Hi Yan,

The issue of using motion correction has been discussed extensively throughout the forum and in some papers. (try searching for motion correction & design matrix). By adding motion regressors to the design matrix, you reduce the degrees of freedom for the analysis, which can significantly reduce the amount of fMRI activation. We addressed this problem by correcting our data for motion prior to running the first-level analyses (search for Art Toolbox by Paul Mazaika). 

Here's a link to the thread:


Hope that's a good starting point.

Vy T.U. Dinh
Research Assistant, Neurological Sciences
Rush University Medical Center
Phone: (312) 563-3853
Fax: (312) 563-4660
Email: [log in to unmask]

From: SPM (Statistical Parametric Mapping) [[log in to unmask]] on behalf of Dr Cyril Pernet [[log in to unmask]]
Sent: Friday, August 05, 2011 1:22 AM
To: [log in to unmask]
Subject: Re: [SPM] motion regressors

Dear  Yan,

As soon as you change the design matrix, results will be different. In general term you can think of the data as a point in a N dimensional space and you project onto a subspace at P dimensions ie the nb of columns of the design matrix (see SPM book). As the design matrix changes, so does the projection.

For the specific case a motion regressors many voxels will have some variance explained by these regressors so it is a good idea to include them, see a neuroimage paper from Lund.


Dr Cyril Pernet
University of Edinburgh

----- Reply message -----
From: "Yan Fang" <[log in to unmask]>
Date: Fri, Aug 5, 2011 00:50
Subject: [SPM] motion regressors
To: <[log in to unmask]>


Can I ask a question about motion regressors generated after realignment? I reorient all the images after realignment and coregistration. If I still add motion regressors to the design matrix, will this affect the result? Thank you!