Dear Ben,
> to evaluate common activity for Neutral stimuli in the three conditions
This sounds like the conjunction null hypothesis (= "set of consistently significant effects").
> Is this analysis approach generally valid?
The contrasts are between-subject effects. To be able to run the conjunction you need a model like the flexible factorial though, which does not allow a valid inference on between-subject effects.
Now either ignore that, as it's nonetheless established ;-) , or turn to something different, e.g. just visualizing the different activation maps side by side, or combine the thresholded T maps to show the overlap with different colours (in your case with three levels/maps, something like i1+i2.*2+i3.*4 should do it, those voxels with sig. effect for 1 and 2 have a value of 3, those with sig. effect for 1 and 3 have a value of 5, those with a sig. effect for 2 and 3 have a value of 6, and so on). If you want to show both positive and negative activations it becomes more complicated, but then you can still treat the three T maps as six separate maps (going either in the positive or the negative direction).
Best
Helmut
|