Dear group,
My question concerns the construction of parametric modulators (PM)
in 1st-level design matrices in SPM5. As has been discussed, the
set of PM variables is by default sequentially orthogonalized. I
did not want this, so (as Karl suggested in a posting about this) I
commented out the call to spm_orth in spm_get_ons. I then expected
that a given PM predictor in the design matrix would be the same
regardless of the presence of other PM variables. But this was not
the case.
To clarify, consider two PM variables: M1 and M2. For a fixed set of
event onsets, it should ideally be so that the resulting fMRI PM
predictor for M2 (call it pred_M2) should be the same whether it is
used alone or in conjunction with M1 (in the sequence M1, M2). But
while pred_M1 had this invariance to the presence of other PM
variables , pred_M2 (and any other PM variables besides M1) did
not. So some sort of sequential modification to the PM variables is
still being done, but it does not seem to be occurring in
spm_get_ons. In particular, the stored values of M2 (in
SPM.Sess.U.P(2).P, which is created in spm_get_ons) were identical.
I am grateful for any ideas or advice.
Thanks,
Eric
|