Print

Print
Dear Vladimir 

Thanks a lot for the quick reply. I could understand this part very well because of your reply.

I have a query in continuation to the same. Now if we have two conditions C1 and C2 and we model the condition specific effects. In that case after we have got our winning model (by BMS) we extract the DCM.Ep.A and DCM.Ep.B and also for C for the winning model. 
Now my query is on these condition specific effects.
(1) What does A matrix show in terms of connection strength. Does it shows the connection strength for average of two condition C1 and C2 (i.e C1+C2)?. Or does the A matrix shows the connection strength for difference waves of two condition (i.e.  C1-C2)?.
(2) What does the B matrix show in terms of connection strength modulation. Is the parameter obtained by fitting the model to Condion C1 minus Condition C2. Suppose I get the value of A(1, 2) = 0.07 and B(1, 2)= 0.02. Does that mean that for condition C1 the strength of a connection = 0.07 and for C2= (.07)+ - (.02).

As I am new to DCM and couldn't clarify this clearly from manual, your answers will clarify my doubt.

Regards
Rajan Kashyap

On Thu, Jan 14, 2016 at 6:19 PM, Martin Dietz <[log in to unmask]> wrote:
Dear Rajan,

Please see Vladimir's response. This should be perfectly fine.

Best
Martin

On 14 Jan 2016, at 09:49, Rajan Kashyap <[log in to unmask]> wrote:

Dear Martin

Since I have not received any reply from the group, so I am emailing you for a guidance on the below mentioned query that I have. (Attached email has the query in detail)
In simple words I want to know that Can I fit different models to only one condition ERP data and then find the best model of all the fitted models by BMS. In general people always fit a model to conditional effect as shown by between Trial Button in SPM GUI. But my interest is to fit a model just to one condition and then check the average parameters from A matrix.

Please Guide, as its important for me to understand the concept of DCM.

Regards
Rajan
---------- Forwarded message ----------
From: Rajan Kashyap <[log in to unmask]>
Date: Wed, Jan 13, 2016 at 3:45 PM
Subject: A basic question: A different approach
To: "[log in to unmask]" <[log in to unmask]>


Dear SPM users

Let me clarify my question in more details.

Suppose we have the ERP data of one condition only. According to the GUI of SPM, we first select the Kind of model as per our data and then we select the time window. Now in the "between Trial effects", since we have only one condition data, we put 1 there and therefore don't model any condition.

In this case after filling in the other required buttons and clicking in invert DCM we can have a model fitted to one condition data. Similarly I can fit another model and then compare the model fits by BMS for many other models and find the best fit.

If my understanding is correct, in this way if somebody is not interested in between trial effects, then one can find an optimum model for one condition.

Please let me know if i am wrong to take this approach.

regards
Raj
---------- Forwarded message ----------
From: Rajan Kashyap <[log in to unmask]>
Date: Wed, Jan 13, 2016 at 11:08 AM
Subject: A basic question
To: "[log in to unmask]" <[log in to unmask]>


Dear DCM experts

We were performing a experiment on patients where we could collect ERP data for only one condition. 
Now we want to construct various family of models and see which model best fits the data of that condition. Finally we want to get the connection strength of the winning model.

Can we do it for only one condition?
How can we do it ?

Regards
Raj