The dataset is represented using a difference estimation plot (seeSTAR Methods) to visualize the effect size for changes in cocaine-biased behavior across tests. Top, raw data points, with each set of four connected points reporting the difference between cocaine- minus saline-paired LED activation for one mouse in each stage; bar charts, average (mean SEM) LED activation preference over animals. Bottom, corresponding effect size, using the mean difference estimation for each stage compared to pre-test. Black dot, mean difference; filled curve: distribution of mean differences; black lines, 95% confidence interval; ###p < 0.001; ## and**p < 0.01; # and*p < 0.05; repeated ANOVA for pairwise stage-stage (#, vs. pre-test;*, vs. extinction) interactions, with Bonferroni correction for multiple comparisons.
