> the inter-regional variability in neural/hemodynamic coupling (specifically,
> onset delay) would seem to pose a major problem for using this method on
> BOLD data. Wouldn't such variable delays make the results uninterpretable?
There are at least two causes for such an onset delay: inter-regional cause-effect relationship and neurovascular differences. The former is what the causality analysis relies upon, while the later is a confounding effect. See some discussion in
Catie Changa, Moriah E. Thomasonb and Gary H. Glover, Mapping and correction of vascular hemodynamic latency in the BOLD signal, NeuroImage, Volume 43, Issue 1, 2008
Gang