Yes, naming task led to sever head movement, and there were really possibilites that motion might contribute my deactivatin results.I had ,however,a block design, so there might not exist reasonable way to separate motion from the fmri signal.And I had browse Rasmus Birn's pernal website,no further help was found.
I had cencored the ourlier time points by 3dToutcount -range,by which the value in the first column bigger than the value in the second column was regarded as outliers, and cencored. I had got some ,but minor,changes.Is there any more effective approaches?Maybe it can be only done before processing?