I ran DCM on a block design with three conditions (so my DCM model has 4 conditions -- driving input + 3 experimental conditions). I have plotted the resulting DCM.U.u for the three experimental conditions and see that they correspond to the experimental timings for the three conditions. I noticed, though, that for each condition, DCM.U.u alternates between two non-zero, non-integer numbers, one positive (corresponding to "on" blocks) and one negative (corresponding to "off" blocks).
My question is: in order to use the scaling factor "u" to calculate linear combinations of the A and B matrices (per the DCM equation), do I just multiply the values in my B matrices to the value in DCM.U.u corresponding to the "on" blocks, or do I need to use some combination of both the "on" and "off" values in DCM.U.u?
Thanks in advance,