Self-connectivity in rDCM

[CogPsycho2023]

Hi,
I have a naive question about regression DCM.
I am going to implement both spectral DCM and regression DCM in my project now. As I understood from spectral DCM, the diagonal elements of matrix-A or matrix-B EC were log-scaled. Then I am thinking about if the regression DCM also has such a scaling transformation in the output.
Any responses would be appreciated.
Best regards,
Shufei Zhang

[ImreKertesz]

Dear Shufei Zhang,

the posterior estimates of the inhibitory self connections (diagonal elements of A matrix) are not log-scaled. The estimated quantities are the actual values of the A matrix.
Regarding the B matrix: rDCM is based on a linear DCM, therefore (in it's current implementation) it's not modeling changes in connectivity. In other words, the values of the B matrix are not estimated.

I hope this is helpful. It not, feel free to come back to us.

All the best,
Imre