I would just like to add 2 thoughts to Peter's good reply.
1. Maybe 3dttest++ is a more useful choice, since it
additionally gives you the mean and t-stat for that
mean. The t-stat for the covariate should be directly
convertible to/from the correlation, meaning they are
equivalent.
So basically, 3dttest++ gives the same plus more output.
2. Assuming you go with 3dTcorr1D, consider using
3dTcat to create the input, rather than 3dbucket. It has
the nice little advantage of accepting both wildcards and
sub-brick selectors, e.g.
3dTcat -prefix AllSubjects 'Subject*+tlrc[condition1#0_Coef]'
Note the quotes around the entire name (including both
the * and the [ characters).
- rick