Multidirectional cells (MDCs) exhibit stable tuning across perceptual changes and adapt to environmental complexity

A-B. Left: Animals successively explored two environments composed of two connected rooms that were either visually and tactilely identical (S1–2id: same visual cue and floor) or distinct (S2–2diff: different visual cues and floors). Right: Polar plots of two example BDCs (BDC1 and BDC2) recorded in both conditions. Peak firing rates are indicated for each cell. Note the emergence of secondary peaks along the 180° axis, regardless of the perceptual similarity between rooms. C. Directional activity of 30 BDCs recorded in S1–2id (top); among them, 26 were also recorded in S2–2diff (bottom). Activity is color-coded from blue (no activity) to orange (peak firing rate), and cells are sorted by their PFD. D. Comparison of BDC properties between S1–2id and S2–2diff. Left: Flip scores, quantifying bidirectionality is performed by computing the difference at 180° vs. 90° of the tuning curve autocorrelation. Right: Peak firing rates. No significant differences were observed between environments. Bars represent mean ± SD; individual data points are shown as black dots. Statistical comparisons were performed using unpaired t-tests. E-F. Left: Animals successively explored two environments composed of either 2 (S1–2diff) or 4 (S2–4diff) visually and tactilely distinct connected rooms Right: Polar plots of two example BDCs (BDC3 and BDC4) recorded in both conditions. Note the emergence of multiple peaks in the four-room condition. G. Directional activity of 42 BDCs recorded in S1– 2diff (top); 32 of these were also recorded in S2–4diff (bottom). Color scale and sorting as in panel C. H. Comparison of BDC properties between S1–2diff and S2–4diff. Left: Flip scores, computed as in D, significantly decreased in the four-room condition, indicating a loss of bidirectionality. Right: Peak firing rates did not change significantly between environments. Bars represent mean ± SD individual data points are shown as black dots. Statistical comparisons were performed using unpaired t-tests. ***p < 0.0001.

Inter-room tuning shifts of multidirectional cells (MDCs) reflect environmental geometry

A-B. Animals successively explored two environments composed of two connected rooms that were either visually and tactilely identical (S1-2id) or distinct (S2-2diff). Polar plots of two example BDCs (A: BDC5; B: BDC6) are shown, plotted separately for each room. Both cells exhibited unipolar activity within individual rooms, regardless of perceptual similarity. C. Directional activity for each room of 30 BDCs recorded in S1-2id (top); 26 of these were also recorded in S2-2diff (bottom). All cells displayed unipolar tuning in each room. D. Polar histograms showing angular differences in PFD between rooms in S1-2id (top) and S2-2diff (bottom). Distributions cluster significantly around 180°, indicating opposite tuning directions across rooms. V: V-test; Dir: mean PFD offset between rooms ± SD. ***p < 0.0001. E-F. Animals successively explored two environments composed of either 2 (S1–2diff) or 4 (S2–4diff) visually and tactilely distinct connected rooms. Polar plots of two example BDCs (E: BDC7; F: BDC8), shown separately for each room. Both cells exhibited unipolar activity within individual rooms, even when the global pattern appeared multidirectional. G. Directional activity for each room of 42 BDCs recorded in S1-2diff (top); 32 of these were also recorded in S2-4diff (bottom). All cells showed unipolar tuning within each room. H. Polar histograms showing angular differences in PFD between rooms for S1-2diff (top) and S2-4diff (bottom). Distributions cluster significantly around 180° in the two-room condition and around 90° in the four-room condition, consistent with the spatial layout. Peak frequencies are indicated on each polar plot. V: V-test; Dir: mean PFD offset between rooms ± SD. ***p < 0.0001.

Head direction cells maintain stable directional tuning across perceptual and structural changes.

