The effect vector specifies how the modulatory B matrix parameters will affect the A matrix parameters. If, for example, you have a clear baseline where you assume no modulation by B matrix parameters, you might want to model your experimental effects as [0 1], where 0 is the baseline condition and 1 is the experimental manipulation that you assume to have an effect on the baseline. This means that for the second condition, the B matrix will be added to the A matrix to determine the connectivity weighting under your experimental manipulation. Alternatively, if there is an implicit baseline that your experimental conditions might up- and downregulate, your effect vector can take a form of [1 -1], where the B matrix parameters will be added to the A matrix parameters for your first condition, and they will be subtracted from the A matrix parameters for your second condition. Similarly, if you have more than one condition and you assume a parametric relation between them, you can specify your effect vector accordingly, just as you would specify a parametric contrast in a GLM.

Best regards,

Ryszard