Quantitative analysis of substantia nigra pars reticulata activity during a visually guided saccade task

Several lines of evidence suggest that the pars reticulata subdivision of the substantia nigra (SNr) plays a role in the generation of saccadic eye movements. However, the responses of SNr neurons during saccades have not been examined with the same level of quantitative detail as the responses of neurons in other key saccadic areas. For this report, we examined the firing rates of 72 SNr neurons while awake-behaving primates correctly performed an average of 136 trials of a visually guided delayed saccade task. On each trial, the location of the visual target was chosen randomly from a grid spanning 40°of horizontal and vertical visual angle. We measured the firing rates of each neuron during five intervals on every trial: a baseline interval, a fixation interval, a visual interval, a movement interval, and a reward interval. We found four distinct classes of SNr neurons. Two classes of neurons had firing rates that decreased during delayed saccade trials. The firing rates of discrete pausers decreased after the onset of a contralateral target and/or before the onset of a saccade that would align gaze with that target. The firing rates of universal pausers decreased after fixation on all trials and remained below baseline until the delivery of reinforcement. We also found two classes of SNr neurons with firing rates that increased during delayed saccade trials. The firing rates of bursters increased after the onset of a contralateral target and/or before the onset of a saccade aligning gaze with that target. The firing rates of pause-bursters increased after the onset of a contralateral target but decreased after the illumination of an ipsilateral target. Our quantification of the response profiles of SNr neurons yielded three novel findings. First, we found that some SNr neurons generate saccade-related increases in activity. Second, we found that, for nearly all SNr neurons, the relationship between firing rate and horizontal and vertical saccade amplitude could be well described by a planar surface within the range of movements we sampled. Finally we found that for most SNr neurons, saccade-related modulations in activity were highly variable on a trial-by-trial basis.