Neural subpopulations in marmoset area MT/MTC detect and discount saccade-related retinal motion

Primates rely on eye movements made 2-3 times every second to scan their visual environment. Each eye movement induces substantial retinal motion, yet observers readily suppress it to stitch together the percept of a stable world. Neurons in the middle temporal (MT) and medial superior temporal (MST) areas respond to motion, but also suppress that response during eye movements (Bremmer et al., 2009; Leopold & Logothetis, 1998; Thiele et al., 2002). This saccadic suppression could reflect the influence of corollary feedback from eye movement related brain structures (Berman et al., 2017). However, visual signals that detect wide-field motion characteristic of a saccade could also contribute to suppression, akin to visual masking (Idrees et al., 2020). We sought to disentangle the relative contributions of visual signals and corollary discharge in driving saccadic suppression in areas MT and MTC in marmoset monkeys. Marmosets freely viewed natural images, blank backgrounds, or externally moving images that simulated saccades, and also made saccades in complete darkness. We found a diversity of saccadic modulation across the population. Some neurons showed short latency excitatory responses to saccades while others showed suppression. Still others showed a biphasic response that began with suppression followed by an excitatory rebound. Neurons with early responses were frequently tuned for the direction of saccades on natural images and retained that same tuning for saccades on blank screens or in complete darkness, thus supporting a role for corollary feedback. However, these neurons also responded with similar directional tuning for retinal motion created by simulating saccades while the eyes were fixed, and thus also appear to integrate visual cues for saccadic motion. More so, these units were biased towards having narrow spike waveforms consistent with putative inhibitory cells, which drive suppression observed in the rest of the population. We also found a subset of neurons that responded robustly to simulated saccadic motion but had negligible response to motion from real saccades, such that they discounted saccade induced motion. Those neurons were biased to have broader spike waveforms, consistent with being putative excitatory projection neurons and could support our percept of stability during eye movements.