Dear Felix,
If I understand your enquiry correctly, it sounds similar to a scenario described in our paper Rae et al (2015) here https://www.ncbi.nlm.nih.gov/pubmed/25589771 and also a similar approach taken by Ewbank et al (2011) https://www.ncbi.nlm.nih.gov/pubmed/21490204.
In both these DCM analyses, the onset of all events is coded in the first column of the design matrix, and applied as C-matrix driving input.
The experimental manipulation of interest is then set as a parametric modulator of the all events regressor in the first design matrix column. It would therefore appear in the following column of the design matrix to the 'all events' regressor. (e.g. 'stopping trials' vs 'go trials', [1 -1] in Rae et al; 'greater change during same-identity blocks compared with vary-identity blocks and vice versa' in Ewbank et al). This parametric modulator of 'all trials' in the following column of the design matrix can then be applied as a B-matrix modulatory input.
Apologies if that is not quite the scenario you are after but hope that is of some help
Best wishes
Charlotte
---------------------------------------------------
Dr Charlotte Rae
Research Fellow
Sackler Centre for Consciousness Science
Brighton & Sussex Medical School
University of Sussex, Falmer
[log in to unmask]
---------------------------------------------------
|