Dear Peter,
what is your hypothesis?
Are you assuming that there is a linear relationship between the rating and the BOLD signal? If so, I would suggest a parmetrical modulation (select 'other' and polynomial order=1). There you just have to enter one value per trial. That will model an additional regressor into your model which is convolve with the basic-function.
In principle, you can enter such an regressor you have mentioned, but that will not explain much variance when you are not
convolving it with your basic function. I guess, you expect an lower or higher BOLD signal in relation to the rating, but this, of course, has the same hemodynamic properties as the BOLD response to your trial. Additionally, you have very long lasting trials, which you have to consider in your manual regressor, as well, while the parameteric-modulation option would consider this.
Good luck,
Karsten
> Hello,
>
> >We have an event related design containing 5 sessions, each session composed of 30 trials, each trial lasting 11
> >seconds, TR 2.5, resulting in 132 WHV per session. For each trial we obtain a rating (1-3) from the subject.
> >Therefore, in addition to comparing different trial types in a classical way, what we would also like to do is
> >perform a regression of the subjective responses with activation patterns. In essence, it would be much like
> >adding in the 132 realignment movement parameters. Our question, however, is that we only have 30 subjective
> >responses, one for each trial (rather than one value per WHV as with realignment parameters), and hence are
> >not sure how to set-up this kind of a regression.
>
> It has been suggested to simply pad the regressor file with 0's. But is it okay to use 0? Specifically, I would
> like to check whether or not SPM models a regressor value of 0 as nothing (null-value). That is, are onset
> scans in the regression that are assigned a value of 0 not included in the regression? I suspect not, but want
> to check with the community and know for sure.
>
> I haven't found a satisfying answer in SPM documentation, but any pointers would be immensely helpful as well.
>
> Many thanks!
> -Peter Hu
>
>
> phu AT bidmc DOT harvard DOT edu
> (trying to cut down my daily intake of spam)
--
-------------------------------------------------------------
Karsten Specht, PhD
Department of Biological and Medical Psychology
& National Competence Centre for functional MRI
University of Bergen
Jonas Lies vei 91
5009 Bergen
Norway
Tel.: +47-555-86279
Fax: +47-555-89872
[log in to unmask]
http://fmri.uib.no/
|