Hello, DCMers (Klaas, Will, et al.)
I have a question about justification for the selection of connections
within a DCM model.
Consulting primate literature and studies of effective/functional
connectivity, I have assembled a network of anatomical nodes and
connections among these nodes. This is not a fully interconnected network,
as not all regions were connected to one another, and not all connections
were reciprocal. However, all included connections are justified by
primate projection studies. I have also assembled a fully interconnected
network, as well as a much sparser, more "conceptual" network of
connections that include all of these regions, for the purposes of DCM
model comparison with these two networks. I created all three networks for
all of my control study participants, averaged them across the group, and
then performed a DCM model comparison between the 3 averaged networks.
If it turns out, after performing the DCM model comparison, that the
network I have assembled has a greater fit to the data, by both Akakies
and Bayesian criteria, is this sufficient justification for selection of
this particular model for a study in which I want to compare alterations
in effective connectivity within said network due to, say, brain disease?
I eagerly await a reply from the DCM experts!
Thanks!
Siri Sonty
|