Hi Haisam,
I probably do not understand fully, but it seems like the first
case is quite appropriate. There is one consistent task that
sometimes occurs before a condition and sometimes after. It is
quite reasonable to break this into 2 separate regressors as in
the first case.
There is no post-condition of the task in the "pre" time period,
and there is no pre-condition of the task in the "post" period.
If you do not agree, maybe you can provide more details about
what you think the problem is.
- rick