Hi Mihaly,
In my opinion this is not a problem of symmetry. Either (i) try the new
version (DCM12), see similar problem in
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1205&L=spm&D=0&P=700815
or (ii) double check the extraction of your VOIs. I’m curious to know
why DCM.Y.X0 is empty in your models (on real fMRI data).
I hope this helps,
Mohamed
On 14/11/2012 15:26, Bányai Mihály wrote:
> Dear DCM users,
>
> I'm trying to estimate DCMs on fMRI data with DCM10, and the EM always
> converges in 4 steps, resulting in parameters very close to zero in B
> and C matrices with low p-values and uniform values in A.
> My models are defined with three areas fully (reciprocally connected)
> intrinsically, a single input, connected to each area, and what is
> varied in the model space is the effect of the input on intrinsic
> parameters (B), but always in such a way that the two edges of a
> reciprocal connection are both either affected or unaffected by the
> input.
> I suspect that my uninteresting results are a consequence of the
> highly symmetric model definitions. Can that be the case?
> I'm also not using any confounds (X0). Can that cause a problem?
>
> Thank you for any help,
> Mihály
>
|