Hi Brenda,
These warnings can be ignored (via -GOFORIT).
The basic problem is simply that you have exactly 2 runs. In the case
of exactly 2 runs, the baseline regressors will be extremely collinear
(even though they may be orthogonal).
For example, with exactly 2 runs, the constant baseline terms will be
"perfectly" collinear, with an r value of -1, since they are effectively
negatives of each other, once the mean is removed.
The warnings seemed like they were all for the baseline (verified using
'1d_tool.py -show_cormat_warnings' on your X-matrix).
Adding -GOFORIT will be necessary, but not harmful.
Note that if your stimuli vary over subjects, you would want to verify
that the only issue is in the baseline for each subject.
- rick