywhere from 1 to 15 trials apart, allowing us to probe learning effects from relevant versus temporally adjacent irrelevant outcomes.