a) Population average of across-session angular difference in preferred direction, relative to the last session for all animals not in Fig. 1; two-tailed p < 0.001 for all animals; n = 16, 18, 16, 16 sessions respectively; median absolute deviations from 0° = 15.9, 7.0, 5.9, 21.3). b) Right: V-test statistic for the population average of across-session angular difference in preferred direction relative to the last session as shown in a) and Fig. 1. V statistic per animal = 14.1, 17.1, 17.3, 14.6, 12.2. Median of V statistic shuffles per animal = 3.0, 3.2, 5.7, 4.8, 2.0. Left: V-test statistic for pooled animals (75.2) versus the distribution of V statistic for within-day shuffled cell identity (median = 18.6). c) Distribution of population-wise angular difference relative to the previous session for all five animals (V-test, not different from 0°, two-tailed p < 0.001, n = 75 sessions). Mean ± standard deviation per animal: 0.4° ± 34.4°; 1.8° ± 24.1°; −0.7° ± 10.8°; 5.1° ± 22.8°; −2.7° ± 33.7° d) same as b) but for across-session difference in preferred direction relative to the previous session. V statistic per animal = 12.5, 15.6, 16.7, 12.9, 10.1. Median of V statistic shuffles per animal = 3.3, 3.5, 6.7, 5.4, 2.7. V statistic for pooled animals = 69.9. Shuffled V statistic median for pooled animals = 21.9. e) Mean variance of absolute change in preferred direction versus number of recordings included, from the first to the last day of the experiment for all animals. Error denotes SEM. f) Absolute change in overall preferred direction, per session, with respect to the last day of the experiment in Fig. 1. Median absolute change per animal in degrees: 15.9, 5.1, 7.9, 26.3, 24.0. One-tailed Wilcoxon signed rank test between each animal and its shuffled cell identity distribution: p = 3.052 × 10−5, 3.81 × 10−6, 3.81 × 10−6, 7.63 × 10−5, 1.53 × 10−4, W statistic = 1.0, 0.0, 0.0, 3.0, 5.0 respectively. g) Left, Difference in angular offset across all HD cell pairs and all recording pairs in the recording protocol, shown for individual animals relative to control datasets of shuffled cell pair identities. n = 25,901 pairs. Filled portion indicates ± 2 s.d. of the shuffled distribution. Right, Distribution of difference in angular offset per animal versus the distribution of medians for their respective shuffled cell pair identity. Median change ± s.d. in angular offset per animal = 12.0° ± 27.6, 9.0° ± 24.1, 12.0° ± 27.1, 24.0° ± 40.9, 12.0° ± 31.1. Median of shuffled median distribution = 51.0°, 54.0°, 54.0°, 54.0°, 54.0°. The median difference in angular offset was compared to the respective shuffled median distribution of each animal. ‘***’ indicates the median value was lower than 1000 shuffled median values (p < 0.001). h) Example population vector correlation between sessions 13 and 17 of Animal 3. Population activity is binned in 12 HD bins. i) Recording-to-recording average population vector correlation per animal. j) Population vector correlation as a function of time, for each animal. Dashed line denotes the mean of all animals.