Dear Hongyu Yang,
There is nothing wrong with the posterior parameter estimates greater than 1. Note that these are not correlation coefficients, but time rates thus in theory they could take any value. In the DCM as defined in Friston (2003) the posterior values will also depend on the prior parameters that will be defined as such that the dynamical system after a perturbation converges to a stable state.
Greetings
Branislava
P.S. You are surely using a DCM10 in place of DCM8.
|