1. Previous studies have shown that dopamine (DA) neurons respond to stimuli of behavioral significance, such as primary reward and conditioned stimuli predicting reward and eliciting behavioral reactions. The present study investigated how these responses develop and vary when the behavioral significance of stimuli changes during different stages of learning. Impulses from DA neurons were recorded with movable microelectrodes from areas A8, A9, and A10 in two awake monkeys during the successive acquisition of two behavioral tasks. Impulses of DA neurons were distinguished from other neurons by their long duration (1.8-5.0 ms) and low spontaneous frequency (0.5-7.0 imp/s). 2. In the first task, animals learned to reach in a small box in front of them when it opened visibly and audibly. Before conditioning, DA neurons were activated the first few times that the empty box opened and animals reacted with saccadic eye movements. Neuronal and behavioral responses disappeared on repeated stimulus presentation. Thus neuronal responses were related to the novelty of an unexpected stimulus eliciting orienting behavior. 3. Subsequently, the box contained a small morsel of apple in one out of six trials. Animals reacted with ocular saccades to nearly every box opening and reached out when the morsel was present. One-third of 49 neurons were phasically activated by every door opening. The response was stronger when food was present. Thus DA neurons responded simultaneously to the sight of primary food reward and to the conditioned stimulus associated with reward. 4. When the box contained a morsel of apple on every trial, animals regularly reacted with target-directed eye and arm movements, and the majority of 76 DA neurons responded to door opening. The same neurons lacked responses to a light not associated with task performance that was illuminated at the position of the food box in alternate sessions, thus demonstrating specificity for the behavioral significance of stimuli. 5. The second task employed the operant conditioning of a reaction time situation in which animals reached from a resting key toward a lever when a small light was illuminated. DA neurons lacked responses to the unconditioned light. During task acquisition lasting 2-3 days, one-half of 25 DA neurons were phasically activated when a drop of liquid reward was delivered for reinforcing the reaching movement. In contrast, neurons were not activated when reward was delivered at regular intervals (2.5-3.5 s) but a task was not performed. 6. With established task performance, neurons lost responses to primary reward and instead were activated in their majority by the conditioned light. Thus the response to primary reward was transferred during learning to the conditioned stimulus that predicted reward and had the capacity to elicit arm and eye movement reactions. 7. Subsequently, each animal was overtrained with 30,000 arm movements. This resulted in automated task performance with shortened reaction and movement times. Responses of 165 neurons to the light were progressively reduced in terms of responding neurons (46 and 34% in 2 successive phases, respectively) and overall response magnitude. 8. DA neurons in areas A8 and A10 showed responses similar to those in A9, where most neurons were recorded. In particular, there was no regional preference for neurons responding to a particular stimulus during any learning phase. Thus the populations of A8, A9, and A10 DA neurons showed homogeneous responses during each phase of experimentation. Each neuron either responded to a particular stimulus during each learning phase or lacked responses to any stimuli. 9. These data suggest that, during acquisition of simple behavioral tasks, DA neurons respond to unconditioned and conditioned salient stimuli that attract the attention of the animal, induce behavioral activation, and are associated with reward. Effective stimuli include 1) novel, unexpected stimuli eliciting orienting reactions; 2) primary reward, when delivered as reinforcer during conditioning; and 3) conditioned incentive stimuli, which predict reward and have the capacity to elicit behavioral reactions. The decreased neuronal responsiveness after overtraining parallels the reduced attentional and incentive processes that occur when the task is performed as a habit and stimuli serve merely as temporal reference for automatic task performance. These data provide further evidence for the involvement of DA neurons in arousing, motivational, and behavioral activating processes that determine behavioral reactivity without encoding specific information about the behavioral reaction.
