Dear all,
I am wondering how SPM8 handles the multiple regressors part of the design matrix differently from the multiple conditions part, aside from mean-centering and not-convolving with the HRF.
Background: for my blocked design study, I need to model a response that gradually increases over my ~2 min. blocks, rather than an on/off boxcar response.
I construct a function to model this 'manually', and feed it into the multiple regressors part of the design matrix. I convolve with the HRF before feeding the regressor to 1st level, and I turn off automatic mean-centering for this regressor.
To test whether my home build is valid, I also home-built a box-car model and compared it to the same model built the conventional way. Apart from the order of conditions/regressors, the design matrices are exactly equivalent (as far as I can tell from the 'review' overview); the HRF convolution also seems in order. However, when I test them, they provide slightly different maps (on 1st level).
It seems that I am missing something here, does anyone know what it might be?
Thanks for your help, best,
Vincent
|