Thanks Rick,
I tired using 'BLOCK(2.5,1)' and it worked well. I initially added the ITI in as a regressor of no interest, and was using the first 12.5s of rest from run 1 as the baseline (which was not modeled). However I have since re-run the analysis using the initial rest + all ITIs as baseline.
The results are not drastically different, but any suggestions on which approach is most appropriate for my design?
Thanks again,
Steve