Dear Gaby,
Maybe you can clarify what you're trying to do? Do you want to set up a model with (among others?) a factor "Basis function" (canonical HRF, time derivative, dispersion derivative)? Probably you would better go with three separate models for each of the basis functions, as usually, you don't want to test whether the estimate for the canonical HRF differs from that of the derivative (and then it should be straightforward to specify a covariate).
Depending on your aims / hypotheses it might also be interesting to integrate the different estimates into a single one, see Calhoun et al. (2004, http://dx.doi.org/10.1016/j.neuroimage.2003.12.029 ), Steffener et al. (2010, Neuroimage, http://dx.doi.org/10.1016/j.neuroimage.2009.11.014 ) and Pernet (2014, Front Neurosci, http://dx.doi.org/10.3389/fnins.2014.00001 and https://github.com/CPernet/spmup/blob/master/spmup_hrf_boost.m ).
Best
Helmut
|