Extended Data Fig. 8: Principal component analysis of hippocampal population dimensionality for various levels of explained variance.

To corroborate the results we obtained on the dimensionality of principal cell (PC) population theta co-firing vectors in laser-off and laser on tests in the cNOR task (Fig. 6e), we repeated the principal component analysis (PCA) for different proportions of explained variance. In Fig. 6e we show the number of dimensions required to explain 80% of the variance, with each data point scaled by the number of PCs in each vector. (a-d) Panels a-d show the same analysis but for 70%, 75%, 85% and 90%, respectively. Only abDGC::ArchT (4-7wpi) mice exhibit reduced dimensionality during laser-on versus laser-off tests (p = 0.001, p = 0.001, p = 0.03, p = 0.001, respectively, two-sided paired permutation tests, n = 10, 6, and 8 sessions in 2 abDGC::GFP, 3 abDGC::ArchT (4-7wpi), and 2 abDGC::ArchT (9-12wpi) mice, respectively). (e) The corresponding absolute change in dimensionality (laser-on minus laser-off) for the three groups of mice at different proportions of explained variance as in a-d. In each case, abDGC::ArchT (4-7wpi), but not abDGC::ArchT (9-12wpi) were significantly lower than the abDGC::GFP group (p = 0.005, p = 0.02, p = 0.01, p = 0.01, respectively, two-sided permutation tests). The mean number of PCs in each vector was as follows: abDGC::GFP 22.7 ± 2.2; abDGC::ArchT (4-7wpi) 31.7 ± 5.1; abDGC::ArchT (9-12wpi) 27.0 ± 4.9. There were no group differences in the number of PCs in each vector (F(2,21) = 1.1, p = 0.4; one-way ANOVA). Panels a-e show Cumming estimation plots as described in Extended Data Fig. 1e. ***p < 0.001, **p < 0.01, *p < 0.05.