Fig. 3: Changes in hippocampal ensemble depiction of Day1Run1 linear track ends/corners after sphere-rearing.

a Group-average correlation matrix depicting neuronal ensemble mapping of track locations for cuboid (top-left), sphere-reared rats (top-right), and difference between groups (bottom). Green/red arrows: higher similarity of population vectors representing track ends in sphere-reared rats. b Higher population vector correlation of place maps between the two track ends in sphere- vs. cuboid-reared rats (p = 0.0018, two-sided rank-sum tests). Dashed lines: medians. c Neural manifold for track across 3 most informative principal component dimensions. d Schematic (top) and values (bottom-left) of population vector correlations between place maps at locations equidistant/symmetric from the two ends of the track (and track middle) vs. distance between the locations during Day1Run1 (Pearson’s correlation (R); best-fit regression line slope (ß)). Stars: significant differences between groups, between closest vs. furthest ___location, 0.24 vs. 0.88 m (right), between four adjacent spatial locations in middle vs. end (towards top). Dashed lines: group-averages at 0.24 m distance. Bottom-right: Correlations (top) and beta-coefficients (ß, slope) of best-fit regression line (bottom) for all individual animals and run directions (colored stars: comparison vs. 0). e Increased similarity of neuronal ensembles active at track ends/corners and its gradual decrease between symmetric track locations as a function of distance from ends in sphere-reared rats, reflecting early latent experiences in cuboidal (line) or spherical (ring) home cages. f (Top) Example individual run direction place rate maps with non-symmetric and symmetric firing in cuboid- (left) and sphere-reared rats (right; rate maps 3, 5, 6, 7 counting top to bottom exhibit some symmetry). (Bottom) Proportion rate maps well-tuned for linear space, exhibiting symmetric track-firing and low tuning for linear space. Symmetric track-firing and low tuning occurred more frequently in sphere-reared rats (p = 0.018 and p = 0.0015, respectively, two-sided Z-tests for 2 proportions). g Left-top two columns: Example Day1Run1 decoded trajectories during Post-Day1Run1 sleep frames in cuboid (left) and sphere-reared rats (right) illustrating confusion/swapping of decoded probabilities of track ends after sphere-rearing. Left-bottom: Reduced differences in cumulative significant frame proportions between cuboid and sphere-reared rats during Pre-/Post-Day1Run1 sleep on removal of 1–3 time bins (20 ms/bin) from beginning or end of non-significant sleep frames. Right-top: Method for determining time-space contingency. Right-center: Average time-space contingency matrices of decoded locations along the linear track; arrowheads: swapping of decoding of track ends in sphere-reared rats. Right-bottom: Specificity index of decoded track ends (p = 0.00050, p = 0.0025, two-sided t-tests). Data in d, g, are means ± SEM. N = 10 (5 rats/group; 2 directions/rat). ***p < 0.005. **p < 0.01. *p < 0.05. ns = not significant. Source data provided as Source Data file. Cartoons in Fig. 3/Panels d, e adapted from U. Farooq, G. Dragoi, Emergence of preconfigured and plastic time-compressed sequences in early postnatal development. Science 363, 168–173 (2019). DOI: 10.1126/science.aav0502. Reprinted with permission from AAAS.