Greetings, FSL experts,
Might someone be able to help me figure out how model a completely within-subjects 2x2x2x2?
Just to clarify, every stimulus is represented at one of two levels on each one of the four factors. Imagine for example, that each stimulus was:
-either a circle or a square, and also
-either red or blue, and also
-either striped or solid, and also
-either big or small
That means that there are 16 possible shapes, and I need to separately extract parameter estimates for each of the 16 conditions for each individual subject.
I'm not concerned with creating brain maps of the higher-level interactions (at least not for the moment). Rather, I'm just aiming to extract those 16 betas (or % signal change estimates) from specific regions of interest (so, imagine a bar chart showing 16 bars indicating how a V1 ROI responds to each of the 16 shapes). The interactions are quite relevant, but I can handle these calculations in Matlab, Excel, or SPSS, after extracting the values for each subject.
All that being the case, I'm inclined to think that the best way forward is to simply ignore the factorial structure of the data for the moment, and create 16 unique EVs, one for each of the 16 shapes described above. Anyone have thoughts on this?
With eternal gratitude,
-Dan
|