According to contemporary learning theories, the discrepancy, or error, bet
ween the actual and predicted reward determines whether learning occurs whe
n a stimulus is paired with a reward. The role of prediction errors is dire
ctly demonstrated by the observation that learning is blocked when the stim
ulus is paired with a fully predicted reward. By using this blocking proced
ure, we show that the responses of dopamine neurons to conditioned stimuli
was governed differentially by the occurrence of reward prediction errors r
ather than stimulus-reward associations alone, as was the learning of behav
ioural reactions. Both behavioural and neuronal learning occurred predomina
ntly when dopamine neurons registered a reward prediction error at the time
of the reward. Our data indicate that the use of analytical tests derived
from formal behavioural learning theory provides a powerful approach for st
udying the role of single neurons in learning.