Hi Vitaly,
> is your suggested approach different mathematically from my approach
> A:
>
> y = A1x1 + A2x2 + e
>
> i.e. for your 3 column suggested approach is there still 2
> regressors in the
> GLM, one for mean vanilla effect and the other for the parametric
> modulation? using the 3 column approach is the model treated
> differently
> somehow than just plugging in 2 EV's?
No, in my suggestion there are still just 2 regressors. I guess my
point was that you have to make sure you create your regressors in the
right way. Basically what you want is to have one vector of "sticks"
of equal height and one regressor with "sticks" of varying height. And
then you want to convolve those with the same HRF (and possibly the
temporal derivative of the HRF, taking the # of EVs to 4).
Good luck Jesper
|