Hi Liana,
Let me just make a few points:
1. Motion will tend to cause outliers, so the TRs above each of the
limits might entirely overlap.
2. Motion censoring censors the previous TR, as well as the current,
since we do not know when the motion occurs. Ah, you noted this.
3. Even with censoring the previous TR, high motion often occurs over
multiple TRs, e.g. with 4 consecutive high motion TRs, only 5 will be
censored, not 8. And if they were to start a run, it would only be 4.
4. The number of counted TRs per stim class are those with a non-zero
regressor (i.e. non-zero expected BOLD response), and BOLD response
periods tend to have overlap across regressors.
5. So 72 and 81 TRs censored per stim class (total of 153) just means
there were 5 (or more) TRs censored when both BOLD IRFs were non-zero.
It's really great that you are paying such close attention to the scripts,
results, output and even program help!
I would be happy to provide more details if you would like them.
- rick