> As an alternative to Step One, would it be okay to use the -errts output from
> the original 3dDeconvolve command, and continue following the instructions?
Yes, this is the more straightforward way to get what you want in this case.
> Once I follow Step One (or use the -errts output?), task has been removed
> from the time series. I no longer need to worry about it, and do not need to
> include any task regressors once I get to the actual correlation analysis
> performed in Step Four. Am I following correctly?
Yes, that's correct.