Science follows, not leads, human perception. We invent the measurements to understand perception, and then fit theories to the measuremens that are not exact. Finally we mathematically manipulate the formulas that follow from the theory to derive/predict new conclusions, which we then attempt to further verify with measurements.
Nothing prevents there from being a real-world condition that we cannot (yet) explain, as others here have noted. It is simply that our models are not good enough yet. You can predict that something is inaudible, but it is inaudible only if you can't hear it.
Posters mistake this all the time. They say "this is too low to be heard". That is not correct -- they mean "our current scientific models, proven to be accurate over years of testing, predict that this can't be heard".
That's why I like listening tests. But because of the placebo effect, and bias, and differences in hearing person-to-person, tests are not easy. And many believe that blind tests introduce bias of their own. Save this fight for another post.
Some examples of theory vs real-world:
It is a correct conclusion of standard audio waveform analysis that all songs never end -- they play for all time, and have been playing since the beginning of time (because the frequency response of your headphones is limited, therefore the Fourier transform of the frequency response curve-- which is amplitude vs time -- must be non-zero from minus to plus infinity ... or else if the music stops then your headphones go to 100KHz and beyond, to infinity).
This is obviously nuts, and many here know the answer -- the amplitude/time function and the freq response function are actually related by much more complicated transform theory -- but standard Fourier analysis is an extremely good approximation to reality, and is very useful.
Our models define attributes of an idealized circuit or sound transducer system that is a close, but not exact, description of reality. And our measurements measure something physical of course, again a close approximation to the attributes we just defined.
Neither the measurement nor the attribute are 100% correct, they are only models.
We are sure to discover more about human hearing and sound perception over the next n years. And then things will be clearer (pun).
For decades -- centuries actually -- we had no clue how homing pigeons found their way back. We worked on this at Bell Labs in the 70's. I mean really no clue. Of course the pigeons didn't care, they flew home. Scientists are now starting to understand the complex processes involved -- the pigeons are amazing.
High end gear makers think we are pigeons and will buy anything shiny that gets good reviews from reviewers with an economic interest in the industry. Bird poop on that -- let's test and test and test -- by listening, not measuring -- until the cows, and the pigeons, come home.