Re: [Marsbar-users] Values fitted event time course

[Matthew B. <mat...@gm...>]

Hi,

Sorry to be slow to reply.

On Thu, Jul 11, 2013 at 2:46 AM, Marek Meier <mar...@gm...> wrote:
> Dear MarsBar experts,
>
> I hope that someone will find the time to help me with my questions. I used
> an HRF function in my model and I have two different conditions (each 4 s)
> and the mean duration of my jittered ISI is 5 s. I want to extract times
> course and I'm not sure whether I have to use the fitted event time course
> or FIR event time course.
>
> 1. Can I use the FIR event time course function or would that be wrong given
> that I used an HRF function in my model?

It would not be wrong, but I recommend you use the FIR model, and then
use the fitted time course in Marsbar.   The FIR event time course
fits a new FIR model just to the event of interest (not the other
events, which continue to have HRF models).  That makes it a little
hard to think about what might have happened because one event has a
richer model than the others.  That's way I personally prefer FIR
modeling in the first place.

> 2. What is the meaning of the numerical values in the "time_course" variable
> if I use "fitted event time course"? With an event duration of 4 s I get 191
> values. How can I use these values to calculate the time course for a
> certain time, e.g. 12 seconds?

You mean why are there 191 values?  SPM, when it generates its model,
uses what it calls a 'supersampled' time series to generate the
regressors, which are then downsampled to the TR.  The supersampled
series has time deltas of TR / 16.  The standard length in time for
the event basis functions is 34ish seconds.  So, say your TR was
around 2.5, then the time course would have; dt = 2.5 / 16; 34 / dt =
217 time points.

What do you mean by 'time course for a certain time?  Sorry if I'm
missing something,

Best,

Matthew