Maryam Azadi¹, Fatemeh Didehvar¹, Tom P. Franken¹; ¹Washington University in St.Louis School of Medicine

An important step in the segmentation of visual scenes into discrete objects is assigning each border to the correct side of foreground (border ownership). Object borders are owned by one side at a time, suggesting a role for inhibitory neurons in the computation of border ownership. While border ownership signals are known to be prominent in primate visual cortex, the relation between border ownership selectivity and cell type is unknown. Here we investigated this using Neuropixels probes in macaque area V4. We inserted the probes orthogonally in the cortex to record from populations of well-isolated neurons while a macaque was viewing visual scenes. We first mapped the aggregate receptive field of the columnar population of simultaneously recorded cells. We then presented a luminance-defined square object such that only one of its borders fell in this receptive field. We varied the side-of-ownership, luminance contrast polarity and square orientation. We identified border ownership-selective (BOS) cells by assessing the statistical significance of the border ownership index (BOI, difference divided by sum of spike rates to opposite sides of ownership) at the border orientation corresponding to the preferred square position. We used spike duration to separate the population into narrow-spiking (putative inhibitory neurons) and broad-spiking (putative excitatory pyramidal) cells. As in prior studies, spike duration was strongly bimodal, and narrow-spiking cells had higher mean firing rates and higher peak-trough ratios than broad-spiking cells. About a third of both narrow-spiking and broad-spiking cells were tuned for border ownership. Broad-spiking BOS cells had significantly higher BOI magnitudes than narrow-spiking BOS cells. Our data suggest that putative pyramidal cells have stronger border ownership selectivity than putative inhibitory cells in V4. Because cells that project to other areas are pyramidal cells, this suggests that border ownership signals are prominent in the projections that leave V4.

Acknowledgements: This work was supported by NIH grant R00EY031795 and a 2024 NARSAD Young Investigator Grant from the Brain & Behavior Research Foundation.