Hi Mirko,
> Thank you very much for you clarification and guidance. Unfortunately I
> am still puzzled with your last statement "So it is this main time
> effect of overall activation that you can mask your interaction
> (inclusively) by in order to distinguish B activating more (i.e. [1 -1
> -1 1] & [-1 1 -1 1]) from B deactivating less (i.e. [1 -1 -1 1] & [1 -1
> 1 -1])." How can I get information on B activation using two rather
> different contrasts (i.e. [1 -1 -1 1] & [-1 1 -1 1]) and similarly B
> deactivation (i.e. [1 -1 -1 1] & [1 -1 1 -1])?
Under SPM5 at least, after selecting the contrast you are interested
in, you are prompted whether you want to "mask with other
contrast(s)", then after selecting a masking contrast, you are asked
for a p-value (uncorrected) for thresholding the mask, and then
whether you want inclusive or exclusive masking.
So, if you select the interaction contrast [1 -1 -1 1] (which could be
significant for B activating more than A or deactivating less), and
then mask this with e.g. the activation contrast [-1 1 -1 1] with a
liberal p-threshold, and inclusive masking, then the results should
only be the regions where B activated more than A, and not the results
where it deactivated less, as the latter would not be included in the
"masking contrast".
It's possible that I have completely misunderstood the whole purpose
and/or practice of contrast masking, and/or the nature of
activation/deactivation in your data; if so, I hope someone will
correct me on the list; if not, I hope the above makes sense and works
for you.
Best,
Ged.
|