A manifold neural population code for space in hippocampal coactivity dynamics independent of place fields

Hippocampus is comprised of [~]20% place cells, discharging in cell-specific locations ("place fields"), standardly interpreted as a dedicated neuronal code for space. However, place cell discharge is temporally unreliable across seconds and days, and place fields are multimodal, suggesting an alternative "ensemble cofiring" spatial code with manifold dynamics that does not require reliable spatial tuning. We evaluated these hypotheses using GCaMP6f and miniature microscopes to image mouse CA1 ensemble activity in two environments, across 3 weeks. Both place fields and ensemble coactivity relationships appear to "remap," being distinct between, and (weakly) similar within environments. Decoding location as well as environment from 1-s ensemble location-specific discharge is effective and improves with experience. Decoding the environment (but not location) from cell-pair coactivity relationships is also effective and improves with experience, even after removing place tuning. Discriminating environments from 1-s ensemble coactivity relies crucially on the cells with the most anti-cofiring cell-pair relationships because ensemble activity is internally-organized on a low-dimensional manifold of non-linear cofiring relationships that intermittently reregisters to environments according to the anti-cofiring subpopulation activity.