t-test, P<0.001 in both contra- and ipsiversive trials; Fig. 3h) increased more strongly in the direction of ACC to FEF than vice versa (GACC→FEF−GFEF→ACC). The results also indicated that the influence of ACC on FEF is greater than that of FEF over ACC during the delay period across alpha and gamma frequency ranges (n=275, paired t-test, P<0.05), as shown in Supplementary Fig. 4C. We performed a similar bootstrapping test as described for the weighted phase lag index (WPLI)-debiased analysis, to investigate the significant difference between baseline and delay periods, across the single channel pairs with flipped Granger-causality sign following the reversal of time series. In the direction of GACC→FEF, respectively, 115/275 (41.82%), 47/275 (17.09%), and 41/275 (14.91%) of channel pairs exhibited concurrent theta and beta increase, theta-band-only increase, and beta-band-only increase in Granger causality in the delay period. On the other hand, respectively, 46/275 (16.73%), 34/275 (12.36%), and 52/275 (18.91%) of channel pairs displayed concurrent theta- and beta-band, theta-band-only, and beta-band-only decrease in Granger causality in the delay period. Within the direction of GFEF→ACC, we observed that, respectively, 100/275 (36.36%), 68/275 (24.73%), and 26/275 (9.45%) of channel pairs showed concurrent theta- and beta-band, theta-band-only, and beta-band-only increase in Granger causality
