Neuronal population estimates of spatial attention are robust to the presence of microsaccades

Neurons in visual area V4 exhibit attention-related changes in firing rate. Recent work (Lowet et. al., 2018) proposed that attentional modulation of V4 activity occurred only after a microsaccade towards the attended location, suggesting that microsaccades gate attention-related effects. However, other work (Gongchen, et. al., 2022) reported that attentional modulation of neuronal activity in the superior colliculus (SC) occurred before and after microsaccades, and in the absence of microsaccades. Thus, microsaccades may not contribute to attention-related effects in the SC. To determine if these contrasting findings emerge because of differences in brain structure or task demands, we investigated population measures of attention in V4 aligned to microsaccade onset while monkeys performed a visual-spatial attention task (Mayo and Maunsell, 2016), similar to the task used in prior SC work. Monkeys detected an orientation change in one of two simultaneously presented oriented Gabors. We cued attention to one target using an 80% valid visual cue on instruction trials that occurred prior to each block of test trials. During each trial, monkeys fixated until they reported a change in orientation by saccading to the changed target. We recorded over 3500 V4 units from two bilaterally implanted Utah arrays across 54 sessions. We used de-mixed principal component analysis (dPCA; Kobak et. al, 2016) to extract an attention-related latent axis from our high-dimensional neuronal activity. We projected our microsaccade aligned neural population activity onto this attention-related axis and found that attention-related population activity was flat aligned to microsaccade onset, suggesting that attention modulates V4 activity regardless of microsaccades. Trials in which microsaccades occurred appeared identical to the unchanging activity observed in trials without microsaccades. Our results indicate that the modulation of V4 neural activity by attention and microsaccades is largely separable, and that attention modulates V4 activity regardless of the occurrence of a microsaccade.