What is reward prediction error?

Reward prediction error is a concept that plays a crucial role in learning and habit formation, closely linked with the release of dopamine in the brain. According to Andrew Huberman, it involves the discrepancy between expected rewards and actual outcomes. If an expected reward is received or exceeds expectations, dopamine release is activated, reinforcing the behavior leading to that reward. Conversely, if a reward is less than expected or not received at all, dopamine levels can drop, altering future behavior patterns.

For instance, when someone expects a reward (like praise, a physical reward, or a personal achievement) and that reward materializes, it reinforces the behavior associated with it. This makes it more likely that the behavior will be repeated. However, if expected rewards fail to materialize or are lower than expected, this leads to a drop in dopamine, which might decrease the likelihood of repeating the behavior.

Reward prediction error guides a wide span of learning processes and habit formation, with deviations from expected outcomes (either positive or negative) strongly influencing subsequent motivation and behavior 1.