With resting state analyses, would it be wise to use 3dREMLfit to reduce the autocorrelations in the data or will this wipe away the low-frequency fluctuations that we are interested in? The reason I had thought to use 3dREMLfit would be to obviate the need to control for the degrees of freedom like Fox et al. (Spontaneous neuronal activity distinguishes human dorsal and ventral attention systems, 2006 PNAS).
However, it's possible that I haven't thought of the repercussions of autocorrelations with resting state and it doesn't look like this is a step recommended from the recent Advanced AFNI course so any opinions you have on the matter would be greatly appreciated!
-Matt