Dear Helmut,
what you observe is a consequence of whitening (ie the computation of
SPM.xX.W*SPM.xX.X).
You can see the various effects of high pass filtering and whitening by
looking at each element of the equation below after model estimation:
load SPM.mat
X = spm_filter(SPM.xX.K,SPM.xX.W*SPM.xX.X);
Best regards,
Guillaume.
On 10/02/15 15:23, H. Nebl wrote:
> Dear everyone,
>
> I've just came across this accidentally, but what does setting the high-pass filter to "Inf" actually mean? Up to now I had that this would correspond to "no filter applied", but something seems to happen, e.g. the constant term is not constant after model estimation / the first few and last vol. have different values than those in between, and the values are not identical to 1 (which they still are when looking at the design after model specification, which seems to reflect the unfiltered time course, as it's always the same independent of setting).
>
> I found this message https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0701&L=spm&D=0&1=spm&9=A&I=-3&J=on&d=No+Match%3BMatch%3BMatches&z=4&P=68718 but I'm not really sure what this means.
>
> Best,
>
> Helmut
>
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|