Hello spm
A question relating to the constrast weights used in analysis of fMRI
data. We have 2 conditions – A and B – where both are “active”
conditions (ie perception of words [A] vs perception of nonwords [B], we
have no rest condition). To look at the comparison words - nonwords,
we use the orthogonal contrast 1 –1.
What we don’t understand, however, is whether the use of these contrast
weights represents
(1) voxels which are active in A but not active in B or
(2) voxels which are active in condition A and deactivated in condition
B
Secondly, for the contrast 1 0, would this thus refer to
(1) voxels which are active in A independent of whether they are active
in B or not or
(2) voxels which are active in A but not active in B
If, in the latter case, option 1 is correct, then in what way would the
weights 1 0 for A vs B differ from the weights 1 0 applied to A vs C,
where C is a putative rest condition in which the subject listens to the
scanner and watches an empty presentation screen? And if they do not
differ, then under what experimental conditions should the contrast 1 0
be used, given that brain activity relating to the scanner noise, random
mentation etc is rarely of theoretical interest?
Or does the contrast 1 0 refer to voxels which are active in A but do
not exceed some given threshold of activation throughout the entire
experiment, given by the mean, or something similar?
Thanks very much in advance
Pilar Salgado, Pere Vendrell
|