Fig. 5: Trial variability and decoding analysis during visual perception.
a A schematic of visual stimulation. Gratings in 8 directions are presented by a screen in a pseudo-random order. b Time-dependent mean trial variance during delivery of grating videos with different contrast, shading stands for SEM, n = 3 mice. c The ranking single-cell trial variance, smaller rank means lower trial-variability. The black pentagram indicates the center of the visual area, which is used as the center of the field of view in (f). d Trial variance in each brain area, the box indicates the range from the bottom quartile to the upper quartile. The centered line shows the median. The whiskers show the maximum and the minimum of data. n = 14445 neurons are included in the total. e Two-sided Wilcoxon rank sum test (with Bonferroni correction for multiple comparisons) result between every brain area. (*p < 7.6 × 10-4, **p = 7.6 × 10-5, ***p < 7.6 × 10−6). f Decoding grating orientation using increasing neuron numbers (bottom), and the corresponding field of view radius (top). The center of view is shown in (c) with a black star. The model result is produced by a setting with spatially ordered functional neurons. g Decoding accuracy changes with the number of principal components (PC). The shading represents ± SEM in repeated simulations with different initialized connections. h Decoding accuracy changes with the number of neurons. The order of neurons is either random or ranked by coding ability. The shading represents ± SEM in repeated shuffled runs (gray), and repeated simulations (purple). Source data are provided as a Source Data file.
