We are doing an experiment with two conditions, A and B, with the main
effect of interest being the difference A-B. We suspect that this
difference may be declining progressively during the trial due to some
sort of habituation. Is there any way to test for this using SPM?
The analysis we've tried so far involves defining a design matrix with a
linear time-dependent paramater applied to both A and B. This creates a
design matrix with successive columns [A Axtime B Bxtime]. We have then
defined contrasts individually for subjects of the sort [0 -1 0 1] to
form the difference (Axtime)-(Bxtime) with an added negative sign to
account for the fact that the paramaterized time dependence increases
with time while we suspect A-B to decrease with time. As my final
intuition-based step I ran a paired t-test on the individual contrast
images. Unfortunately, my intuation and my reason don't rhyme too well
at this point and I'm at a loss to translate the results even though I
suspect that this approach is nearing the mark. Am I on the right
track? Or is there another preferred way to perform this type of
analysis?
Thanks in advance for any input.
Samuel McClure
|