The resolving ability of the human ear is largely dependent on the inertia of its component parts. Do any of these component parts have low enough mass to be accelerated at rates high enough resolve the difference between the 44.1 kHz and 192 kHz sampling rates?
How about tinnitus -- the "noise floor" of the central nervous system, so to speak, and which is present in every human -- does that by any chance mask the minute difference between sampling rates?
It seems to me that high sampling rates are useful not for your listening pleasure, but for digital signal processing. The more data that is available to the signal processor, the better that signal processing will be able to work. Also for post-production work: if the audio is to be slowed or sped in time, greater resolution will result in lower "pixelization" of the end product. When combined with other signals (mixing), greater resolution will result in a more accurate final product. But once the product if finalized, it can likely be downsampled with no discernible loss of quality. In other words, high sampling rates are useful on the recording end, not so much on the listening end.