afni.nimh.nih.gov

No, adding a column of [-7 -5 -3 -1 1 3 5 7] as a covariate would not work because it would have multi-collinearity problem with those dummy variables in the original design matrix. It can be done by specifying a different design matrix by including the linear trend weighting as one of the dummy variables using something like contr.poly(8), but that would involve adding a new set of option for 3dLME.R

Gang