Dear SPM-Experts,
I am trying to setup a meaningful DCM model, therefore I have read some
papers for understanding how to include inputs (both direct and modulatory)
in the model. I have read the very intersting paper of Noppeney et al (Cereb
Cortex 2006;16:437-445), where a DCM model was built for analyzing changes
in effective connectivity for category-selective responses. However, it is
not clear to me how DCM inputs were defined. In the "Conventional analysis"
section, it is written that 18 conditions were modeled, according to the
2x3x3 factorial design. Then, in the "DCM analysis" section, it is written
that a visual input (words written + pictures) was connected to OCC, an
auditory input (words spoken) to STG, ... and so on for modulatory
connections. My question is: is this "visual input" the sum of 12 of the 18
conditions defined in the "Conventional analysis"section? (i.e. Tools
pictures 1st task + Tools words written 1st task + Animals pictures 1st task
+ Animals words written 1st task + .. and so on for 2nd and 3rd task). If
yes, did you create a new condition in the SPM.mat, calling it e.g.
"visual"? Did you have any problem by estimating it? I had some problems
when I defined an SPM.mat where one regressor was a linear combination of
other regressors. SPM gave me an error and did not estimate any contrast,
saying that beta parameters were not uniquely identified.
Thank you very much for any help in this issue
Best regards,
Paola
|