We are trying to test several a priori hypotheses
about connectivity in hemianopes, using DCM (SPM5). I
have read several papers on DCM and gone throught the
practical example in the manual, and feel I have a
fair conceptual understanding, but I have some
practical questions that I hope some of you can help
me with. I've searched through the archives, but
haven't come across much that relates to my questions.
If there is other documentation that someone can
suggest (in addition to the Friston papers and the
manual), that would be great as well.
In the output DCM structure, I don't think I
understand what some of the components are. According
to the manual, DCM.A contains the intrinsic
connections, DCM.B the modulatory inputs, and DCM.C
the driving inputs. Are the DCM.pA, DCM.pB, and DCM.pC
then p-values for these parameter estimates? Or are
these Bayesian posterior probabilities or something
like that? Am I hoping for high or low for these
numbers? The practical example in the manual doesn't
really go over this, it procedes directly to testing
two models against each other.
There are also many other components of the DCM
(i.e., DCM.Ep). What are these things? AIC and BIC is
obvious, but what is vA, vB, vC, K1,K2, etc?
Any other guidance in interpretation of output or
things to watch out for would also be appreciated.
Thanks in advance.
Tim Martin, Ph.D
University of Rochester Medical Center
Department of Ophthalmology
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around