Dear Suzanne,
I am not sure that I understand your question. You created the DCM's (basically you crated models which need to be estimated). The inputs are in SPM.U.u. They are defined by your SPM design I presume?
Thus in your SPM design, for on conditions you use positive numbers, and for off condition negative.
Usually for DCM the SPM design is created so that 1 is attributed to inputs (starting at onset, long as duration) and 0 for no input.
There was a half a year ago (up to a year) discussion on the spm-mailing list about the effects of using other values for inputs. As I remember it complicate to rethink (calculate) how different scaling of input affects results. Probably the easiest is to create a separate SPM design with inputs adjusted to [0,1] coding.
I hope this helps. Out of curiosity, why would you calculate linear combinations of A and B matrices?
|