The estimation of the reward an action will yield is critical in decision-making. To elucidate the role of the basal ganglia in this process, we recorded striatal neurons of monkeys who chose between left and right handle turns, based on the estimated reward probabilities of the actions. During a delay period before the choices, the activity of more than one-third of striatal projection neurons was selective to the values of one of the two actions. Fewer neurons were tuned to relative values or action choice. These results suggest representation of action values in the striatum, which can guide action selection in the basal ganglia circuit.

The basal ganglia are neural structures within the motor and cognitive control circuits in the mammalian forebrain and are interconnected with the neocortex by multiple loops. Dysfunction in these parallel loops caused by damage to the striatum results in major defects in voluntary movement, exemplified in Parkinson's disease and Huntington's disease. These parallel loops have a distributed modular architecture resembling local expert architectures of computational learning models. During sensorimotor learning, such distributed networks may be coordinated by widely spaced striatal interneurons that acquire response properties on the basis of experienced reward.

The basal ganglia have been implicated in motor planning and motor learning. In the study reported here, we directly tested for response plasticity in striatal neurons of macaque monkeys undergoing Pavlovian conditioning. To focus the study, we recorded from the tonically active neurons (TANs) of the striatum, which are known to respond to conditioned sensory stimuli that signal reward delivery and elicit behavioral reactions. The activities of 858 TANs were recorded extracellularly from the striatum in alert behaving macaque monkeys before, during, and after the acquisition of a classical conditioning task. Two monkeys were trained to lick reward juice delivered on a spoon simultaneously with the presentation of a click. Almost no licks were triggered by the cues at the start of training, but by the fifth day more than 90% of licks were triggered, and values were near 100% for the remainder of the 3 week training period. In the striatum, only a small number of TANs responded to the clicks at the start before conditioning (about 17%). During training, the numbers of responding TANs gradually increased, so that by the end of training more than 50-70% of the TANs recorded (51.3-73.5%) became responsive to the clicks. The responses consisted of a pause in firing that occurred approximately 90 msec after the click and that was in some cells preceded by a brief activation and in most cells was followed by a rebound excitation. Prolonged recordings from single TANs (n = 6) showed that individual TANs can acquire a conditioned response within at least as short a time as 10 min. TANs retained such responsiveness after overtraining, and also after a 4 week intermission in training. When the monkey was trained to receive rewards in relation to a new conditioning stimulus, TANs were capable of switching their sensory response to the new stimulus. Histological reconstruction showed that the TANs that became responsive were broadly distributed in the region of striatum explored, which included the dorsal half to two-thirds of the caudate nucleus and putamen over a large anteroposterior span. We conclude that, during the acquisition of a sensorimotor association, TANs widely distributed through the striatum become responsive to sensory stimuli that induce conditioned behavior. This distributed change in activity could serve to modulate the activity of surrounding projection neurons in the striatum engaged in mediating learned behavior.

The projection from the thalamic centre médian-parafascicular (CM-Pf) complex to the caudate nucleus and putamen forms a massive striatal input system in primates. We examined the activity of 118 neurons in the CM and 62 neurons in the Pf nuclei of the thalamus and 310 tonically active neurons (TANs) in the striatum in awake behaving macaque monkeys and analyzed the effects of pharmacologic inactivation of the CM-Pf on the sensory responsiveness of the striatal TANs. A large proportion of CM and Pf neurons responded to visual (53%) and/or auditory beep (61%) or click (91%) stimuli presented in behavioral tasks, and many responded to unexpected auditory, visual, or somatosensory stimuli presented outside the task context. The neurons fell into two classes: those having short-latency facilitatory responses (SLF neurons, predominantly in the Pf) and those having long-latency facilitatory responses (LLF neurons, predominantly in the CM). Responses of both types of neuron appeared regardless of whether or not the sensory stimuli were associated with reward. These response characteristics of CM-Pf neurons sharply contrasted with those of TANs in the striatum, which under the same conditions responded preferentially to stimuli associated with reward. Many CM-Pf neurons responded to alerting stimuli such as unexpected handclaps and noises only for the first few times that they occurred; after that, the identical stimuli gradually became ineffective in evoking responses. Habituation of sensory responses was particularly common for the LLF neurons. Inactivation of neuronal activity in the CM and Pf by local infusion of the GABA(A) receptor agonist, muscimol, almost completely abolished the pause and rebound facilitatory responses of TANs in the striatum. Such injections also diminished behavioral responses to stimuli associated with reward. We suggest that neurons in the CM and Pf supply striatal neurons with information about behaviorally significant sensory events that can activate conditional responses of striatal neurons in combination with dopamine-mediated nigrostriatal inputs having motivational value.