Adjusting the for the effects of interest removes the effect of the null
space regressors (e.g. motion, bad points, etc.). This step is necessary as
you want to remove any non-neuronal sources of signal in your data before
you deconvolve the data to get the estimate of neural activity. For the
omnibus test, you should include all tasks. This is done by default.

Please see my other email for details on individual VOIs.