@rotemb9 suggested this while trying to create NARPS model

"2. The model I was working on has a quite common feature: it uses a different duration for the regressors than the duration of the trial given in the tsv file (specifically, it uses the average TR). Jeanette told me it can be done by assigning the desired duration to the variables, but we should definitely show examples of that for users because this could be confusing."