Fig. 1
Odor selection and transreinforcer reversal learning task. a From an initial set of eight food odors, one sweet (SW) and one savory (SV) odor was chosen for each participant (n = 23) such that they were matched in rated pleasantness. A high-intensity and low-intensity version of each selected odor was then established. b High-intensity odors (SWH, SVH) were rated as more pleasant than low-intensity odors (SWL, SVL), and matched in pleasantness within intensity level. Error bars depict within-subject s.e.m. *p < 0.05 in post-hoc paired t-tests. c Two visual symbols (randomly selected for each participant) were used as conditioned stimuli (CS). The four food odors, comprising a 2 × 2 factorial design with value and identity as factors, were used as unconditioned stimuli (US). d Specific CS–US associations conformed to four unique states in the reversal learning task. Transitions between states followed either identity reversals (iREV) or value reversals (vREV). e On each trial, participants first saw a pre-choice “cue” to indicate whether it was a forced-choice or free-choice trial. They then chose one of the two CS’s to receive its deterministically paired odor US. f A given state persisted for either 9 or 12 trials, and states were separated by iREV or vREV. An 84-trial sequence for one example fMRI run is depicted. Participants completed three fMRI runs
