Hi,
> I'm new to this forum and I was wondering if the marsbar FIR time-course
> estimation is based on a non-shape assumption even if the original
> SPM-design was modeled with a canonical gamma HRF? So in other words, is the
> time-course analysis dependent on the modelled HRF or is it a different
> disentangeled analysis?
The FIR analysis is inherited from SPM, and what SPM (therefore
Marsbar) does, is to take the original design (with HRF modeling),
extract the event your are interested in, replace the HRF model for
that event with an FIR (non-shape) model, and reestimate. So, the
event of interest has an FIR model, and all the other events have
their original HRF model. It's a bit unsatisfactory, because bad fits
in the other events can have odd effects on the event-of-interest FIR,
and it might be better to use a full FIR model for all the events
(although this can soak up lots of degrees of freedom).
Best,
Matthew
|
