Dear AFNI experts,
I understand that the assumption of Gaussian shape for spatial autocorrelation function (ACF) of the noise is essentially wrong.
Anyway, I was wondering how spatial restrictions affect the estimation of acf/fwhm values. To give an example, the usage of a mask with separate brain regions (say PCC, dlPFC and mPFC) in a GLM (and in the consequent 3dClustSim) will lead to less discrepancy between the results obtained with acf/fwhm? Why?
Thanks.
Simone