Hi Michael,
I don't think that scaling the timing for regression analysis would generate the correlationship between what subjects perceived as pain and their real response during the experiment.
Instead I suggest that you go ahead and run 3dDeconvole with the original stimulus timing files (without scaling), and then at group level run regression analysis with the pain perceptual ratings, or run ANCOVA with the pain perceptual ratings as a covariate if you have other factor(s) to test. Let me know if I can be of any help in this regard.
Gang