parametric modulation with
arousal
I'm analyzing a
study where different emotional stimuli were shown and the question of
interest concerns areas whose activity covaries with arousal.
Our first (random effects) analysis used categorical contrasts
between fixation, low, med, and high arousal conditions (based
on normative data). However, we also collected arousal ratings
from each subject for each stimulus mini-block, and I would now like
to follow up with another random effects analysis based on
individuals' ratings.
From reading the
parametric modulation paper cited on the SPM site, I decided to model
the experiment with one condition (stimulus: all 3 stimulus types
together, vs. the implicit fixation baseline) and parametric
modulation by the ratings (0, 1, 2, or 3). Using the linear expansion
as a starting point, I examined the [0 1] contrasts for each subject
and used a one-group t-test to carry out the 2nd level
analysis.
Is this analysis
sensible? The user-specified regressor option also seemed an option,
but it is scan-based vs. condition-block based and it was not clear
how one would use it here.
regards,
Stephan
--
Stephan Hamann, Ph.D.
Department of Psychology
Emory University
Atlanta, GA 30322
voice:(404)727-4261
fax: (404)727-0372