Hi all,
I have a question about how you would go about testing/describing the
overlap of clusters produced by similar contrasts between different
tasks. In this case, I want to see whether a cluster produced by an
Old>New contrast in a memory task using images occurs in the same area
as one I get from an Old>New contrast in a word memory task. I imagine
this overlap depends on the thresholding used to a large degree, so this
may be a descriptive measure rather than a test? A purely descriptive
result is fine, I'm just trying to figure out if there is a better way
to do this than a simple overlay.
Would a correlation of voxels activated for each task within an ROI
(obtained from an f-test of all 4 conditions?) be a reasonable thing to
do? Otherwise, maybe comparing voxels common to both tasks versus those
active in only one task?
Any ideas would be much appreciated.
Thanks!
Jeremy
|