Figure 2. Excess correlations across familiar and novel environments. *A*, Neural activity was recorded using tetrodes implanted in the dorsal CA1 area of the hippocampus. *B*, *C*, Example pairs of pyramidal cells simultaneously recorded during free exploration of a familiar environment with significant positive (*B*) and negative (*C*) excess correlation *w* (gray histogram, distribution of correlation coefficients derived from the null model; red dashed line, measured raw pairwise correlation). *D*, Experimental paradigm. Aix animals explored a familiar environment, then rested in a sleep box (rest data not used), after which they explored a novel environment (20–40 min for each condition). Each animal contributed to one experimental session. *E*, Proportion of significant excess correlations for different cell types (triangle, pyramidal cell; circle, putative interneuron), positive (green) and negative (red) excess correlations, for both the familiar (top row, blue) and the novel (bottom row, orange) environment (stars, significant difference under binomial test at *p* < 0.001). Error bars indicate SEM across *N* = 6 animals. Shaded regions mark the fraction of interactions detected in the familiar environment that remain in the novel environment. *F*, Paired comparison (colormap, binned pair count) between excess correlations *w*_{ij} detected in familiar versus novel environment for each cell-pair within EE (left), EI (middle), and II (right) subnetworks. *G*, Example of an estimated excitatory subnetwork. Circles show the place field of each neuron, with edges showing significant excess correlations (green, positive; red, negative excess correlations); line thickness corresponds to interaction strength. *H*, Left, Proportion of significant excess correlations in the excitatory subnetwork increases with place field overlap, measured as the Pearson correlation of 2D firing rate maps, for positive interactions (blue, familiar environment; orange, novel environment; shaded area, 99th percentile confidence interval for the mean). Right, Analogous plot for negative excess correlations. *I*, *J*, Left, Distribution of log node-degree of E cells normalized by the total number of E cells in each session, for the novel (*I*) and familiar (*J*) environment. Inset, Quantile-quantile plot comparing this distribution to the normal expectation. Right, Excitatory subnetwork has a significantly higher clustering coefficient (orange/blue line, data) compared with the expected distribution for an Erdos–Renyi (ER) network with a matched connection density. See Extended Data Figures 2-1, 2-2, 2-3, 2-4 and Extended Data Tables 2-1, 2-2 for more information.