The entire argument is based on the premise that the ear can't detect content above the limit of 20khz. While it is no doubt that few if any ears can hear indep

The entire argument is based on the premise that the ear can't detect content above the limit of 20khz. While it is no doubt that few if any ears can hear independent tones above 20khz, what about detecting the shape of tones below that point? I don't think anybody's proven that the overtones of notes below 20khz are unimportant, or that the ear doesn't use edge detection for example to determine phase differences for location determination. These would require faithful reproduction beyond 20khz. Also to clarify a technical point, sampling at 192khz doesn't extend the frequency response to 192khz, it only extends it to 96khz (Nyquist in action).