The WPLI-debiased method outputs a single value for a given trial set and thus, we cannot perform a traditional statistics to test whether there is a difference in WPLI-debiased coherence between the baseline and delay period of a single channel pair. To test for a statistically significant difference in coherence between baseline (700 ms before fixation onset) and delay (400–1,100 ms after stimulus onset) period in a single channel pair, we used a bootstrapping approach. We created 1,000 bootstraps with replacement of the original trial set and obtained a distribution of the WPLI-debiased for the baseline and delay periods. The difference between baseline and delay was determined to be significant if the mean of the WPLI-debiased in the delay period fell above or below the 99.9th percentile of the WPLI-debiased distribution of the baseline period. To further confirm the results of the bootstrapping procedure, we pooled the baseline and delay WPLI-debiased results and randomly assigned them into two groups. Then, we tested whether the mean difference between the original baseline and delay period falls above or below the 99.9th percentile
