Hello Nicolas:
A couple suggestions:
Have you tried using the raw binary sequence of 1's and 0's as the input
stimulus function (i.e., without pre-convolving with waver)? So, with
a block design (20 sec. ON, 20 sec. OFF, 2 sec TR), the input stim. function
would look like:
1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 etc. etc.
In this case, you could use (say) maxlag = 5. The assumption is that the
system can be modeled as a linear time invariant system, which assumption
is probably more appropriate for random event-related designs than for
periodic block designs. However, note that even for a linear system, depending
on the data, there can still be overshoot or ringing in the fitted response.
Alternatively, you could simply average the response over the separate blocks.
In this case, the input stim. function would look like:
1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 etc. etc.
Here, you could use (say) maxlag = 15. The accuracy of the fit will depend
on the similarity of the response across blocks.
As always, use the Deconvolution plugin to plot the fitted response on top of
the actual fMRI data.
Doug Ward