Hi -
I have an fMRI design/protocol question for the list.
How many conditions is it statistically sound to include in one SPM model?
We are looking at fMRI activation during inductions of specific emotions.
Given a block design, emotion1, neutral, emotion2, neutral, emotion3,
neutral, emotion4, neutral....how many can we/should we/is it sensible to
include in a single study?
Our aims, to identify regional activation associated with appraisal themes
of specific emotions, don't really map onto a 'cognitive subtraction'
design, since few things really function as 'exactly the same as a given
emotional experience but without the emotion', and we're not really
parametric...since we're looking at different emotions not varied levels of
an emotional state.
At some level, it seems better to include lots of different states because
then 'baseline' activation will include ample variability, i.e. it won't be
biased towards the valence or arousal levels of other emotions. On the
other hand, do we shoot ourselves in the foot statistically/physiologically
by including too many conditions in a single run and hoping to find
activation related to only 1 of, say 10 conditions?
Your thoughts on this will be greatly appreciated.
|