Hi Rick,
It seems that you and Bob had different opinions about the question. I can understand that the polort bases are able to take care of the mean and trends of the EPI time series. Will they also be able to take care of the "step function" shape of motion regressors? I assume that AFNI would just treat the concatenation of the motion parameters across runs as a single regressor. Ideally if there is no head motion within each run, then correlating to a straight line must be different with correlating to a step function?
Thanks,
Yisheng