OB performances in the observation box and in the maze depended on ACC.

(A) Behavioral apparatus consisting of an observation box (top), a continuous T-maze and a rest box (bottom). Red: reward site (water port); arrow: running direction. OB/Demo: observer/demonstrator rat.

(B) Coronal brain sections after NMDA and vehicle infusion. Arrow: lesion in ACC. Number: anteroposterior coordinate from the Bregma.

(C) Average (mean ± SEM) poke performance curves of OBs in the box for the 3 sessions before (Before) and the first 3 testing sessions 2 weeks after NMDA or vehicle infusion (After).

(D) Maze performances of OBs in each of Before and After sessions. Thin/thick line: individual rat/average performance.

***: P < 0.001, **: P < 0.01, *: P < 0.05, n.s.: not significant (same for all figures).

ACC selective cells were activated during delay periods in the observation box.

(A) Spike raster of an ACC left- (top) and an ACC right-selective (bottom) cell on the left (magenta) and right (blue) sides of the T-maze, and their average firing rates along the (linearized) trajectories on the two sides (right curves). Each tick is a spike. Each row is a trial. Color markers show maze landmarks: choice point (CP), reward site (R), end of side-arm (End). The two cells were from two different sessions. Error trials (top: 2 left and 4 right; bottom: 0) are excluded.

(B) Similar to (A), but for the same two cells during the 2-s delay period of each trial in the box (dash line: time 0 – first rewarded poke in the box). Note the higher rate on the same side of each cell’s selectivity.

(C) Trial-by-trial firing rates of the same two cells in (A-B) during running on the left/right side of the maze and during delay period on the left/right side of the box. Note the similar selection indices (SIs, numbers) in the maze and in the box.

(D) SIs in the maze and in the box for all ACC selective cells under Demo, Object and Empty. Each dot is a cell. Black line: liner regression between maze and box SIs. Colored dots: cells with significant same-side selectivity in the box on the left (magenta) or right (blue).

(E) Percentages of ACC cells with significant same-side selectivity in the box (left/right combined, red lines) under Demo, Object and Empty, compared to random distributions (black) obtained by shuffling each cell’s firing rates among all delay periods of left/right trials.

See also Figure S1, S2.

ACC cells with same-side selectivity in the box were less activated during delay periods of error trials.

(A) Left: illustration of a correct and an error left trial (the Demo chose left in the maze). A trial was considered an error if the OB chose the opposite (right) side in the maze, after getting rewarded (on the left side) in the box. Red: reward site. Right: trial-by-trial spike raster (top) of an ACC left-selective cell (in both the maze and box) and firing rates of this ACC cell in each correct/error trial (bottom) during delay periods of an example session. Each tick represents a spike. Magenta/blue: spikes in correct/error trials. Red dash line: OB’s poke time (time 0). Number: firing rate difference index (DI) between correct and error trials for this cell.

(B) Similar to (A), but for an ACC right-selective cell in correct and error right trials. Blue/magenta: correct/error trials.

(C) Comparison of ACC cell DIs described in (A-B) during delay periods on their preferred side under Demo, Object and Empty.

(D) Percentages of ACC cells with same-side selectivity in the box that had significantly higher firing rate during delay periods of correct than error trials on their preferred side (left-selective and right-selective cells combined, red lines) under Demo, Object and Empty, compared to their shuffle-generated distributions (black).

See also Figure S3.

Activities of ACC cells were correlated with those of CA1 same-side ensembles during delay periods in the box.

(A) Trial-by-trial spike raster of an example ACC right-selective cell and its associated CA1 left and right ensembles during delay periods of left (magenta) and right (blue) trials of a session. Each tick is a spike. Red dash line: time of first rewarded poke in the box (time 0).

(B) Firing rate cross-correlation between the example ACC cell and its associated CA1 ensemble in (A). Red/black curves: average values over all trials in the session. Shaded areas: SEM. Note the peak correlation around time lag 0 (dash line) between the ACC cell and its same-side CA1 ensemble.

(C) Average correlations (mean ± SEM, at time lag 0) during delay periods between ACC cells and their CA1 same- (red) and opposite-side (black) ensembles under Demo, Object and Empty. Note the significant same-side correlation under Demo, but not under Object or Empty.

(D) Similar to (C), but for correlations during running laps in the maze. Note the significant same-side correlation under all three conditions, as expected.

See also Figure S4.

Activities of ACC cells were correlated with those of CA1 same-side ensembles during water consumption in the box.

(A) Spike raster of an example ACC right-selective cell during water consumption in the box for a left and a right trial of a session. Each tick is a spike.

