Fig. 3: Responses to novel and adapted stimuli and neural orthogonalization.

a, Distribution of neurons with d′ ≥ 0.3 between the leaf2 and circle1 corridors in task mice and unsupervised mice, either when new or after learning, as well as in fully naive mice. b, Summary of distribution changes in panel a across regions (n = 5 task mice and n = 6 unsupervised mice). c, Licking behaviour to leaf2 when new and after learning (n = 5 mice). d, Distribution of neurons with d′ ≥ 0.3 or d′ ≤ −0.3 between leaf1 and leaf2 in task mice and unsupervised mice after training with leaf2, as well as in fully naive mice. e, Summary of panel d across regions (n = 5 task mice; n = 6 unsupervised mice; n = 9 naive mice, 11 sessions; n = 3 unsupervised with gratings mice, 6 sessions). f,g, Example projections on the coding direction between leaf1 and circle1 in V1 (f) and the medial region (g). h, Similarity index (from Fig. 2i) for neural responses to leaf2 (same n as panel e). The bottom inset shows a schematic of the orthogonalization effect observed. Neural vectors are referenced with respect to the centre of the leaf1–circle1 axis. All data are mean ± s.e.m. *P < 0.05, **P < 0.01 and ***P < 0.001 from paired, two-sided t-tests, except panels e,h, which were independent two-sided t-tests.