Dear Manuela
Thanks for sending this. This isn't a valid design matrix, because the first regressor is the sum of the second and third regressors, making it rank deficient. In general, you can't compute a contrast where any of the regressors involved are the linear combination of other regressors. I am not quite sure how this worked when you had trials of zero duration. I appreciate that you copied the attention example, but unfortunately that's a bad example of how to generate an efficient design matrix, for the same reason.
Please could you tell us a bit more about your experimental design? (Depending on the design, you might want to use a more efficient design matrix for extracting timeseries, and then use this rank-deficient design matrix to specify the input timing for the DCM.)
Best
Peter
-----Original Message-----
From: Manuela Sellitto <[log in to unmask]>
Sent: 11 March 2020 10:28
To: [log in to unmask]; Zeidman, Peter <[log in to unmask]>
Subject: Re: duration regressors DCM analysis
Dear Peter,
thanks for your reply. Attached you can find a pic of my design matrix.
One participant, three runs, three regressors per run. It is a hierarchical model, the first regressor contains the onsets of both the second and the third regressor.
Cheers,
Manuela
|