Hi ye,
> Suppose i am not to use any fixed model. With 3ddeconvolve
> options: for blocks--stim_minlag 0, stim_maxlag 0. for events--
> stim_minlag 0, m_maxlag 6.
If you analyze your blocks this way, you are assuming a fixed hrf for them. And you would have to run 'waver' and then provide those output files from 'waver' as input for 3dDeconvolve.
> Then for GLT, i would like to compare between those stimuli , the
> glt file is (say there are 2 runs): 0 0 0 0 1 -1 -1 -1 -1 -1 -1 -1
Are you comparing one block with one transient event? If so, the glt vector looks fine, but you may want to start with a quadratice fitting (-polort 2) for the baseline and drift instead of linear.
Gang