Mindfulness meditation modulates reward prediction errors in a passive conditioning task

Reinforcement learning models have demonstrated that phasic activity of dopamine neurons during reward expectation encodes information about the predictability of reward and cues that predict reward. Self-control strategies such as those practiced in mindfulness-based approaches is claimed to reduce negative and positive reactions to stimuli suggesting the hypothesis that such training may influence basic reward processing. Using a passive conditioning task and fMRI in a group of experienced mindfulness meditators and age-matched controls, we tested the hypothesis that mindfulness meditation influence reward and reward prediction error (PE) signals. We found diminished positive and negative PE-related blood-oxygen level-dependent (BOLD) responses in the putamen in meditators compared with controls. In the meditator group this decrease in striatal BOLD responses to reward PE was paralleled by increased activity in posterior insula, a primary interoceptive region. Critically, responses in the putamen during early trials of the conditioning procedure (run 1) were elevated in both meditators and controls. Overall, these results provide evidence that experienced mindfulness meditators are able to attenuate reward prediction signals to valenced stimuli, which may be related to interoceptive processes encoded in the posterior insula.