Dear Fabienne,
There is large amount of anatomical evidence that M1 is agranular. This means that it lacks a segregated layer 4 which is, canonically, the target of feedforward connections. From the point of view of the neural mass models used in DCM, this means that a connection targeting M1 is less likely to be feedforward. Rather, the nature of the connections between motor areas is more likely to look like a lateral connection. In DCM for evoked responses (ERP neural model), forward, backward and lateral connections are different in nature and you should therefore decide whether you want to model interhemispheric connections using one or the other type of connection, in light of the anatomical evidence. Using the B-matrix, you can model the gain or change in the strength of any connection specified in the A-matrices. You would also use the B-matrix to model changes in intrinsic connections.
I hope this helps
Martin
On 18 May2015, at 17:23, Fabienne Jung <[log in to unmask]> wrote:
> Dear Martin,
>
> just one more question. I think I still don't get it :-|
>
> If I'd like to model bidirectional connections without hierarchy, would I use 4 connections for both M1 cortices (forward, backward, forward, backward) in the A matrix or a lateral connection? Is it the same? From a logical perspective, I would say yes. Does it make sense to model a B effect with 4 connections or even lateral connections?
>
> Best,
> Fabienne
>
> -----Ursprüngliche Nachricht-----
> Von: Martin Dietz [mailto:[log in to unmask]]
> Gesendet: Montag, 18. Mai 2015 15:56
> An: Fabienne Jung
> Cc: [log in to unmask]
> Betreff: Re: [SPM] DCM for ERPs: Specification of connections, neural model
>
> Dear Fabienne,
>
> If you are unsure about the type of connection, you should turn your doubt into a model comparison problem and simply compare a model with lateral connections to a model with forward and backward connections between left and right M1.
>
> Best wishes
> Martin
>
>
>
> On 18 May2015, at 15:37, Fabienne Jung <[log in to unmask]> wrote:
>
>> Dear Martin,
>>
>> thank you for your response! For me, it just makes more sense, that there is no hierarchical connection between the motor cortices, before the go stimulus for a movement is presented. That's why I initially preferred a lateral connection in the A-matrix. Isn't this possible?
>>
>> Best
>> Fabienne
>>
>> -----Ursprüngliche Nachricht-----
>> Von: Martin Dietz [mailto:[log in to unmask]]
>> Gesendet: Montag, 18. Mai 2015 15:21
>> An: Fabienne Jung
>> Cc: [log in to unmask]
>> Betreff: Re: [SPM] DCM for ERPs: Specification of connections, neural model
>>
>> Dear Fabienne,
>>
>> In this case, you should simply specify a forward and a backward connection between left and right M1 in the A-matrices and specify the same connections in the B-matrix.
>>
>> Best wishes
>> Martin
>>
>>
>>
>> On 18 May2015, at 15:09, Fabienne Jung <[log in to unmask]> wrote:
>>
>>> Dear all,
>>>
>>> I have a question regarding the specification of the connections in DCM for ERPs:
>>>
>>> is it possible to have a lateral connection between for example left M1 and right M1 that changes into a forward/ backward connection with stimulus presentation in the B-matrix? That means the B-effect is forward from right to left and backward from left to right?
>>>
>>> Thank you in advance!
>>>
>>> Best
>>> Fabienne
>>
>
|