Supplementary Figure 3: The relationships between spatial receptive field overlap and spike-time correlations were also preserved for other MEC cell-type pairs.

(a) As with the grid cells analyzed for Fig. 5, which included both directionally sensitive grid cells (“conjunctive cells”; CJ) and non-directionally sensitive grid cells (“pure grid”; GC), pairs of pure grid cells (n = 182, GC/GC, top) and pairs containing both pure grid and conjunctive cells (n = 24, GC/CJ, bottom) show positive correlations between rate-map correlation coefficients (x-axis) and spike-time cross-correlation coefficients (y-axis). All rho values presented are Pearson’s r. Linear regressions for pairs of pure grid cells (GC/GC, top) indicated that spatial correlation coefficients significantly predicted spike-time cross-correlations for all three behavioral states (RUN: r = 0.754, p = 1.228e-34, R² = 0.568; REM: r = 0.481, p = 6.189e-12, R² = .232; NREM: r= 0.565, p = 9.662e-17, R² = 0.319). Regression results were similar for pairs of pure grid and conjunctive cells (GC/CJ) for RUN (r = 0.491, p = 0.015, R² = 0.241) and NREM (r = 0.471, p = 0.020, R² = 0.222) but failed to reach significance for REM (r = 0.239, p = 0.261, R² = 0.057), likely due to the low number of samples. (b) Spike-time correlations were also related to similarity of receptive fields in pairs of conjunctive and head-direction (HD) cells (n = 56, CJ/HD, top) and in pairs of head-direction cells (n = 103, HD/HD, bottom). Closeness in preferred firing angle, quantified as the relative angular difference (x-axis), negatively correlated with spike-time similarity (y-axis) across waking and sleep states such that cells which preferentially fired in similar directions were more likely to fire together during both waking and sleep states. For conjunctive cells paired with head direction cells (CJ/HD, top), linear regressions indicated that relative angular distance predicted spike-time cross-correlation coefficients during RUN (r = 0.488, p = 0.0001, R² = 0.238), REM (r = 0.364, p = 0.006, R² = 0.133), and NREM (r = 0.374, p = 0.004, R² = 0.140). Similarly, linear regressions for pairs of head direction cells (HD/HD, bottom) revealed that relative angular distance significantly predicted spike time cross-correlations during RUN (r = 0.516, p = 2.457e-8, R² = 0.266), REM (r = 0.228, p = 0.020, R² = 0.052), and NREM (r = 0.315, p = 0.001, R² = 0.099).