Life Above 20 Kilohertz!"
If you think sound ends at 20 kHz,
then please see James Boyk's paper
Life Above 20 Kilohertz!"
Boyk measures real world musical instrument spectras out to 102 kHz
and very clearly shows harmonics of real sounds way above 20 kHz.
Whether we perceive this information is an open question,
but the sounds themselves probably need 192 kHz sampling for accurate recording.
Within this atmosphere of novel experimentation there can be controversy, since the question often comes down to what a listener or listening panel judges to "sound good". This is especially the case with new developments such as high resolution digital audio, upsampling, new digital filter algorithms, etc. With these truly new areas, the old references of vinyl sound, analog tape sound, and even conventional Compact Disc sound can be less than useful or even potentially misleading to an open-minded and keen-eared listener.
That's not because science can't measure everthing we can hear, but because science hasn't yet come up with workable theories for all that we can hear. Science will never be able to explain everything, but it's the only reliable tool we have available to attempt to find those answers. Just because science can't yet explain something doesn't necessarily mean that something doesn't exist or that a scientific explanation of it is impossible. It may simply mean that there is no scientific explanation yet.
For example, I find that upsampling seems to let the music on CDs through better. I definitely hear more detail, spaces and general clarity, while at the same time it's smoother and more liquid. The general consensus seems to be that raising the sampling frequency moves filter artifacts further away from the desired signal and this may be some of what we're hearing. In other words upsampling lets the analog filters do less damage to the music.
At first I was uncertain about the sound of the upsampled music. But a few things changed that for me. First, native 96 x 24 DVDs like the Cheskys or Classic Records ones have a smoothness to the sound that is unlike anything else I have heard, whether it's vinyl, CD, or analog tape. Upsampling regular CDs seems to get closer to that sound so the filter argument above may have some merit. Second, downsampling 96 x 24s to 48 x 24 moves the sound in the direction of non-upsampled 44.1 CD: less depth, harder, grainier, etc. Third, I came to realize that I had associated the harder CD sound with "quality" because I had been listening to CDs that way for so long and had assumed it was "right". Well I'm pretty positive it's not "right" now and that higher sampling rates and to a lesser extent upsampling bring us closer to the original event and are therefore higher fidelity.
I think the final point is very important. It's very easy for perceptions to be influenced by prior experience and present expectations. It's important to have an open-minded approach and to give new things some time for the brain to "relearn" and adjust. Hearing live music often also helps. How do we know when we're hearing truly better sound? What's our reference? Live music probably makes a much better and safer reference than CD, LP, or even 15 IPS analog master tape....