Abstract
The striatum, the major input structure of the basal ganglia in the brain, and the superior colliculus (SC) play distinct roles in reinforcement learning. With training, tonically active neurons (TANs) in the striatum develop a pause in their tonic firing following a reward-related sensory stimulus. The timing of the pause is coincident with the phasic activation of the dopamine neurons in the substantia nigra pars compacta (SNc), which also projects to the striatum. Although the pause response is believed to be important for learning, how the relatively rare and sparsely-distributed TANs can develop a synchronised pause across two hemispheres and how the pause is involved in learning is unclear. In our experiments, the firing patterns of the TANs under urethane anaesthesia were entrained by the local field potential (LFP), which is a global signal in the striatum. Moreover, the firing pattern of the TANs represented the summation of their inputs, including those from cortex, intralaminar thalamus and SNc. By manipulating TANs with cortical stimulation to induce a pause, and activating phasic dopamine release by visual stimulation through the dis-inhibited SC, we provide the first in vivo evidence that the pause of the TANs needs to match the phasic dopamine activation to potentiate corticostriatal synaptic connections onto striatal spiny projection neurons.
The SC plays an important role by sending short-latency signals to striatum following a salient sensory event. The response to a stimulus recorded in SC habituates rapidly when not reinforced. An important property of reward is its ability to block sensory habituation or, indeed, potentiate sensory responsiveness to reward-related stimuli. Therefore, we tested whether SNc stimulation, which is known to induce dopamine release, as a reward could attenuate habituation of visual responses in the rat SC. We found that SNc stimulation paired 1 sec after each light flash at 0.1 Hz for 10 minutes induced a new short-latency component of the collicular VEP that persisted for up to 10 minutes following SNc stimulation. This new sensory responsiveness was greatly reduced when SN stimulation either preceded the light flash by 1.0-1.5 sec or followed it by 3 sec, suggesting that an eligibility period exists for sensory reinforcement. Enhanced visual responsiveness was attenuated following local injection of a dopamine D1 receptor antagonist (SCH23390) or a serotonin 5HT1A antagonist (WAY100635) but not a serotonin 5HT2A antagonist prior to pairing. These data indicate that rewarding stimuli can activate dopamine and serotonin-dependent sensory reinforcement mechanisms that act on the SC to confer salience to previously neutral or habituated stimuli. Together with our finding regarding TANs in the striatum, we propose that the reward-sensitised VEP in the SC and the pause response in TANs are two major physiological mechanisms determining how the brain processes delayed reward, reinforcing actions that lead to a primary reward or secondary reinforcer during reinforcement learning.