Fig. 4: Emergence of cue-evoked ACh increases in the anterior dorsal striatum during partial extinction.

a Schematic of the partial extinction paradigm. b Mean lick index on light and tone cue trials for each day of training for one representative mouse. Shaded regions, S.E.M. across trials for each day. c Rasters of normalized licking aligned to light cue onset for all trials from the beginning of extinction training for a single mouse. d Lick indices across all mice (n = 8) for the light and tone cues associated with different reward probabilities during initial learning (100%) and partial extinction (20% and 80%). Mean across mice shown in red, individual mice in gray. Error bars, S.E.M. Repeated measures one-way ANOVA, followed by Tukey post hoc test. *: p < 0.05, ***: p < 0.001, ****: p < 0.0001. T: tone, L: light. e Top: Mean ∆F/F aligned to the light cue onset for a representative fiber with a significant decrease in the early peak from learning to extinction for trials where light was associated with 100% (post-learning), 80%, and 20% (extinction) probabilities. Shaded regions, S.E.M. Red dots in insets indicate the fiber locations in the coronal plane. Bottom: Light-triggered ∆F/F for all trials included in the triggered averages at top. f Left: percent of all fibers with significant increases or decreases, no change, or no significant component from post learning to the 20% light extinction phase for the early peak ∆F/F at light cue onset. Right: histogram of number of fibers per mouse with significant early peak changes. Each bar is the fiber count for one mouse for each condition indicated by colors at left. g Mean ∆F/F aligned to the light cue onset as in (e) for a representative fiber with a significant increase in the late peak and decrease in the dip from late learning to the extinction phases. h Left: percent of all fibers with changes from post learning to extinction as in (f) for the dip and late peak components. Right: histogram of number of fibers per mouse with significant dip (top) and late peak (bottom) changes. Each bar is the fiber count for one mouse for each condition indicated by colors at left. i Maps (axial, top; sagittal, bottom) showing spatially weighted means across locations of differences between the 20% extinction and 100% post learning phases (20%−100% ∆F/F) for the mean early peak ∆F/F at light cue onset. Lines indicate the axes of maximal variation and arrows indicate the direction of peak decreases from smallest to largest changes. White contours indicate regions with changes in the highest 10^th percentile. j Same is (i) but for dip (left) and late peak (right). k Model coefficients indicating the relative magnitude and direction variation of the mean ∆F/F differences between the 20% extinction and 100% post learning phases for the early and late peak components for light and tone cues across each striatal axis (AP: anterior-posterior, ML: medial-lateral, DV: dorsal-ventral). The coefficients represent fixed effects derived from a linear mixed-effects model, with individual recording sites as the unit of analysis and mouse identity included as a random effect. The sign of the coefficient indicates the direction of the largest differences (see arrows). Error bars, S.E.M. *p < 0.05, **p < 0.01 ***p < 0.001, ****p < 0.0001, two-tailed Wald t-test followed by bonferroni post hoc analysis on model coefficients. Significant interaction terms (black) indicate difference in coefficients between early and late peak for a given axis. l Same as (k) but comparing spatial coefficients for tone and light for all three components. m Mean light cue-evoked late peak ∆F/F (green) and lick index (blue) for all trials following the transition from 100% reward probability to 20% reward probability for a single fiber. The lines indicate where each measure significantly (CUSUM algorithm, see Methods) changed relative to the 100% probability phase. n Histogram showing the # fibers with relative latencies between the significant increase in light cue-evoked late peak ∆F/F and the decrease in lick index following high to low reward probability transitions. Vertical line indicates the median of the distribution. o Same as (m) but for tone. p Same as (n), but for tone. The exact p-values and additional statistical details can be found in Supplementary Table 1. Water drop schematic in (a) adapted from SciDraw (scidraw.io), licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Source (https://doi.org/10.5281/zenodo.3925935). Brain schematics in (e) and (g) were adapted from the Allen Mouse Brain Common Coordinate Framework (CCFv3) (https://atlas.brain-map.org/). Source data are provided as a Source data file.