Hi spm pros

first, thanks to roberto for answering my question (totally correct).
Therse's another problem:

What can I do when my condition starts before my first scan? In a resting
state experiment the BOLD signal does not follow an overshot at scan # 0.
The BOLD fluctuation in this state started ~half a minute earlier. The
regressor should be convolved since the condition alters later (eyes
open(closed).

What I should have is a regressor that gets through the standard procedere
but starts 20 seconds before the first scan of the analysis. I tried a
negative value as onset but it gets interpreted as zero in spm.

Thanks for any help

Rafael