Figure 1. Experimental design.

(A) Experimental design. Participants underwent a pain calibration that identified temperatures corresponding to maximum tolerable pain (high pain; 8), pain threshold (low pain; 2), or medium pain (5). They were then positioned in the fMRI scanner and randomly assigned to group. Participants in the Instructed Group were informed about contingencies, while participants in the Uninstructed Group were told to pay attention to the associations between auditory cues and heat but were not informed about the specific cue-outcome contingencies. (B) Trial structure. On each trial, a 2 s auditory cue preceded heat delivered to the participants left forearm. Participants rated perceived pain following offset using an 8-point continuous visual analogue scale. Trials were 48 s long. (C) Instructed and experience-based reversals. Participants first underwent a brief conditioning phase of 5–6 trials in which Original Low Cues (gray) were followed by heat calibrated to elicit low pain (level 2) and Original High Cues (black) were followed by heat calibrated to elicit high pain (level 8). Conditioning was immediately followed by intermittent test trials, in which we delivered medium heat following each cue to test the effects of predictive cues on perceived pain. Following the initial test phase, participants in the Instructed Group were informed about reversals and we delivered medium stimuli to test the effects of instructions. We then paired high heat with the Original Low cue and low heat with the Original High cue, which should act as an experiential reversal, and again administered medium heat to test whether pain reverses upon experience. (D) Example trial order. There were three reversals across the entire task. We used two trial orders that were counterbalanced across participants.