But we're talking about CD sampling/recording rates, not turntables here.
The original frequencies aren't necessarily below the noise floor of the recording process, and can introduce harmonics during the playback process. Many do have enough "energy"; there's no magical cutoff at 20kHz just because that's pretty much the upper range of human hearing.
You can argue - somewhat correctly - that those harmonics within audio range would be captured by the lower (CD-quality) sampling rate.. but herein lies one of the things that, depending upon conditions, can be heard. It's not the same sound profile as playing back the actual higher frequencies because (shockingly enough) playback conditions impact how it sounds. Note I'm not saying one would sound better than the other, just noticeably different.
Also, if you have inferior AD conversion and low-pass filters when recording, anti-aliasing can be a problem caused by higher frequencies. Again, I'm talking strictly about higher sampler rates and storage at "better than we can hear" quality. AC (not sure if it was you) argued it was completely identical under all cases unless you were some kind of mutant. That just isn't true. It's mostly true, but not completely.
Finally, if we were to do what you suggest with a vinyl record, it... doesn't prove much of anything. Why? You're only passing the higher frequencies. The harmonics occur at a lower frequency in terms of the music. Ostensibly, you're playing inaudible frequencies against line noise, which will of course sound like line noise. This, of course, assumes that the recording wasn't at some point or another passed through some kind of EQ or digital recording (same sample rate discussion) that didn't remove them. Digital recording goes back much further than most people think..
Emphasis here on something that is mostly irrelevant: the difference between OP saying it's impossible for there to be any distinguishing characteristic for humans, and then talking about what the practical implications are. You really should read the short article I linked, it gives a fair, if high-level, explanation.