 |
 |
Dear Sandra, Yes, I think I didn't receive any responses indeed. However and as far as I recall the obeservation "this produces different results" was due to different slice time settings which are implicitely expected in that script from Wager lab, which I hadn't adjusted to my settings before. In your case, make sure that the time courses of the fixed-duration and variable-duration regressors are constructed the same way. Once you have the two time courses (already convolved with the canonical HRF or any other function you want to rely upon) you can go with spm_orth. The first column remains the same, the second is adjusted. In the final model you only need these two columns (leaving aside any other condition regressors, realignment parameters, dummy regressors, ... of course). The variable-duration regressor will account for activation changes unexplained by the other regressor then, as the shared variance is attributed to the first one. Best Helmut