Grid cells in opposite hemispheres of two rats kept their open-field share of toroidal phase coupling in REM and slow-wave sleep in three testable modules

Question: do grid cells in the left and right medial entorhinal cortex keep one shared toroidal state when there is no shared spatial input? Approach: we used Gardner et al. 2022 recordings from two rats with one Neuropixels probe per hemisphere (six bilateral grid modules). For each module and state (open field, wagon wheel, REM sleep, slow-wave sleep), we measured the slope of pair spike-count correlation (50 ms) on open-field grid-phase similarity, for pairs across hemispheres and within a hemisphere, with cell-bootstrap intervals. We also decoded the toroidal phase every 100 ms from each hemisphere separately. Result: in the three modules that we could test in sleep, cross-hemisphere phase coupling stayed positive in REM and slow-wave sleep (all 95% intervals above zero). The cross/within slope ratio was 0.53 to 0.79 in the open field and 0.56 to 0.81 in sleep. A gain-normalised ratio changed from open field to sleep by factors of 0.92 to 1.16; all six intervals include 1. Phases decoded from the two hemispheres agreed better than all 200 time-shift controls, with best agreement at zero lag. Limitations: two animals; in one rat most probe-1 cells are conjunctive cells, so cell class and hemisphere are confounded; one module was untestable because its probe-1 cells had no phase coupling among themselves, so the strict pre-specified rule failed; the data show a shared state, not the path (commissural axons, common input, or one bilateral attractor). The gain-normalised and state-invariance statistics are post hoc.

grid cells · medial entorhinal cortex · continuous attractor · interhemispheric coupling · sleep · REM · slow-wave sleep · toroidal topology · Neuropixels · structure-function