(B) CA1 LFPs filtered within the ripple band (top) and spike raster of CA1 cells in the left (middle) and right ensemble (bottom) during the same time periods in (A). Arrows: SWRs. Note the population bursts associated with SWRs.

(C) Correlation (r) between trial-by-trial firing rates of the example ACC cell and its CA1 same- and opposite-side ensemble within SWRs during water consumption in the session. Each dot is a trial.

(D) Average correlations (mean ± SEM) for all ACC selective cells and their CA1 same- and opposite-side ensembles, as computed in (C), under Demo, Object and Empty.

See also Figure S5.

Activation of ACC selective cells during delay periods was associated with subsequent activation of CA1 same-side ensembles within SWRs during water consumption in the box.

(A) Spike raster of an ACC right-selective cell during delay periods of left trials in an example session. Each tick is a spike. Red dash line: OB’s first rewarded poke in the box (time 0).

(B) Spike raster of CA1 left and right ensembles within SWRs during water consumption periods of the same (left) trials in (A). Only spikes within SWRs of a 1 s window are shown for clarity.

(C-D) Similar to (A-B), but for the same ACC (right-selective) during delay periods and same CA1 ensembles during water consumption periods in the right trials of the same session.

(E) Correlation (r) between trial-by-trial firing rates of the ACC cell during delay periods and its CA1 same- or opposite-side ensembles within SWRs during water consumption in the session. Each dot represents a trial.

(F) Average correlations (mean ± SEM) for all ACC selective cells and their CA1 same- and opposite-side ensembles, as computed in (E), under Demo, Object and Empty.

See also Figure S6.

Activation of ACC selective cells during delay periods was associated with the replay of CA1 same-side ensembles during water consumption in the box.

(A) Spike raster (middle) of an example forward/reverse replay of the CA1 template cells corresponding to the right outbound/inbound trajectory (top). The Bayesian-decoded probability of trajectory positions at each time bin is shown at the bottom. x: decoded (peak probability) positions. Dashed line: replay order of spikes.

(B) Trial-by-trial correlation (r) between an example ACC cell’s firing rate during delay periods and the number of replays for CA1 templates on the same- or opposite-side inbound trajectories of the ACC cell’s selectivity. Each dot represents one trial.

(C) Average correlation values (mean ± SEM) for all ACC selective cells and the corresponding replay numbers for same- or opposite-side templates, as computed in (B), under Demo, Object and Empty. Note the higher same-side correlation under Demo, but not under Object or Empty.

(D) Similar to (C), but for outbound trajectories.

Behavior of OBs during delay periods in the observation box, related to Figure 2.

(A) Illustration of the delay time window of a trial: time interval between the Demo passing the T-maze choice point (CP) on a left (magenta) or right (blue) trajectory and the OB’s first poke with reward at the corresponding water port in the box (not drawn proportionally).

(B) Distribution of the delay window duration for all trials of all recorded OBs under the Demo condition. Note the peak at 1.97 s.

(C) Top: an example OB’s head direction within the 2 s (Start) delay period before first rewarded poke (End) in the box for a left and right trial. Head direction was computed in every 0.1 s time bin. Downward arrow shows the direction (270°) facing the T-maze.

Bottom: distribution of the OB’s head direction in the 2 s delay periods of all trials in a session with left/right rewarded pokes. Number inside circle: radius length (count of time bins). Red rectangles: reward sites (at ∼210°, 330°).

(D) Similar to (C), but for the OB’s head position in the same left and right trial (top) and head position density during delay periods of all trials in the session. Bottom color bar: count of time bins. Gray trace: all head positions in the session.

ACC cells with opposite-side selectivity in the box were fewer than the chance level under Demo, related to Figure 2.

(A) Numbers of ACC selective cells (white) and corresponding percentages among all recorded ACC cells under Demo, Object and Empty.

(B) SIs in the maze and in the box for all ACC selective cells under Demo, Object, and Empty. The plots are the same as in Figure 2D, but now the colored dots highlight those left- (magenta) or right- (blue) selective cells in the maze that had significant opposite- side selectivity in the box.

(C) Percentages of ACC selective cells with significant opposite-side selectivity in the box (red lines) under Demo, Object and Empty, compared to their random distributions (black) obtained by shuffling each cell’s firing rates among all delay periods of left and right trials in the box. Demo: 14%, Z = -4.1, P = 2.1 x 10-5, Z-test; Object: 27%, Z = 0.47, P = 0.32; Empty: 22%, Z = -1.2, P = 0.89.

ACC cells with same-side selectivity in the box did not differ in firing rate during delay periods between correct and error trials on their non-preferred side, related to Figure 3.

