Fig. 1: Plasticity in the visual cortex after supervised and unsupervised training.

a, Illustration of the virtual reality (VR) task with a sound cue at a random position in each corridor. Water was available after the sound cue in the rewarded corridor. b, Task training timeline in task mice. Mice in the unsupervised cohort experienced the same stimuli or the gratings stimuli without water rewards. c, Lick distribution in an example mouse after task learning. Trials were sorted according to the sound cue position. d, Performance quantification of anticipatory licks before water was delivered (error bars on all figures represent s.e.m.; the centre values are average; grey lines denote n = 5 mice). e, Example field of view of the mesoscope (left) and zoomed-in view (right) to illustrate cellular resolution. AL, anterolateral; LM, lateromedial; MMP, mediomedialposterior; RL, rostrolateral; RLL, rostrolaterallateral; RSP, retrosplenial. f, Selectivity index d′ of neural responses inside the two corridors. g, Single-trial responses of a single circle1-selective neuron as well as the entire population from an example mouse. h, Same as panel g but for leaf1-selective neurons. i, 2D histogram of selective neuron distributions across the field of view, aligned to a map of visual areas. The top and bottom rows are task mice and unsupervised mice, respectively. The left and right columns are before and after learning, respectively. A, anterior; L, lateral; M, medial; P, posterior. j, Percentage of neurons with high selectivity in each of the four visual regions defined in the inset (n = 4 task mice, n = 9 unsupervised mice and n = 3 unsupervised with gratings mice (5 sessions); statistical tests in all figures are two-sided Student’s t-tests, paired or independent as appropriate; see Methods for full details). All data are mean ± s.e.m. *P < 0.05, **P < 0.01 and ***P < 0.001 from paired, two-sided t-tests.