Hi all,
I'm doing DCM analyses following the SPM tutorial and examples. The model I create on purpose is very similar to the one for the "Attention to Visual Motion fMRI dataset", with the exception that my timing unit is not "scans" but "secs" (it's event-related). I therefore use duration=0 for my three regressors. The model runs until it's time to create contrasts: at these point, regressors look correlated, and so no contrasts are possible. If I change the duration of the first regressor to any number (e.g., 18, the max possible length of a trial) or leave it at 0 and model the other two regressors with actual durations rather than stick function, everything works fine until Bayesian comparison and beyond.
What am I missing or doing wrong? And does any of the solutions I found make sense? Am I still able to capture HRF properly?
Many thanks for all your help in advance,
Manuela
|