Fig. 3: Population coding of CS1 and US1, despite individual neurons’ variability within and between sessions.
A Field of view of a representative animal used for neuronal tracking during appetitive Pavlovian conditioning. B Number of neurons presenting each type of response of tracked D1-MSNs or C D2-MSNs on day 1, 5 and 10 of conditioning, showing variability in individual CS1-US1 responses over days (nD1 = 272, nD2 = 172). D Data from individual D1-MSNs’ activity was used to train a SVM decoder for the identification of CS1 and US1 (features: single neuron activity during stimulus; observations: average activity during given trial; output: CS1 or US1). Neurons are sorted and colored by decoding accuracy on day 1; the same was done for E D2-MSNs. F, G D1- or D2-MSNs with higher decoding accuracy on day 1 did not necessarily have relevant decoding accuracies throughout days. Colored dots represent accuracies on day 1 and gray dots represent the decoding accuracy of the same cells in days 5 or 10. H, I Decoding accuracies for all D1- or D2-MSNs and for top or bottom 20% cells in terms of accuracy. The decoding accuracy of the top 20% cells on day 1 largely overlapped with the entire population on days 5 and 10 for both populations. J Cartoon depicting Population-based SVM decoder (features: population activity during stimulus; observations: average activity during given trial; output: CS1 or US1). K SVM classification accuracy using population activity during CS1 and US1 (0–3 s from CS or US onset) for D1-MSNs (One-way ANOVA; Bonferroni post hoc) or L D2-MSNs (nD1-cre = 11, nA2A-cre = 9). The population decoders efficiently identified CS1 and US1 events for both populations. M Population vector correlation of CS1 responses of D1- or D2-MSNs within and between sessions (nD1-cre = 10, nA2A-cre = 9; Two-Way ANOVA, Bonferroni post hoc). N Population vector correlation of US1 responses of D1- or D2-MSNs, suggesting representational drift of US encoding between sessions (nD1-cre = 9, nA2A-cre = 9; Two-Way ANOVA, Bonferroni post hoc). Data are presented as mean values +/−SEM. *p < 0.05, **p < 0.01. Source data are provided as a Source Data file.
