Supplementary Figure 4: Structured events prior to experience are also absent with place cells randomly downsampled or downsampled by L-ratio

(a) Distribution of spike correlations of 8310 candidate events during Sleep 1 after down-sampling to 15 cells. 15% candidate events in Sleep 1 did not produce real number correlation after down-sampling cells thus were not included in the histogram. Open bars indicate spiking events versus the original place fields templates; filled bars indicate spiking events versus 5000 shuffled templates scaled down 5000 times. Left, both place fields templates in two running directions were included (absolute correlation medians: 0.35 versus 0.36, data versus shuffle; one-tailed Wilcoxon rank sum test: p=0.75; one-tailed Kolmogorov-Smirnov test: p=0.81; Monte-Carlo P-value>0.4 with criteria of absolute correlation >0.5 or 0.6 or 0.7, see Methods). Right, only the biggest absolute spike correlation resulting from the two templates for each candidate event and shuffle were included (absolute correlation medians: 0.44 versus 0.47, data versus shuffle; one-tailed Wilcoxon rank sum test: p=1; one-tailed Kolmogorov-Smirnov test: p=1; Monte-Carlo P-value=1 with criteria of absolute correlation >0.5 or 0.6 or 0.7, see Methods). (b) Histogram (bin size=0.001) of L-ratio for 442 manually clustered, putative excitatory units in Sleep 1. Red line indicates the median value. (c) Distribution of spike correlations of 8051 candidate events during Sleep 1 after sub-sampling cells that have an L-ratio equal to or lower than the median value. 18% candidate events in Sleep 1 did not produce real number correlation after sub-sampling cells thus were not included in the histogram. Open and filled bars are as described in (a). Left, same as in (a); absolute correlation medians: 0.36 versus 0.36, data versus shuffle; one-tailed Wilcoxon rank sum test: p=0.40; one-tailed Kolmogorov-Smirnov test: p=0.60; Monte-Carlo P-value>0.1 with criteria of absolute correlation >0.5 or 0.6 or 0.7. Right, same as in (a); absolute correlation medians: 0.48 versus 0.50, data versus shuffle; one-tailed Wilcoxon rank sum test: p=1; one-tailed Kolmogorov-Smirnov test: p=1; Monte-Carlo P-value=1 with criteria of absolute correlation >0.5 or 0.6 or 0.7.