Recordings of neuronal activity from the CA1, DS, and OFC during the temporal bisection task.

(A) Schematic of the temporal bisection task. Rats run on the treadmill for long (10 s) or short (5 s) time intervals. They then have to choose the left or right arm of the Y-maze, which are associated with long- or short-time intervals, respectively. A water drop is provided as a reward at the end of the correct arm. Test trials (7.07 s) are randomly inserted between the discrimination trials of short and long intervals. In the test trials, the rats are also required to choose the left or right arm of the Y-maze after forced running during the interval period, but no reward is provided in either arm. (B) Behavioral performance across sessions in the temporal bisection task. Performance of every five trials was averaged across sessions for all animals (black lines, mean ± S.E.M.; n = 31 sessions) or individual animals (colored lines, mean). (C) Average ratio of selected the arm associated with long-interval trials in the test trials in all sessions. The error bar indicates 95% confidence intervals (t-test). The ratio of each session is indicated by a blue dot. (D) Recording sites with silicon probes in the CA1, DS, and OFC of rat s71. Arrowheads, the traces of the probes on green fluorescent Nissl-stained sections. Scale bar: 0.5 mm.

Representation of elapsed time in CA1, DS, and OFC neurons in the temporal bisection task.

(A, a-i) Spiking activity of nine representative time cells in the neurons in the CA1 (a-c), DS (d-f), and OFC (g-i). Top, raster plots of time cells in long and short interval trials. Bottom, PETHs in long (solid line) and short (dotted line) interval trials. (B) Spiking activity of time cells in the CA1 (left), DS (middle), and OFC (right) (n = 298, 282, and 208 units from 3 rats, respectively). Each row represents the normalized PETHs of individual neurons in long-interval trials. The color scale shows the firing rate of each neuron (red represents the maximum rate of each neuron, and blue represents 0 Hz). Neurons are ordered according to the time of their peak firing rates. (C) Box plot of information rates of time cells in the CA1, DS, and OFC (n = 298, 282, and 208 units from 3 rats, respectively). (D) Box plots of standard deviation (SD) of Gaussian kernels fitted to PETHs of time cells in the CA1, DS, and OFC. (E) Box plots of maximum firing rates of PETHs of time cells in the CA1, DS, and OFC. *P < 0.05; **P < 0.01; ns, not significant. One-way ANOVA followed by Tukey-Kramer post-test.

Theta modulations of LFPs in CA1, DS, and OFC during the temporal bisection task.

(A) Time-resolved wavelet power spectrum of LFP in CA1, DS, and OFC during the interval period in a single rat (s82; mean of all sessions; n = 12 sessions). (B) Wavelet power spectrum of LFP in CA1, DS, and OFC during the interval period. Mean (solid line) ± SD (dashed line). n = 31 sessions from 3 rats. (C) Time-resolved wavelet coherence of LFPs between CA1 and DS, CA1 and OFC, and DS and OFC during the interval period in one rat (s82; mean of all sessions; n = 12 sessions). (D) Wavelet coherence of LFPs between CA1 and DS, CA1 and OFC, and DS and OFC during the interval period. Mean (solid line) ± SD (dashed line). n = 31 sessions from 3 rats.

Entrainment of firing activity of time cells by theta oscillation.

(A, a to i) Firing rates and theta phases of nine representative time cells in CA1, DS, and OFC. Top: PETHs of long (solid line) and short (dashed line) interval trials. Middle: Plots pf theta phase as a function of elapsed time. Each dot represents an action potential from the long (light color) and short (dark color) interval trials. Bottom: Histograms of theta phase plots as a function of elapsed time. (B) Ratios of time cells in CA1, DS, and OFC that are significantly phase modulated by hippocampal theta oscillations (P < 0.01, Rayleigh test). Clopper-Pearson confidence intervals for 99% are also shown. (C) Ratios of time cells in CA1, DS, and OFC that have significant time-phase correlations (P < 0.01, statistics of linear-circular correlation). Clopper-Pearson confidence intervals for 99% are also shown. (D) Distribution of preferred theta phases for significantly modulated (P < 0.01) time cells in CA1, DS, and OFC.

Synchronous activity of time cells within and across the CA1, DS, and OFC.

(A, a–g) Synchronous activity of six representative time-cell pairs within and across the CA1, DS, and OFC. Left: PETHs of long (solid line) and short (dashed line) interval trials of the pairs of time cells. Middle: Cross-correlograms (CCGs) of the pairs of the time cells (jitter methods). Right: CCGs of neuronal pairs as a function of theta phase. (B) CCGs of time-cell pairs showing significant peaks within and across the CA1, DS, and OFC. The CCGs are z-scored in the plots. (C) Ratios of time-cell pairs with significant peaks within and across the CA1, DS, and OFC (jitter methods). Clopper-Pearson confidence intervals for 99% are also shown. (D) Ratios of time-cell pairs that are significantly modulated by theta oscillations within and across the CA1, DS, and OFC (Rayleigh test). Clopper-Pearson confidence intervals for 99% are also shown. (E) Ratios of time-cell pairs that have significant peaks within and across the CA1, DS, and OFC (jitter methods) during running on the treadmill only (time only; bule), running in the Y-maze only (Y-maze only, orange), and both treadmill and Y-maze (both time & Y-maze; green).

Activity of time cells in CA1, DS, and OFC reflects the decisions of rats.

(A) Representative data of Bayesian decoding analysis. Top, simultaneously recorded spiking activity of time cells (11, 22, and 8 units from CA1, DS, and OFC, respectively) in a single normal trial of the temporal bisection task. Bottom, the probability density of temporal information is computed from the simultaneously recorded activity of time cells in the trial. (B, a-d) Decoded time in the selected-long and -short test trials per 1 s interval duration using spiking activity in CA1, DS, and OFC (a), only CA1 (b), only DS (c), and only OFC (d). Mean ± S.E.M. ##P < 0.01 for effects of trial type. *P < 0.05, **P < 0.01 for simple effects of trial type at each period. Two-way ANOVA with Bonferroni post-test. (C) Mean differences in decoded time across CA1, DS, and OFC in each trial. *P < 0.05, ***P < 0.001, t-test compared to trial-shuffled data.