Dear Liang,
Sometimes DCM completely fails to fit the data and you should be able
to see that by looking at the 'ERPs (mode)' plot to see whether the
prediction fits the observed data well. You might see that the
prediction is flat. There are several possible reasons for that.
Basically DCM-ERP is optimized to fit ERPs about 500ms long with most
power in the alpha range. If you have something very long or with slow
drifts DCM might be unable to fit it or may require more than one
external input. If this is the case for your data try to increase the
'detrend' parameter if high-pass filtering your data prior to DCM. You
might also try shortening the time window and using the 'hanning'
option to limit your ERPs in time.
Best,
Vladimir
On Fri, Dec 30, 2011 at 10:26 PM, liang wang <[log in to unmask]> wrote:
> Hi DCM experts,
>
> I run my electrophysiological data using DCM for ERP. First I used a script
> included in SPM8 to convert the data into SPM format and then setup all the
> parameters required and the assumed connections. However, after estimating,
> I found all the posterior probability for the connections (A,B,C matrix
> shown) is 0.5. Initially I think maybe it is due to local minimum
> convergence. I tried to continuously use the estimated parameter to estimate
> again, but the results are the same (i.e., 50% still). Could anyone who
> experiences this give some solutions about this. Thanks.
>
> Best,
> Liang
>
> --
> Liang Wang, PhD
> Neuroscience of Attention and Perception Laboratory
> Princeton Neuroscience Institute
> Princeton University
> Princeton, NJ, 08540
>
>
|