Dear Susanne,
you should compare all different models to each other (or families of
these) in order to test, which model most likely generated the data.
This can be done with the Bayesian model selection (BMS) procedure
implemented in SPM.
The endogenous connectivity parameters (A matrix) usually reflect the
average coupling strengths between regions (i.e. under average
perturbation with respect to your experiment).
So you are right that the "parameters of the C matrix scale the direct,
extrinsic influence of inputs on brain states in particular regions" and
the "intrinsic [or endogenous] connection parameters refer to the
coupling of neural states in different regions".
In order to estimate if the impact, that one region exerts over another,
changes under the influence of the experimental conditions, however, you
need to specify these in the matrix B, containing the modulatory inputs.
This is exactly what the modulatory inputs do: esimating the rate of
change in the coupling that one region exerts over another - depending
on external input or context.
*All* connectivity parameters in DCM pertain to effective
connectivities.
Best wishes,
Thilo
On Sat, 2012-09-15 at 16:31 +0200, Susanne Dietrich wrote:
> Dear SPMers,
>
>
>
>
>
> I get distinct intrinsic connectivity patterns, when I use different
> stimulus categories as input (C matrix) into a fully connected model
> (A matrix). Thereby, modulatory input (B matrix) has not been defined.
>
>
>
> How should such kind of DCM be interpreted? Is it a functional or an
> effective connectivity model?
>
>
>
> The parameters of the C matrix scale the direct, extrinsic influence
> of inputs on brain states in particular regions. The intrinsic
> connection parameters refer to the coupling of neural states in
> different regions.
>
>
>
> Allow these parameters to estimate the impact that one neural state
> exerts over another under the influence of the experimental conditions
> (C matrix)?
>
>
>
> If yes, can I say that the connectivity is effective?
>
>
>
> Susanne
>
>
>
|