There is enough scientific evidence that it is bullsh*t
Up until now there is no evidence of a human being able to hear frequencies over 20kHz, and the Nyquist/Shannon theorem still hasn’t been proven wrong. For recording/production environments it makes perfect sense to use higher bitrates and resolutions to have some margin for processing, but after the final mastering there is no need to use anything higher than 16bit/44.1kHz.
Some more reading material (with scientific backup!):
numeralnine.wordpress.com/2013/ … l-consumer