Dopamine: Context and counterfactuals.

Dopamine and Reward Prediction Errors How does the brain learn what feels good? How does it adapt to new foods, bad bets, and played-out songs? How is it tricked into wanting ever-increasing doses of heroin or cocaine? Is there a simple rule that links neural signaling to the fundamental approach and avoidance responses on which more complex behavior is built? It is now almost 20 y since Schultz, Montague, and Dayan (1) gave a provocative answer to these questions, an answer that is now recognized as one of the cornerstones of reinforcement learning and neuroeconomics. In a series of prior papers (reviewed in ref. 2), Schultz and collaborators had demonstrated that dopamine-releasing cells in the ventral