Hi Gang,

You are right, now for SPMG1 I have only one regression coefficient for I and B and two for S (and for SPMG2 I have only two regression coefficient for I and B and four for S).

So in this case, if I still want to see the contrast between the S regressor and I/B regressors considering the parametric modulation effect, what will be an appropriate approach?

1. for SPMG1: I should do S[1] - I[0] OR will be better to do S[1]-I[1] including the parametric modulation for the regressor I too (by using the same parametric modulation value which is used for the S regressor, since both regressors - S and I - are composing the same trial from the run for the subject, just the task of I and S regressor is different)? Also, should I consider the S[1] - B[0] OR will be better to do S[1]-B[1] including the parametric modulation for the regressor B too (again, using the same parametric modulation value which is used for S regressor)?

2. for SPMG2: I should do S[2]-I[0] and S[3]-I[1] OR will be better to do S[2]-I[2] and S[3]-I[3] including the parametric modulation for the regressor I too, like for SPMG1? Also, should I consider the S[2]-B[0] and S[3]-B[1] OR will be better to do S[2]-B[2] and S[3]-B[3] including the parametric modulation for the regressor B too, like above?

Thanks in advance for your precious answers!

Ruben.