Hi Rick and Wolfram,
I am having the same problem with my data (3dFWHMx is returning a FWHM smaller than the one I applied to the data). My data (resting-state) is also bandpassed. The 3dFWHMx command I'm using is:
3dFWHMx -dset ${n}_whiteresids+tlrc.BRIK -mask 1mm_automask+tlrc.BRIK -detrend -combine >${n}_noiseFWHM.1D
The input is the Rwherr file from running 3dREMLfit on the data (178 TRs). Input of the Rwherr or the Rerrts residuals did not make an appreciable difference in the output.
Am I mistaken in the way I'm running 3dFWHMx?
Cheers,
Laurel