in no-swap trials between sites contralateral and ipsilateral to the sample location (Figures 8C and 8D). Between-region differences in signal-to-noise ratio (SNR) can result in false-positive Granger causality. As our results were pooled across left-to-right and right-to-left hemispheres, any between-hemisphere SNR differences should have averaged out. To further rule out this possibility, we recomputed Granger causality on time-reversed data, which preserves SNR differences but should reverse the direction of true causal signal flow (Bastos and Schoffelen, 2016; Haufe et al., 2013). This is exactly what we observed (Figure S4). These results indicate that during the time period of putative WM trace transfer, signals flow in the predicted direction, from the sender to the receiver hemisphere.
