In 322 Allen Neuropixels probes of mouse visual cortex, the unipolar LFP alpha-beta/gamma crossover was in layer 5, 220 µm below CCF layer 4; bipolar and CSD signals did not show the gradient

Question: Mendoza-Halliday et al. (2024) proposed that the depth where relative alpha-beta and gamma LFP power cross marks cortical layer 4. In mouse, they assigned layers from this pattern itself. We tested the crossover against anatomical layers. Approach: We used 322 Neuropixels probes in visual cortex of 57 mice (Allen Brain Observatory Visual Coding). We read grey-screen and flash-block LFP by HTTP byte ranges. We labelled each channel with its CCFv3 layer and found the flash CSD early sink. A FLIP-like search found the pattern and its crossover in unipolar, common-average, bipolar and CSD-domain signals. Intervals are session-bootstrap 95% intervals. Results: The pattern occurred in 81% of probes (unipolar) and 77% (common average), but in 11% (bipolar) and 4% (CSD), against 2% for shuffled depth order. The unipolar crossover was a median 218 µm (205-237) below the CCF layer-4 centre. It was in layer 5 in 149 of 170 probes and inside layer 4 in 2.4%. Common-average crossovers were 310 µm deeper than layer 4. An atlas rule without LFP predicted layer-4 depth with a 20 µm median error; a calibrated crossover gave 54 µm. The crossover tracked layer-4 depth only through cortical thickness. The flash CSD early sink was 90 µm above the layer-4 centre and about 300 µm above the crossover. Flash stimulation and running moved the crossover by only 5 and 20 µm. Limitation: CCF layers come from track registration, not local histology, and the CSD sink and CCF layer 4 disagree by about 90 µm. Our algorithm is not the published FLIP code. The result applies to head-fixed mouse visual cortex only.

local field potential · spectrolaminar motif · cortical layers · layer 4 · current source density · Neuropixels · Allen Brain Observatory · mouse visual cortex · CCFv3 · volume conduction · FLIP · laminar identification