Fig. 3: Decoder results are explained by changes in the ratio of selective cells across schedules.
A Each row plots the difference in trial-averaged activity of left trials minus right trials triggered by the time of choice. Cells plotted are all imaged neurons where we observed a significant effect of side, sorted from the most negative beta coefficient to the most positive one. The activity of the neurons was range-normalized. A value closer to 1 indicates more selectivity for the left side trials, and a value closer to − 1 indicates more selectivity for the right side. B, C Same as A, but the difference is the averaged activity of unrewarded trials minus rewarded trials, centered at time of reward cue (B) or reward retrieval (C). D Ratio of side selective neurons is overall larger in M2 (nM2 = 18, nOFC = 24, p < 10−4), but it increases with schedule only in OFC (nSch1 = 8, nSch2 = 8, nSch3 = 8, p = 0.0001). Group sizes from left to right are: n = 6, n = 6, n = 6, n = 8, n = 8, n = 8. E Ratio of outcome selective neurons increases with uncertainty only in OFC (nSch1 = 8, nSch2 = 8, nSch3 = 8, p = 0.0091). Group sizes from left to right are: n = 6, n = 6, n = 6, n = 8, n = 8, n = 8. F Ratio of reward selective neurons is overall larger in M2 (nM2 = 18, nOFC = 24, p = 0.033) but increases with schedule only in OFC (nSch1 = 8, nSch2 = 8, nSch3 = 8, p = 0.032). Group sizes from left to right are: n = 6, n = 6, n = 6, n = 8, n = 8, n = 8. G Example of the z-score normalized activity of a significant side-selective M2 neuron on left and right trials. H Same as (G) but for a significant outcome-selective M2 cell. I Same as (G), but for a significant reward-selective OFC neuron. The center of each boxplot is the median. All statistical tests performed using a Generalized Linear Model. *p < 0.05. **p < 0.01 ***p < 0.001. Source data are provided as a Source Data file.
