Vinny,
If you want to exclude a time point in a time series from the analysis, create an intervention regressor with all 0's except that you put a 1 at the censored time point. Then feed this intervention regressor as a covariate in the causality analysis. Basically it's the same strategy as the -censor option in 3dDeconvolve.
Gang