Dear SPMers,
Based on some previous analyses that we conducted which revealed that our
response signal may differ from the canonical HRF, we are interested in
doing an FIR analysis with our data. Since this is our first attempt, lots
of questions have arisen:
1. Is it okay to have your time bins be longer than your TR? Our TR is 1.5
sec, but we have specified our bins to be 3 secs long. Or is it convential
for your bin to equal your TR?
2. In our paradigm, we have subjects perform a task that contains a working
memory (WM) component and a control (C) component. The subjects perform
this task under two different conditions (on two different days). What is
the best way to test for task component x condition interactions for the
entire sample?
3. If we believe we have an idea of when the peak signal occurs, is it okay
to only take that bin (for WM-C) forward to the second-level RFX analysis
level and perform a one-way t-test?
I have read the face data example, but it doesn't appear to address
anything beyond main effects of faces. It would be helpful to know how to
test for famous vs non-famous...and even more helpful if the task was
completed under two different conditions by the same group.
Thanks for any guidance!
Rachel
|