Dear Ning,
while it is most likely some physiological noise that shows in this contrast, filtering or modelling the noise might not be sufficient. I guess that you would expect some brain activation that is explained by this condition instead. If your model was properly set up and there is a measurable effect, this activation should somehow show up. Since it does not, I would suggest to review your model specification. Pay attention to timing details such as synchronisation between scanner and experiment, dummy scans, TR, slice timing (if applicable), onsets and durations. Even if you specified your design in "scans", the first scan will start at t=zero, the second at t=one.
Hope this helps
Volkmar
|