Hi Rik,
Let me have a go at your questions.
See interleaved text here under.
Le 17/03/2011 11:09, Rik Henson a écrit :
Dear DCMers –
A few, hopefully simple questions for DCM(10) for fMRI:
1. Does it make sense to have a model with 4 regions, in
which 2 regions within each of 2 pairs are interconnected,
but there are no connections across pairs (ie, two
“isolated” subnetworks; eg, with regions E1<->E2
F1<->F2, DCM.a = [1 1 0 0; 1 1 0 0;
0 0 1 1; 0 0 1 1])?
Yes it does.
You're "simply" modelling 2 parrallel and independent processes a
single model.
2. A related question to above: does it make sense to
compare the free energy of two models, one in which a region
is “isolated” (ie has no connections apart from a
self-connection) with one in which it has connections to
other regions? And could this
be used to ask whether that region needs to be in the model?
(I realise one cannot use the free energy to compare models
with different regions – ie different data – but wonder
whether this approach could be a useful heuristic answer to
that question?).
The residual term of the isolated region will be its own signal (no
drive -> flat modelled activity).
So your question would rather be: is the activity in my
last/isolated region better explained, or not, when it is driven by
the rest of the network? The model comparison will thus be between a
simple model where the activity of one area is not explained at all,
and another one with an extra parameter and a bit more signal
explained.
This doesn't really answer your question whether the region should
(or not) be part of the network though...
I would say that choice of regions to include in the model is more
empirical: you should include the areas necessary to build a model
which models as accurately as possible (or sufficiently
realistically) the "brain function" you want to study.
3. Does it make sense to have a model with no modulations
(eg, DCM.b = zeros(4,4,2), for two inputs)?
Yes.
You would then be comparing the possible different intrinsic
connectivity of a network.
4. If the GLM in an SPM.mat file has two event-related
regressors for conditions G and H (and a jittered SOA so
that responses vs the inter-event baseline are estimated
efficiently):
4.1 does it make sense to use
one of these as a driving input (eg, to both of two regions,
eg, DCM.c = [1 0; 1 0]) and the other as a modulatory input
(eg, DCM.b(:,:,1) = zeros(2,2); DCM.b(:,:,2) = [0 1; 1 0])?
An event-type modulatory input would only modulate the intrinsic
connectivity for a very brief instant, while neuronal activity lasts
much longer. Mathematically it is ok but I don't think it makes much
sense.
I would be happy to hear Klaas (or other experts) opinion about
this.
4.2 if instead one wants to
make the driving input both G and H (ie treating any event
vs baseline equivalently), and modulate by just H, is it
sufficient to set:
DCM.U.u =
[full(SPM.Sess(ses).U(1).u)+full(SPM.Sess(ses).U(2).u)
full(SPM.Sess(ses).U(2).u)];
DCM.U.name =
[sprintf('%s+%s',SPM.Sess(ses).U(1).name{1},SPM.Sess(ses).U(2).name{1})
SPM.Sess(ses).U(2).name];
(and with DCM.c = [1 0; 1 0] and DCM.b(:,:,1) =
zeros(2,2); DCM.b(:,:,2) = [0 1; 1 0], as above), rather
than having to re-parameterise and refit the GLM? (ie, are
there any other fields in the DCM structure that would be
affected by this re-definition
of inputs, hence causing different results compared to
reparametrising and refitting the GLM?)
Yes from a coding point of view this seems ok. Only 1 driving input
which are the 2 types of events put together, i.e. with the same
drive on the network, and the 2nd input made of only one sort of
events.
Though the same worry as previous comment applies here: Does it make
sense to have an instantaneous modulation of connectivity ?
Apologies if some of the above questions illustrate an
incorrect understanding of the DCM code.
Sounds rather like deep understanding of DCM machinery!
HTH,
Chris
BW,R