Dear AFNI experts,

I have a quick question regarding specification of TENT function time window in GLM.

For now, I have the parameters set in a conventional way (e.g., TENT(0,20,11): to get the full response time course from 0–20s; 2s TR) after an auditory event. But as I increase the number of events (to model condition-specific events, such as animal versus speech sounds), modelling 0–20s adds a lot of regressors, which could potentially complain about multicollinearity.

So I was wondering whether restricting the time window of interest using TENT is a good approach (e.g., TENT (4,8,3)). I was uncertain about it because with the less number of regressors in the model, whether baseline be different from using longer-TENT vs. shorter-TENT. That is, the time points not included in the shorter-TENT model GLM (0-4s and 10-20 s) gets included as implicit baseline.? Just one note, I do not want to use HRF like GAM or BLOCK, because I would like to capture a sustained delay activity.

Your help would be greatly appreciated!

SJ