Distinct organization of visual and non-visual signals in visual cortex

Information from rich visual signals and ongoing behavioral variables simultaneously drive the same neurons in the cortex without corrupting each other (Niell et al. 2010, Stringer et al. 2019, Shimaoka et al. 2019). How does the anatomical organization of the cortex support the joint coding of these streams of information? We hypothesized that stimulus-driven and non-stimulus activity follow different anatomical organizations in the visual cortex. To investigate the physical distribution of these signals, we used the Light Beads Microscope (LBM, Demas et al. 2021) to volumetrically image the activity of large populations of neurons in awake mice. We developed a novel 3D cell extraction pipeline to accurately determine the 3D position and morphology of cells in large volumes, which was not possible with existing 2D analysis pipelines. First, we investigated the organization of visual signals by asking whether cells that occupy the same cortical minicolumn have similar orientation preferences. As previously reported (Kondo et al. 2016, Ringach et al. 2016), we found a columnar organization with a sharp dropoff: cells that were within 10 um of each other in the x-y plane had similar stimulus responses across depths, but there was no similarity at larger laminar distances. Next, we investigated the non-stimulus activity of the same cells by looking at the structure of noise correlations. We found that non-stimulus activity also maintained a similar 10 um columnar structure extending up to 100 um in depth. Additionally, we found strong, long-range laminar structure in the pairwise correlation of non-stimulus activity extending across 1 mm. This organization was not present in stimulus responses. The distinct anatomical organization of these two streams of information is analogous to their orthogonal coding in high-dimensional neural space, suggesting that distinct local circuit patterns may act as a mechanism for this separation.