(A) Average DIs (mean ± SEM) of ACC cells with same-side selectivity in the box on their non-selective side between delay periods of correct and error trials under Demo, Object and Empty. The DIs were not significantly different from 0 (Demo: -0.051 ± 0.030, N = 44, P = 0.11, two-sided t-test; Object: 0.032 ± 0.021, N = 84, P = 0.13; Empty: -0.026± 0.016, N = 124, P = 0.11). There was a significant difference among the 3 conditions (One-way ANOVA: F(1,251) = 3.54, P = 0.030).

(B) Percentages of ACC cells with same-side selectivity in the box that had significantly higher rate during delay periods in correct trials on their non-preferred side (red, left-selective and right-selective cells combined) under Demo, Object and Empty, compared to shuffle-generated distributions (black). The percentage was not significant from the chance level under Demo (2.3%, Z = -0.77, P = 0.78, Z-test), Object (4.8%, Z =-0.0062, P = 0.50) or Empty (2.4%, Z = -1.2, P = 0.88).

CA1 cells with same-side selectivity in the box were not significantly different from the chance level, related to Figure 4.

(A) Firing rates of two example CA1 cells during running in each trial on the left/right side of the maze and during each delay period in the box. Note different SIs (numbers) between the maze and box. (Left-selective cell: Maze: P = 3.3 x 10-4, two-sided t-test; Box: P = 0.11; Right-selective cell: Maze: P = 0.0022; Box: P = 0.15)

(B) SIs in the maze and in the box for all CA1 cells active in the maze under Demo, Object and Empty. Colored dots: cells with significant same-side selectivity in the box on the left (magenta) or right (blue). Black line: linear regression between the maze and box SIs. There was no significant correlation under any of the conditions (Demo: R = -0.10, P= 0.93, Pearson’s r; Object: R = -0.025, P = 0.59; Empty: R = 0.061, P = 0.24).

(C) Percentages of CA1 cells with significant same-side selectivity in the box (left/right combined, red lines) under Demo, Object and Empty, compared to their distributions (black) obtained by random shuffling of each cell’s firing rates among all delay periods of left/right trials. The actual percentage was not significantly different from the random distribution under any of the conditions (Demo: 23%, Z = -0.54, P = 0.70, Z-test; Object: 22%, Z = -1.3, P = 0.91; Empty: 24%, Z = -0.38, P = 0.65).

Activities of ACC selective cells and CA1 ensembles during water consumption in the box differed between same- and opposite-side of their selectivity, related to Figure 5.

(A) Trial-by-trial firing rates of a left-selective ACC cell during water consumption in the box on the same- and opposite-side of its selectivity in an example session. Each dot is a trial. Horizontal lines: mean values. Number: firing rate difference index (DI) between same- and opposite-side for this ACC cell.

(B) Average firing rate DIs over all ACC selective cells during water consumption under Demo, Object and Empty. The DIs were significantly higher than 0 under Demo (0.042 ± 0.013, N = 200, P = 9.1 x 10-4, two-sided t-test), but not under Object (0.0053 ±0.018, N = 99, P = 0.76) or Empty (-0.015 ± 0.018, N = 130, P = 0.41). There was a significant difference among the 3 conditions (One-way ANOVA test: F(1,427) = 4.0, P = 0.019).

(C-D) Similar to (A-B), but for firing rates of CA1 ensembles on the same- and opposite-side of ensembles. The average DIs in (D) were significantly higher than 0 under Demo (0.020 ± 0.0053, N = 19, P = 1.3 x 10-3, two-sided t-test), but not under Object (- 0.011 ± 0.012, N = 11, P = 0.41) or Empty (0.0048 ± 0.0074, N = 13, P = 0.53). There was a significant difference among the 3 conditions (One-way ANOVA test: F(1,42) = 3.9, P = 0.028).

Activities of ACC selective cells during delay periods were correlated with their activities during water consumption periods in the box, related to Figure 6.

(A) Trial-by-trial correlation (r) between the firing rates of an example ACC cell during delay periods and during water consumption in all trials of a session. Each dot represents a trial.

(B) Average correlation values (mean ± SEM) for all ACC selective cells under Demo, Object and Empty. The correlations were significant under all the 3 conditions (Demo:0.26 ± 0.019, N = 200, P = 5.7 x 10-32, two-sided t-test compared to 0; Object: 0.25± 0.028, N = 99, P = 2.1 x 10-14; Empty: 0.24 ± 0.023, N = 130, P = 2.1 x 10-19). There was no significant difference among the 3 conditions (One-way ANOVA: F(1,428) = 0.32, P = 0.73). The result indicates that the high (or low) rate of an ACC selective cell during the delay period in a trial was accompanied with its high (or low) rate during the water consumption of the same trial.