I am studying resting state MEG with csd-DCM, and I am interested in dynamics of effective connectivity, thus I estimated a large number of time-window level DCMs.

However, there are too many underestimated DCMs, which show under 80% explained variance, therefore I want to re-estimate those underestimated DCMs with better prior.

For saving computing time, I will only re-estimate underestimated DCMs.

When I tested with two DCMs, different prior results the spm_dcm_peb.m code error, and all of the second level parameters have NaN values.

At this point, I have two questions.

1. Is it possible to running spm_dcm_peb.m code with DCMs using different priors?

2. In order to match the priors, can I use spm_dcm_reduce.m or spm_log_evidence_reduce.m code for changing prior and posterior of re-estimated DCM?

If the two priors have the same size of parameters, I guess that the code might be working, but I am not sure that this is the right way to do.

I would be thankful that you could help me solve this problem.

Best Regards,

Junho