It's because the highest frequency a 8khz can capture is 4 khz, so it doesn't record anything above that, I don't know the science and theory behind that, but the sampling rate must be the highest frequency you expect to record multiplied by two, and since the highest frequency most humans can hear is around 22 khz, 44 khz is usually enough.
With 96 khz the highest frequency you can capture is 48 khz but thats way beyond what someone can hear, so maybe a cat or a mutant could tell a difference between 44 and 96 khz but most people can't
I think you're getting mixed up. Sample rate is the number of samples in a second. It's measured in frequency. The higher the frequency, the more samples are played a second.
What you say about humans not hearing over 22 khz is correct. But this is only the case when we are talking about pitch.
Frequency is a measure of the number of occurrences of a repeating event per unit time
the repeating event you are refering to is the sine wave oscillating up and down to quickly for us to hear it, which is known as pitch.
When measure sample rates, it's not about the sine wave getting closer and closer together to generate a high pitch, but the amount of digital samples per second, that give the representation of sound...
I'm going back to the painting....
A higher sample rate would mean that the dots of colour from a printer are closer together forming a more solid colour on the page. If the lower the sample rate, the more infrequent the dots would print out, giving a poorer resolution.
Kind of like the difference between a monitor with a high resolution (compared to a DVD or something with a high number of samples per second) to a monitor with a low resolution (compared to something with a low sample rate like a telephone or something)