For SS analyses, neural population activity was projected onto the first three PCs of PCA space. These analyses allowed us to determine the time course of alterations in the pattern of firing rate. Similar patterns of population activity reside close to each other in 3-dimensional space, and when different are further apart. Differences in distance between 3-dimensional population activity vectors were evaluated on a bin-by-bin basis via Euclidean distance analyses (Ames et al., 2014). The mean distance between each trial and every other trial in that comparison type were made (for example drinking trial 1 vs all null trials, drinking trial 2 vs all null trials, etc.), and the mean and variance of the (non-redundant) distances were used for plotting and statistical analyses. We were specifically interested in differences between drinking and non-drinking trials (vs null trials), and therefore evaluated Euclidean distance between these groups and null trials on a bin by bin bases using Benjamini–Hochberg FDR-corrected rank-sum testing.