A-B. Left: Animals successively explored two environments composed of either visually identical (S1–2id) or visually and tactilely distinct (S2–2diff) connected rooms. Right: Polar plots of two example HDCs (HDC1 and HDC2) recorded in both environments. Peak firing rates are indicated for each cell. C. Directional activity of the 77 HDCs recorded in S1–2id (top); among them, 40 were also recorded in S2–2diff (bottom). Activity is color-coded from blue (no activity) to orange (peak firing rate), and cells are sorted by preferred firing direction (PFD).D. Comparison of HDC properties between S1–2id and S2–2diff. Left: Mean vector length. Right: Peak firing rate. No significant differences were observed between environments. Bars represent mean ± SD; individual data points are shown as black dots. Statistical comparisons were performed using unpaired t-tests. E-F. Left: Animals successively explored environments composed of either two (S1–2diff) or four (S2–4diff) visually and tactilely distinct connected rooms. Right: Polar plots of two example HDCs (HDC3 and HDC4) recorded in both conditions. G. Directional activity of 145 HDCs recorded in S1-2id (top); among them, 113 HDCs were also recorded in S2-4diff (bottom). Activity is color-coded as in panel C, and cells are sorted according to their PFD. H. Comparison of HDC properties between S1–2diff and S2–4diff. Left: Mean vector length. Right: Peak firing rate. No significant differences were observed between environments. Bars represent mean ± SD; individual data points are shown as black dots. Statistical comparisons were performed using unpaired t-tests. ***p < 0.0001

Head direction cell tuning remains consistent across room boundaries

A-B. Animals successively explored two environments composed of two connected rooms that were either visually and tactilely identical (S1-2id) or distinct (S2-2diff). Polar plots of two example HDCs (A:HDC5; B: HDC6) are shown separately for each room. Both cells exhibited the same unipolar activity in each room, regardless of perceptual similarity. C. Directional activity in each room of 77 HDCs recorded in S1-2id (top); 40 of these were also recorded in S2-2diff (bottom). All cells maintained unipolar tuning in each room. Activity is color-coded from blue (no firing) to orange (peak firing rate), and cells are sorted by PFD. D. Polar histograms showing angular differences in PFD between rooms in S1–2id (top) and S2– 2diff (bottom). In both environments, PFDs cluster significantly around 0°, indicating stable directional tuning across rooms. V: V-test; Dir: mean PFD offset between rooms ± SD. ***p < 0.0001. E-F. Animals successively explored environments composed of either two (S1– 2diff) or four (S2–4diff) visually and tactilely distinct connected rooms. Polar plots of two example HDCs (E: HDC7; F: HDC8) are shown separately for each room. Both cells maintained consistent unipolar activity across all rooms, regardless of the number of connected rooms. G. Directional activity in each room of 145 HDCs recorded in S1-2diff (top); 113 of these were also recorded in S2-4diff (bottom). All cells exhibited consistent unipolar tuning in each room. H. Polar histograms showing angular differences in PFD between rooms in S1–2diff (top) and S2–4diff (bottom). In both cases, distributions cluster significantly around 0°, confirming that HDCs preserved consistent tuning across rooms. Peak frequencies are indicated. V: V-test; Dir: mean PFD offset between rooms ± SD. ***p < 0.0001.

Non-directional RSC cells exhibit room-specific spatial patterns aligned with environmental geometry

