> I want to do a correlational analysis for a seed region within condition A to
> time points A, and then a correlational analysis for a seed region within
> condition B to the B time points.

If you just want to find out the regions in the brain that have similar response to your seed region, then simple correlation is the way to go. On the other hand, if you'd like to find out the network underlying the similar response, you may want to try a different approach:

[afni.nimh.nih.gov]

> should I input my stim onset times, and use 'BLOCK'?

This is up to you to decide which basis function set to use. I guess you most likely have already tried different models for your individual subject analysis, and that should indicate what to put here for the correlation analysis.

Gang