[JFreak]

IN THEORY, digital domain and higher sampling rates ONLY mean processing of higher frequencies are possible. (EDIT: just the same, digital domain and higher bit rates ONLY mean higher dynamic range, but that is not necessary as 24bit can handle 144B and best converters can only do 124dB which still gives us a nice 20dB headroom to record below digital full scale)

IN REALITY, if you ADDA caps at 20k that is taken away from you. The only advantage of higher sample rate session lies in whether the plugin you use works better in 96k compared to 48k. If you only store/process frequencies below 20kHz, there is ZERO advantage in recording at zillion megahertz sampling rate

EDIT:
assuming you have perfect converters. the analog filters that cap to 20k make the more imperfections the less perfect converters you use. this is why people can hear a difference between 48k and 96k in blind tests. if we had 64k sampling rate available, nobody with human hearing could make a difference between 64k and 96k