Dear SPM Members,
Although these issues were already discussed I still found myself
struggling in understanding the differences within the results under each
type of analysis.
I have a 4 condition block design (let us say C1 C2 C3 and C4), repeating
themselves in 4 scans (each block is presented twice in each scan thus, a
total of 8 time). I expect my manipulation to generate a linear increase
in activation so that C1<C2<C3<C4.
As far as I understand, in a standard design matrix (no parametric
modulation), masking contrast C1<C2 with C2<C3 and afterwards masking one
of them with C3<C4 should reveal voxels with the parametric nature I'm
looking for. Nevertheless, results of this type of analysis quite differ
from a [-3 -1 1 3] contrast (or any other slope). Moreover, looking at
the time courses, I find they don't necessarily demonstrate this
parametric characteristic, mostly the latter type of analysis (although I
assume the mean activation of each condition differ significantly...
should I assume that?)
As far as the parametric modulation goes, I have a couple of questions:
First, regarding the design itself, is there a difference between the
following set of coefficients, lets say [1 2 3 4] and this one [0 1 2 3]?
if I understand correctly, then there shouldn't be, right?
Second, regarding the contrast of a one condition parametric modulation, I
really find it hard to figure out how should the contrast look like, in
case my design matrix looks like the attached file...
Thank you for the help,
Ronen.
|