Hi,
I recently had a discussion with Rik Henson about possible ways to
incorporate our behavioural data in the analysis of fMRI data.
We have an experimental design which is as follows:
Rest - Stimulation - Rating - Rest - Stimulation - Rating - .....and so
on.
(during each rating epoch subjects rate the experience of the previous
stimulus epoch)
When designing the fMRI model I used 2 conditions (stimulus (A), rating
(B)) and input the behavioural data acquired during the rating epoch via
parametric modulation (other -> linear) of condition A ("trial 1").
Thus for each subject the design matrix has three columns:
A PARA.MOD. B
A PARA.MOD B
A PARA.MOD B ...........etc
My questions relates to comparison of the activity in A with the
PARA.MOD data.
1.) Am I right in assuming that the parametric modulation step is
essentially a linear regression of activity in A with the behavioural
data?
2.) In comparing A and PARA.MOD is it valid to proceed to a 2nd level
(RFX) analysis by making the appropriate con* images and performing a
paired t-test for all subjects?
Hope you can help.
Yours,
Jon.
Jonathan Brooks (Ph.D.) - Research Fellow
MARIARC, University of Liverpool, Liverpool, UK
Tel: +44 151 794 5629 Fax: +44 151 794 5635
Web: http://www.mariarc.liv.ac.uk
|