Importance of unpredictability for reward responses in primate dopamine neurons

Abstract

1. We used single neuron recording techniques in two behaving monkeys to investigate the conditions in which dopamine neurons respond to primary rewarding or potentially rewarding stimuli. Animals received drops of liquid either outside behavioral tasks or as rewards during learning or established performance of an auditory reaction time task. 2. Three quarters of dopamine neurons showed a short-latency, phasic response to liquid that was delivered outside the task without being predicted by phasic stimuli. The same neurons responded to liquid reward during learning but not when task performance was established, at which time the neuronal response occurred to the conditioned, reward-predicting, movement-triggering stimulus. 3. These data suggest that the responses of dopamine neurons to rewarding or potentially rewarding liquid are due to the temporally unpredicted stimulus occurrence. A known, reward-predicting, tonic context does not prevent dopamine neurons from responding to the rewarding liquid. The responses during learning apparently occur because reward is not yet reliably predicted by a conditioned phasic stimulus. Because the unpredicted occurrence of reward is of central importance for learning, these responses allow dopamine neurons to play an important role in reward-driven learning.