Other's have commented (and my tests have confirmed) that my 1D4 and 5D2 cannot actually see light from a cone that's faster than f/1.6.
This is incorrect. There is a loss of light due to the sensor but there is a definitively different blur with f/1.2 vs f/1.4 and f/1.6.
Pi - Ok I'll bite.
We seem to agree that many digital cameras are insensitive to light arriving from large apertures. I know this to be a fact with my 5D2 and 1D4 and all the f/1.2 and f/1.4 lenses I've tested. I invite you to test this if you disagree.
Nevertheless, you have a persistent belief that apertures wider than about f/1.4 decrease the depth of field even though very little additional light reaches the photodiode.
I would like to understand how a camera's photodiode can simultaneously detect and not detect light. It is a very clever trick.