Extended Data Fig. 8: Characterizing behaviour probability-encoding ACC neurons.

(a) GRIN lens implant locations in each mouse. (b) Maps of registered ROIs in each mouse. (c) Diagram of procedure to identify behaviour probability-encoding neurons from GLM over representative trace of a lick probability-encoding PC aligned to licks. (d) auROC of GLMs predicting each behaviour across sessions (paired t-tests, FDR < 0.01: pAll comparisons <0.0001). (e) Average PSTHs of lick-encoding neurons over onset of each behaviour, sorted by Lick-evoked activity and z-scored to 1-s prior to behaviour initiation. (f) Average activity across pooled behaviour probability-encoding neurons around their preferred behaviour. From left to right: neurons activated vs. inhibited by that behaviour. From top to bottom: behaviour-evoked activity in still, walk, rear, groom, and lick-probability encoding neurons. (g) Permutation test p-values within each animal testing if Fisher decoding accuracy for each behaviour is significantly higher than chance in each session (n = 1000 shuffles). (h) auROC of Fisher decoder in each animal and session in real and shuffled data. (i) Average Fisher decoding accuracy of pain vs. non-pain states in capsaicin (left) and capsaicin + morphine (right). (j) Permutation test p-values within each animal testing if Fisher decoding accuracy for states is significantly higher than chance in each session (n = 1000 shuffles). (k) auROC of Fisher decoder in each animal and session in real and shuffled data. (l) Selectivity of positive (left) and negative (right) pLick neurons for lick compared each other behaviour in capsaicin and capsaicin+morphine sessions measured by d-prime (Two-Way ANOVA, Tukey correction: positive pLick neurons pTreatment = 0.037, pBehaviour <0.0001, negative pLick neurons, pTreatment = 0.0007, pBehaviour <0.0001). ⋆ = p < 0.05. Bars, lines, or dots are mean; small dots are individual animals; error bars and shaded areas are SEM. See Table S1 Rows 111–119 for statistics.