Fig. 7: The activity of hippocampal time cells and cell assemblies coded for left-turn vs right-turn trials after exiting the delay zone, but not during the delay.

a Left, schematic of defined maze regions. Right, proportion of turn-selective cells or cell assemblies in the delay zone. No significant turn selectivity was observed during the delay period (cells: on/10 s, p = 0.50; on/30 s, p = 0.37; off/10 s, p = 0.50; off/30-s, p = 0.37, n = 196 cells; assemblies: on/10 s, p = 0.36, n = 61; on/30 s, p = 0.40, n = 64; off/10 s, p = 0.10, n = 95; off/30 s, p = 0.08, n = 105 assemblies; chance level = 5%, blue dotted line, binomial tests) except when the analysis segment was expanded to include the entry into the delay zone (i.e., with the head, but not the body inside the delay zone; inset: cells, on/10 s, p = 0.002; on/30 s, p = 0.15; off/10 s, p = 4.1 x 10^-5; off/30 s, p = 0.055, n = 206 cells, binomial tests). b Proportion of turn-selective cells (p < 2.3 x 10^-308, n = 148 cells for all treadmill/delay conditions, binomial tests) and turn-selective assemblies (on/10 s, p = 2.7 × 10^-6, n = 10; on/30-s, p = 2.3 × 10^-6, n = 15; off/10-s, p = 1.0 × 10^-17, n = 28; off/30-s, p < 2.3 × 10^-308, n = 32 assemblies; chance level = 5%, blue dotted line, binomial tests) in the T zone. c Two example population events detected in the reward area and during the delay in the treadmill-off condition. Both CA1 population events (spike rasters with yellow background) occurred concurrently with SWRs (CA1 LFP trace on top of spike rasters, filtered at 150–250 Hz). d Overlap between population events and SWRs. Left: proportion of identified CA1 population events that were associated with SWR (treadmill-off delay: 54.5% ± 4.25; reward area: 52.6% ± 5.67, n = 12 sessions). Right: the proportion of detected SWRs in the CA1 LFP associated with CA1 population events (treadmill-off delay: 29.5% ± 4.64; reward area: 48.8% ± 5.76, n = 12 sessions). e Top, example SWR that did not co-occur with a population event. Bottom, example population event that did not co-occur with a SWR. f Population event-associated firing rates were significantly turn-selective in the reward area (left; on/10-s, p = 4.6 x 10^-13, n = 86; off/10-s, p < 2.3 × 10^-308, n = 139; on/30 s, p < 2.3 × 10^-308, n = 92; off/30-s, p < 2.3 × 10^-308, n = 86 population events, binomial tests), but not during delay intervals (right; off/10 s, p = 0.090, n = 92; off/30-s, p = 0.064, n = 194 population events; chance = 5%, blue dotted line, binomial tests). Note that only treadmill-off, but not treadmill-on trials had a sufficient number of population events during the delay for analysis of turn selectivity. g CA1 cell assemblies, but not single CA1 cells were significantly turn-selective when rats traversed in the maze stem. Top left, average head position along the maze stem (5 cm bins, aligned to the running direction) during left- and right-turn trials (mean ± SD, n = 18 sessions). Top right, proportion of turn-selective cells per spatial bin along the stem (10 cm bins; 0 cm, p = 0.24, n = 79; 10 cm, p = 1.0, n = 88; 20 cm: p = 0.18, n = 88; 30 cm, p = 0.07, n = 85; 40 cm, p = 0.50, n = 93; 50 cm, p = 0.12, n = 99; 60 cm, p = 0.0001, n = 15; 70 cm, p = 1.8 x 10^-10, n = 119; 80 cm, p < 2.3 × 10^-308, n = 114, binomial tests). Bottom, significant turn selectivity was observed specifically for cell assemblies in treadmill-off trials (cells: on/10 s, p = 0.51; on/30 s, p = 0.11; off/10-s, p = 0.11; off/30 s, p = 0.27, n = 211; assemblies: on/10 s: p = 0.34, n = 24; on/30-s: p = 0.12, n = 24; off/10 s: p = 0.002, n = 27; off/30-s: p = 1.2 × 10^-6, n = 30; chance = 5%, blue dotted line). All statistical tests are two-sided without post-hoc adjustments for multiple comparisons. Box plots: central line, edges, whiskers and plus signs indicate median, the 25th/75th percentile, maximum/minimum and outliers. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.