Hello all,
I have a problem with computing contrasts. I am using SPM99 for
analysis. My design is as follows: event-related design with 6
sessions (3 attention to visual sessions, 3 attention to auditory
sessions). In each session there is, amongst others, a bimodal
congruent condition and a unimodal condition for whichever modality
you are attending to in that session, so across all 6 sessions there
are 6 bimodal conditions, 3 unimodal auditory conditions and 3
unimodal visual conditions. Now I want to compute a superadditivity
contrast (that is, to see whether the activation in my bimodal congruent
condition {BI-CON} is more than the sum of the two unimodal
conditions). Having modelled the null events this would be
(BI-CON-null)-[(unimodal auditory-null)+(unimodal visual-null)]. What
is difficult is deciding what the contrast weights would be for these
conditions for each of the sessions. If anyone has computed such a
contrast before I would be glad of advice here.
If this can be expressed as BI-CON - Unimodal Auditory - Unimodal
Visual +NULL then contrast weights would be 1, -1, -1, 1, respectively,
for these conditions over the 6 sessions. Does this actually compare
superadditivity or is it asking if the BI-CON activation is more than the
average of the unimodal conditions together?
Many Thanks,
Anthony
|