(A) Trial structure for the no-reward control task. Gray shaded bar represents the response period analyzed for tone-responsive activity. (B) Mean percent of tone-responsive PV-INs from the no-reward task. Violin plots show null distribution of percentage of responsive neurons made by resampling mice and shuffling the session 1000 times (same as Figure 2). The circle represents the mean percentage of tone-responsive neurons. In the absence of reward, PV-INs did not show significant responses to tone. Bootstrap, day 1: p = 0.11, day 7: p = 0.11. (C) The mean percent of tone-responsive PV-INs during no-reward control do not change between days 1 and 7. Paired t-test, n.s., nonsignificant, p = 0.89. (D) Trial structure for the nonpaired control task. Tone was followed by a randomly varied 40- to 80-s period before reward delivery and then a randomly varied intertrial interval (ITI) between 15 and 25 s. First gray shaded bar shows the response period analyzed for tone-responsive activity (2.5 s from tone onset). Second shaded bar shows the response period analyzed for reward-responsive activity (2.5 s from onset of first lick after reward delivery). (E) Mean lick rate during the tone from the nonpaired task. Mice did not show increased anticipatory licking during the tone on day 7. One-way analysis of variance (ANOVA), p = 0.097. (F) Mean lick rate following the randomized water rewards from the nonpaired task. Mice increased lick rate to consume the reward. The lick response was greater on day 7. One-way ANOVA, p < 1 × 10−3. (G) Mean percent of tone-responsive PV-INs during the nonpaired task. Violin plots show null distribution of percentage of responsive neurons made by resampling mice and shuffling the session 1000 times. The circle represents the mean percentage of tone-responsive neurons. When tone was not paired with reward, PV-INs showed significant responses to tone on day 1 but not day 7. Bootstrap, Bonferroni correction for multiple comparisons, day 1 p = 0.003, day 7 p = 0.069. (H) The mean percent of tone-responsive PV-INs during the nonpaired task did not change between days 1 and 7. Paired t-test, n.s., nonsignificant, p = 0.43. (I) Mean reliability index of PV-INs that were responsive to the tone during the nonpaired task. Cells were divided into High or Low Reliability Group based on the 50th percentile of tone reliability. The reliability of neither High nor Low Reliability PV-INs changed from days 1 to 7. Paired t-test, n.s., nonsignificant, High: p = 0.86, Low: p = 0.19. (J) Mean percent of reward-responsive PV-INs during the nonpaired task. Violin plots show null distribution of percentage of responsive neurons made by re-sampling mice and shuffling the session 1000 times. The circle represents the mean percentage of reward-responsive neurons. PV-INs responded significantly to randomly timed rewards on days 1 and 7. Bootstrap, Bonferroni correction for multiple comparisons, day 1: p < 1 × 10−3,day 7: p < 1 × 10−3. (K) The mean percent of reward-responsive PV-INs during the nonpaired task did not change between days 1 and 7. Paired t-test, n.s., nonsignificant, p = 0.51. (L) Mean reliability index of PV-INs that were responsive to the reward. Cells were divided into High or Low Reliability Group based on the 50th percentile of reward reliability. The reliability of neither High nor Low Reliability PV-INs changed from days 1 to 7. Paired t-test, n.s., nonsignificant, High: p = 0.23, Low: p = 0.24. (M) Mean percent of reward-responsive VIP-INs during the nonpaired task. Violin plots show null distribution of percentage of responsive neurons made by resampling mice and shuffling the session 1000 times. The circle represents the mean percentage of reward-responsive neurons. VIP-INs responded significantly to randomly timed rewards on days 1 and 7. Bootstrap, day 1: p < 1 × 10−3, day 7: p < 1 × 10−3. (N) The mean percent of reward-responsive VIP-INs during the nonpaired control did not change between days 1 and 7. Paired t-test, n.s., nonsignificant, p = 0.57. (O) Mean reliability index of VIP-INs that were responsive to the reward. Cells were divided into High or Low Reliability Group based on the 50th percentile of reward reliability. The reliability of neither High nor Low Reliability PV-INs changed from days 1 to 7. Paired t-test, n.s., nonsignificant, High: p = 0.53, Low: p = 0.22. **p < 0.01, ***p < 0.001.
No-water rewards: n = 357 PV-IN cells from three mice. Nonpaired rewards: n = 324 PV-IN cells from five mice; n = 208 VIP-IN cells from three mice. Error bars show standard error of the mean (SEM).