I may have an embarrassing simple question: My fmri data was collected using 3-mm voxels (fairly standard). Now that the pre-processing is mostly done, I'm ready to transform into mni-152 space. The transformation module in SPM (Normalise: Estimate & Write), while transforming to mni space, also permits changing voxel size. SPM12 defaults to 2-mm isotropic voxels ... and most published papers appear to use this voxel size.  Is there any advantage to changing my voxel size from 3-mm to 2-mm? What might the advantage be of doing this? To my mind, doing this comes with no advantages and a number of disadvantages: (1) It significantly increases the size of the 4D fmri volume, (2) it requires significantly more memory when loading the entire volume into Matlab, and (3) one now needs to run the linear model on more voxels (and correct for this).

Kind regards,

-jim