Fig. 3: Predictive dopamine signals across behavioral strategies.
From: Striatal dopamine signals reflect perceived cue–action–outcome associations in mice
a, Top, average normalized dLight fluorescence in the VS in correct trials split by early and late novice sessions during the location rule. Data are shown as mean ± s.e.m. across sessions. Bottom left, average fraction of correct trials (fraction corr.) for early and late novice sessions. Bottom right, average normalized dLight fluorescence peaks in trial epochs cue, spouts and outcome (out.) in correct trials split by early and late novice sessions (n = 12/12 sessions for early/late). Data were analyzed by Kruskal–Wallis tests across pooled sessions (performance (perf.), P = 0.021; cue, P = 0.002; spouts, P = 1.4 × 10−4; outcome, P = 0.525). b, Same as a for the DMS (n = 10/10 sessions; performance, P = 0.059; cue, P = 0.028; spouts, P = 1.6 × 10−4; outcome, P = 1.0). c, Same as a for the DLS (n = 12/12 sessions; performance, P = 0.018; cue, P = 0.817; spouts, P = 0.003; outcome, P = 0.033). d, dLight fluorescence and response bias across performance levels in the frequency reversed task. Top left, average normalized dLight fluorescence in the cue epoch in the VS split by error and correct trials. Top right, average normalized dLight fluorescence peaks in the cue epoch. Data were analyzed by two-sided Wilcoxon rank-sum tests across pooled sessions with a Bonferroni correction for multiple comparisons (novice, n = 6 sessions, P = 0.046; intermediate, n = 9 sessions, P = 0.566; expert, n = 8 sessions, P = 0.234). Bottom left, average absolute response bias. Bottom right, average normalized dLight fluorescence peaks in the cue epoch in the VS; the difference between error and correct trials across performance levels is shown; DA, dopamine; pref., preference. e, Same as d for the DMS (novice, n = 10 sessions, P = 0.021; intermediate, n = 11 sessions, P = 0.590; expert, n = 10 sessions, P = 1.0). f, Same as d for the DLS (novice, n = 11 sessions, P = 0.196; intermediate, n = 14 sessions, P = 0.034; expert, n = 8 sessions, P = 1.0). See Supplementary Table 2 for linear mixed model analyses with animal as the grouping factor.
