(A) Task design for the operant-based reversal-learning task. (B and C) Acquisition curves for discrimination (B) and reversal (C) demonstrate a selective impairment in reversal learning in CNO-treated MDhM4D mice. (Discrimination: repeated measures ANOVA, time p<0.001, group p=0.3593; GFP-sal n=23, GFP-CNO n=7, hM4D-sal n=42 and hM4D-CNO n=11; Reversal; repeated measures ANOVA followed by Bonferroni post-hoc tests, *p<0.05, **p<0.01; GFP-sal n=15, GFP-CNO n=14, hM4D-sal n=26 and hM4D-CNO n=25). (D) S- lever pressing during the reversal phase. All mice decreased responding over time (repeated measures ANOVA, main effect of time, p<0.001). However, only CNO-treated MDhM4D mice made more perseverative errors (repeated measures ANOVA followed by Bonferroni post-hoc analysis *p<0.05, ***p<0.001). Error bars: s.e.m. See also Figure S3.