Extended Data Fig. 3: Dopamine signals across learning.

a) (left 3 columns) jRCaMP1b dopamine signals in the nucleus accumbens core (NAc, black, n = 9) and simultaneous recordings in the ventral tegmental area (VTA, purple, n = 3) and dorsal striatum (DS, green, n = 6), for cued reward trials in the trial bins indicated across training. (right) Reward or omission signals in NAc (top), VTA (middle), and DS (bottom) in trials 600–800, for cued (red), uncued (black), and cued but omitted (blue) trials. b) Mean signals from the data in panel (a) during the 1 s following cue delivery (left) and 2 s following reward delivery (right) across training for NAc (black, n = 9), VTA (purple, n = 3), and DS (green, n = 6). c) Example simultaneous recordings from NAc + VTA (top) and NAc + DS (bottom). d) Mean cross correlations for simultaneously measured NAc + VTA signals (top row, n = 3) and NAc + DS signals (bottom row, n = 6) in trials 1–100 (left) and trials 700–800 (right) within trial periods (1 s before cue to 3 s after reward). e) Peak cross correlation coefficients between NAc + VTA (purple, n = 3) and NAc + DS (green, n = 6) signal pairs across training, within “trial periods” defined as the time between cue start and 3 s after reward delivery. f) Correlation of initial NAc–DA reward signals with final combined preparatory behaviors (see Methods) (Pearson’s r = −0.85, p = 0.004). g) No correlation of initial NAc–DA reward signals with final combined reactive behaviors (Pearson’s p = 0.25). h) Correlation of initial NAc–DA reward signals predicted from behavior measures in trials 700–800 (see Methods) to observed initial NAc–DA reward signals (Pearson’s r = 0.99, p < 0.0001). i) Reward collection latency for simulations with low (cyan) and high (magenta) initial reward-related sensory input. j) Preparatory cued licking for simulations with low (cyan) and high (magenta) initial reward-related sensory input. All error bars and bands represent +/− SEM around the mean.