Re the literature (e.g."Stochastic Designs in Event-Related fMRI," Friston
et al., NeuroImage 10) on maintaining efficiency of detecting main effects
when using short SOAs by using stochastic designs with null events included:
Suppose one wants to make a contrast for the main response to trial type #1
(as opposed to a differential response of some kind). Would the contrast
used be [1 0 0 ... 0], or would it be [1 0 0 ... 0 -1], where the "-1" is
for a null event? The former assumes that a null event is modelled as
having a hemodynamic response of zero. The latter makes a little more sense
in the context of the paper cited above, which seems to treat null events
the same as any other, but I'm really not sure. A derivative question is,
is the construction of such a contrast automatically done by SPM99b? (I'm
just getting started with it.)
Regards,
Steve Fromm
NIDCD/NIH
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|