 |
 |
Dear Rob, No, this should be alright. Adding a parametric modulator means you obtain an additional regressor (if you chose 1st order) predicting linear changes in amplitude relative to the average response for that condition. And the derivatives are incorporated for any of the regressors convolved with the HRF. Based on the description this will indeed result in 8 + 4 regressors reflecting the canonical HRF (8 for the average response cond1-8, 4 for the parametric modulators), 8 + 4 regressors for the time derivative, 8 + 4 regressors for the dispersion derivative, 6 movement parameters plus a constant. The "conditions per session" is misleading, it just refers to "conditions" specified in the GUI, not to the no. of predictors. "regressors" from the GUI (= predictors which are not convolved) are never considered as "conditions", nor are PMs or derivatives. The order should be Sess1 Cond1 x HRF Sess1 Cond1 x Time Sess1 Cond1 x Dispersion Sess1 Cond1 x PM x HRF Sess1 Cond1 x PM x Time Sess1 Cond1 x PM x Dispersion Sess1 Cond2 x HRF Sess1 Cond2 x Time Sess1 Cond2 x Dispersion Sess1 Cond2 x PM x HRF Sess1 Cond2 x PM x Time Sess1 Cond2 x PM x Dispersion, [repeat for Cond3, 4] Sess1 Cond5 x HRF Sess1 Cond5 x Time Sess1 Cond5 x Dispersion Sess1 Cond6 x HRF Sess1 Cond6 x Time Sess1 Cond6 x Dispersion Sess1 Cond7 x HRF Sess1 Cond7 x Time Sess1 Cond7 x Dispersion Sess1 Cond8 x HRF Sess1 Cond8 x Time Sess1 Cond8 x Dispersion Sess1 R1 Sess1 R2 Sess1 R3 Sess1 R4 Sess1 R5 Sess1 R6 [repeat for session 2-5] Sess1 Constant Sess2 Constant Sess3 Constant Sess4 Constant Sess5 Constant Or in short, first the regressor for session 1 condition 1, followed by its derivatives, followed by the PM regressor and its derivatives, then session 1 condition 2 and so on, followed by session 2 and so on. Constant terms indicating different sessions are always the last few columns of the design matrix. Best Helmut