Dear SPMers,
I have a study in which all subjects show a similar behavioural effect
relating a dependent variable (VarDep) with an independent variable
(VarInd). This effect was nicely modeled with a sigmoidal curve. Now I
want to know if some brain regions display the same pattern of
activation, i.e. if their response pattern to VarInd follows the same
sigmoidal shape. I'm thinking about two methods, but I'm not sure which
one will give me the most accurate answer.
The first method would be to model the brain response as a sigmoidal
function of VarInd at the first level, using a parametric regressor, and
then look at the corresponding Beta (or T-stat) at the group level
(reflecting a linear correlation between the brain response and the
"sigmoidized" VarInd).
The second method would consist in modeling each level of VarInd as a
separate regressor at the first level, extracting the corresponding
Betas in every voxel, and regressing these values against the
behavioural sigmoidal curve at every voxel. After proper group averaging
the resulting map would then be a Rē-map.
The first method only gives me a Beta (and also a T-stat) at each voxel,
which reflects -if I'm not mistaken- the slope of the linear fit between
the BOLD and the "sigmoidized" VarInd, and hence gives no hint on the
goodnes of fit. On the other hand the second method gives me a Rē at
each voxel, and seems to better answer my question, but the results I've
got so far show very weak correlations...
Am I right about the interpretations of the results of these two methods
? Am I missing something ?
Thank you very much in advance for you advice.
Guillaume Sescousse
--
___________________________________
Guillaume Sescousse, PhD student
'Reward and decision making' group
Centre de Neuroscience Cognitive
CNRS UMR5229 - UCB Lyon 1
67 Bd Pinel, 69675 Bron, France
tel: 00 33 (0)4 37 91 12 38
fax: 00 33 (0)4 37 91 12 10
http://www.isc.cnrs.fr/dre/
___________________________________
|