Brain-wide normalization of task activation contrast maps

Hello all,

In our lab we are currently discussing normalization methods, and I am having trouble finding papers on this topic. More specifically, is there any literature on normalizing brain-wide task activation maps after a contrast, for example between 2-back and 0-back. By normalization, I mean, taking all values (in this case the difference in beta weights between the two task conditions) across the brain (either voxel-wise or ROI-wise), subtracting the mean and dividing by some scaling factor, so that voxels/ROIs now represent relative activation. This would be done separately for each run of the task.

In the past with my (non-fMRI) data I have done this despite the perhaps questionable theoretical justification, and group differences have ended up cleaner. We also think it would help with data reliability within subjects. Does anyone else do this and if so, are there papers to cite on this topic?