Fig. 4: abGCs improve odor discrimination by MCs in awake mice.

a PCA representation from one mouse before (left) and after (right) CNO administration. For clarity, only six odors are shown in a 3-dimensional space of neural responses. Each dot is a trial; each color is an odor. Variance between different trials of the same odor (distance within) is represented as spheres, with a radius equal to the average distance of the dots from the centroid of their cluster. Separation between a certain pair of odors (distance between) is represented by a black arrow and is calculated as the Euclidean distance between the centroids of their clusters. The clusters of odors #1 (blue) vs. #3 (purple) are emphasized for the purpose of visualization only. b All matrices are calculated for the data shown in a only. Left 2 matrices showing all d-prime values in a pairwise manner before and after CNO. Colored arrows represent the odors shown in a. The red squares mark the example emphasized in a. Odor numbers as in Fig. 2b. Right—discrimination change index (DCI) matrix, in which each entry is calculated according to the presented DCI formula, from the data depicted in the two left matrices. c Matrices of d primes for all odor pairs, calculated in n-dimensional space (n = number of cells) for each mouse and then averaged over all mice, before and after CNO. d Same as c for saline treatment. e DCI following either CNO or saline averaged over all mice for all possible pairwise comparisons. Both conditions were significantly different than 0 (N = 66 DCIs for all comparisons, CNO vs. saline: p << 0.0001, CNO vs. 0: p << 0.0001, saline vs. 0: p < .0001, t-tests followed by Bonferroni correction). Statistical tests are two-sided, and error bars are SEMs.