Hi Andrew,
If you mean the warnings from 1d_tool.py in the afni_proc.py script,
the "IDENTICAL", "high" and "medium" warnings are just that, warnings.
It does not know whether they are problems, but you are warned to at
least think about it.
It is basically impossible for the program to "know" whether there is
a problem.
By default, afni_procp.py gives various warnings for anything above 0.4,
as you mentioned.
---
Note that the constant polort terms are special. Ignoring any censoring,
and assuming all run lengths are the same, then the correlation between 2
constant polort (Pol#0) terms depends only on the number of runs. For
R>1 runs, r = -1/(R-1).
So for 2 runs, r = -1.0, and for 3 runs, r = -0.5. Those will both
generate warnings, but they are not important. I could take out the
warnings for such terms, but I find them educational.
In any case, do not worry about warnings for the constant terms.
Warnings for regressors of interest especially are worth thinking about.
- rick