Reward Prediction Error (RPE)

« Back to Glossary Index

Reward prediction error (RPE) is a specific type of prediction error involving expected and actual outcomes related to reward, motivation, or meaningful consequences. A positive RPE occurs when outcomes are better than expected, increasing the likelihood that associated behaviors will be repeated. A negative RPE occurs when outcomes are worse than expected, reducing that tendency. Research suggests dopamine signaling plays an important role in communicating these prediction differences and updating future expectations. Reward prediction errors do not involve pleasure alone; the brain may also learn from outcomes involving relief, safety, fairness, social approval, or avoidance of harm. Also see Reward Prediction, Reinforcement Learning, Prediction, and Striatum.