Your block effects will run for about 20s after the last block ends. Thus, the block regressor runs into the next run. This is problematic and is not addressed by using spm_fmri_concatenate. The only concatenation code that truly builds the correct concatenation model is encoded in the gPPI toolboxes function spm_estimate_ppi. If you are good at MATLAB coding, you can extract the relevant sections and make stand alone concatenation code. Alternatively, you could just model each block separately and then correlate the block amplitudes with ERP data. With having multiple runs, you could have differences in block signals unrelated to the ERP amplitude and would need to account for that possibility in the correlation approach. You could use the slope of the correlation approach at the group level. Best Regards, Donald McLaren, PhD On Wed, Aug 5, 2015 at 8:40 AM, Ilaria Mazzonetto < [log in to unmask]> wrote: > Dear Donald, > Thank you for your prompt reply. I am working on EEG-fMRI integration and > I would like to add a parametric modulation with a specific ERP amplitude. > I know run concatenation is not a trivial matter because of AR(1) term, > high-pass filter and linear trends, but I found the function > spm_fmri_concatenate that suits me well, I think. Furthermore I have > about 10 s from the beginning of a session and the beginning of the first > block of that session and from the end of the last block in a session and > the end of that session, so if I concatenate the sessions, I will not have > HRF going across runs block. What do you think? > Thank you for your attention to this matter. > Best regards. > Ilaria > > >