A. Schematic of three hypothetical spatial firing patterns (top) and the corresponding spatial correlations (bottom) in two-room environments. B. Two examples of spatial ratemaps from non-directional RSC neurons recorded in two connected rooms that were either visually and tactilely identical (S1–2id) or distinct (S2–2diff). Both cells show oppositely repeated spatial patterns across rooms. C. Comparison of spatial correlations between unrotated and rotated ratemaps in S1–2id and S2–2diff. Rotated maps show significantly higher correlations than unrotated maps, regardless of room similarity. Dots represent individual data points; bars indicate mean ± SEM. Statistical comparisons were performed using a mixed ANOVA followed by Tukey’s post hoc test. D. Schematic of three hypothetical spatial firing patterns (top) and the corresponding predicted correlations (bottom) for unrotated and rotated comparisons between adjacent and opposite rooms. E. Two examples of spatial ratemaps from non-directional RSC neurons recorded in S1–2diff and S2–4diff environments. Cells exhibit oppositely aligned patterns in S1–2diff and orthogonally aligned patterns in S2–4diff. F. Comparison of spatial correlations between unrotated and rotated ratemaps in S1–2diff and S2–4diff. Rotated maps yield significantly higher correlations in both environments. Dots represent individual data points; bars indicate mean ± SEM. Statistical comparisons were performed using a mixed ANOVA followed by Tukey’s post hoc test. G. Top: Schematic representation of adjacent (black arrows) and opposite (green arrows) room comparisons in the 4-room environment. Bottom: Mean spatial correlations between adjacent and opposite room pairs. Unrotated maps show negative correlations between opposite rooms and near-zero correlations between adjacent rooms, whereas rotated maps yield positive correlations for both. Dots represent individual data points; bars indicate mean ± SEM. Statistical comparisons were performed using repeated-measures ANOVA followed by Fisher’s post hoc test. ***p < 0.0001.

Divergent spatial coding in RSC and hippocampus

A. Two examples of spatial ratemaps from non-directional RSC neurons recorded across four consecutive sessions alternating between two environments composed of two connected rooms that were either visually and tactilely identical (S1-2id, S3-2id) or distinct (S2-2diff, S4-2diff). Only RSC neurons with a high spatial information content (≥0.9) are preserved for this analysis. They show oppositely repeated spatial patterns that are preserved across sessions. B. Mean spatial correlation values for repeated environments (S1–2id × S3–2id and S2–2diff × S4–2diff). No significant differences were observed between the two conditions. Bars represent mean ± SEM. Statistical comparison was performed using a paired t-test. C. Mean spatial correlations across consecutive sessions in the full protocol (S1–2id × S2–2diff, S2–2diff × S3–2id, S3–2id × S4–2diff). Correlations remained positive and significantly higher than shuffled controls. Bars represent mean ± SEM. Statistical comparison was performed using one-way ANOVA. D. Two examples of spatial ratemaps from non-directional RSC neurons recorded across four successive sessions alternating between two environments composed of four connected rooms that were either visually and tactilely identical (S1–4id, S3–4id) or distinct (S2–4diff, S4–4diff). Cells exhibit orthogonally aligned patterns that remain consistent across sessions. E. Mean spatial correlation values for repeated environments (S1–4id × S3–4id and S2–4diff × S4–4diff). No significant differences were observed between the two conditions. Bars represent mean ± SEM. Statistical comparison was performed using a paired t-test. F. Mean spatial correlations across successive sessions in the 4-room protocol. Correlations remained positive and significantly higher than shuffled controls. Bars represent mean ± SEM; statistical comparison was performed using one-way ANOVA. G. Examples of hippocampal place cells recorded in S1–2id and S2–2diff. Cells show either repeated or remapped firing fields across rooms, with most exhibiting remapping. Opposite firing fields were never observed. H. Examples of hippocampal place cells recorded in S1–4id and S2–4diff. Place cells almost exclusively remap across rooms, with no evidence of orthogonally repeated fields. I. Left: Distribution of spatial correlation values between S1–2id and S2–2diff for all place cells, overlaid with kernel density estimates. The bimodal distribution indicates the coexistence of stable and unstable spatial activity across sessions. Right: Mean spatial correlation values between S1-2id and S2-2diff. Bars represent mean ± SEM. Statistical comparison was performed using a one sample t-test against 0. ***p < 0.0001. J. Same as in I, but for the four-room condition (S1–4id × S2–4diff). The distribution similarly reveals clusters of stable and unstable place cell activity. Bars represent mean ± SEM. Statistical comparison was performed using a one sample t-test against 0. ***p < 0.0001.