^I was asking irt design differences as I've been so busy I haven't finished reading both 8801 and 4520 threads but I am under the impression that the 8801 is not just a 4520 minus amps. As to the real-world import of the differences irt SQ, that remains a difficult question in my mind
I agree with you that with so many similarities in design and components of these two similar high-end models (8801 and 4520), one could expect the SQ to be quite similar.
In fact, it seems to me that at this level of XT32-equipped processors the likes of Denon 4311, 4520, AVPA1, Marantz8801, one could reasonably expect overall SQ to be quite exc across the board from all of them, and forSQ differences to be fairly small. I'd expect pretty much the same of the XT32-equipped Onk/Integras.
Comparing such models, if one hears "night and day" differences, the first assumption should be that something went wrong procedurally in the A/B. After all, these are complex machines with many settings. And Autosetup mic placements can produce slightly different-sounding calibrations in the same unit and are hard to standardize.
IME reliable evaluation of subtle SQ differences requires level-matched repeated critical listening of very well-recorded short passages with little delay between A and B. Of course the rest of the system has to be revealing as well. Such subtle SQ differences are more subject to expectation bias, both conscious and unconscious. Even if you do level-matched BAB tests, processors are particularly challenging. As Kal pointed out, you really need a switcher.
So I agree with you that psychoacoustics come into play. However I don't assume that any reported SQ differences between models are due to placebo. I just take most of them with a grain of salt.