Participants included treatment seeking methamphetamine dependent adult males (N=46). All individuals underwent fMRI three to four weeks after cessation of substance use. Of the 40 subjects who were followed a median of 370 days, 18 relapsed and 22 did not. The main outcome measure was BOLD activation during a simple two-choice prediction task. During the prediction task, a house was presented, flanked by a person on its left and right. The participant decided on which side of the house a car would appear. Each trial was self-paced to maximize self-determined action, thus the subject determined the number of trials by the latency to select a response. Immediately following the subject’s response, the car was presented for 300 ms on the far left or right side. The screen provided the feedback whether the prediction was correct. Unbeknownst to the participant, the computer determined the response based on the participant’s selection. Three error rate block types included a high chance level (20% of responses were “correct”), a 50% chance-level, and a low (80% of responses were “correct”) chance level. The task captures
