Hi,
I am struggling a bit with the FIR/PSTH code, and was hoping for some
help from the team.
My question is about the meaning of the units for the PSTH plot in SPM
graph. I'll lead up the question with some exposition of what I think
I've gleaned so far...
Just to be clear, I'm going to refer to 'dt' in the following; this is
just the TR divided by the (number of bins/TR) default value (usually
16). Thus, dt for a TR of 1 is usually 1/16. I'll refer to the (number
of bins/TR) value as fMRI_T.
The standard FIR model in SPM, estimates the Impluse Response to one
dt's worth of the event. So, if you specify one event with a 1 second
duration, with and FIR basis function model, the SPM machinery will go
away and make a set of regressors which represent a new event for every
dt timebin. So in fact the model for one event of 1 second duration
will resolve to 16 events of duration dt (TR 1, fMRI_T 16).
In general, in SPM2, when you model events of duration set to 0, SPM
puts in a regressor which is very similar to that for duration one
second (This is unlike SPM99, where an event of duration 0 was assumed
to be of duration dt. This is also so for FIR models).
The spm_graph code for the Peri-Stimulus Time Histogram (PSTH) uses the
FIR basis function machinery to make a new FIR model for the event being
plotted. After it estimates the new betas for this model, it takes the
betas for the FIR model columns, and divides by dt, before plotting the
values.
Here's the question: I think that the above means that the values in the
PSTH plot can be interpreted as activation (at a given time after event
onset) for one second's worth of event. Is this correct?
Thanks a lot,
Matthew
|