I tried an experiment. I used the grouping tool to make a cornea prop from V4's eye, made the original cornea invisible and slapped a reflection/refraction setup on the cornea prop and on the sclera.  After a few test renders and fixing some little annoyances like whatever issue it is that causes AO on reflective/refractive surfaces in my setup to make a cloud of tiny black gnats all over the middle of things, I got a very weird result--my entire cornea was some kind of skin color.  I had skin-colored all other zones before I texture-mapped the face, lips, nostrils, sclera and iris...there are no other maps involved aside from the lash transparency and bumps for what's texture-mapped.

I couldn't tell what that was from, so I assigned different colors to the head (red), neck (orange), body (yellow), and eyesocket (blue and red tiles, because I was so sure that was what it was!).  The new render shows that the mystery image is the Head zone.  Even though I double-checked to make sure that my cornea prop, which is parented to the right eye, is in front of the eye itself and does not intersect with the iris or sclera, it is reaching right through them to refract that Head materal. 

Now why in the world is it doing that?  If I put a refractive glass on a table top, even if they are touching or even intersecting a little, the glass does not ignore the table to show the floor beneath it.  I'm missing something!

On an unrelated note--see how grainy my shadows are? I have my min bias down at .2 for those raytraced shadows, with a blur radius of .25.  What am I doing wrong?  If I lower the bias any more I end up with mesh artifacts. This has been driving me crazy...it doesn't show on a render farther away from the subject, but do I really need to adjust the bias that much just to do closeups?