Basic Digital Sampling:
The Nyquist/Shannon Sampling Theorem developed by mathemetician Harry Nyquist and later proven and implemented by Claude Shannon at Bell Labs circa the 1940's is a very complicated piece of mathematics. It's not "simple math", but the basics of it aren't too complicated in laymans terms.
Basically in order to sample a frequency and reproduce it accurately in all its detail, you need to sample at twice the frequency of interest. In other words, any frequency that you want to accurately reproduce needs at least 2 samples per cycle. At what is normally considered the upper limit of 20 kHz for human hearing, that means 20,000 Hz or 20,000 cycles per second. So to capture 20k, we need to sample at 40k, minimum to get those 2 samples per cycle. The "Nyquist Frequency" is the half way point, so again if we're doing 40k sampling, the Nyquist limit is 20k.
A reconstruction filter is needed upon playback through the D/A converter to get rid of any frequencies above Nyquist. If there are less than 2 samples per cycle you get aliasing, which is a form of digital garbled nonsense that will ruin everything. We need the samples, but we need to filter out any frequencies above Nyquist to prevent aliasing. This is done with a very steep low pass filter, or reconstruction filter. When Sony and Philips developed the standard for (absolute bare minimum) full range digital audio CD's, they specified 16 bit, 44.1 kHz audio. So at 44.1 kHz, the Nyquist limit becomes 22.05 kHz. The space between 20k (end of the road for human ears) and 22.05 (Nyquist) is used to build the filter.
Since we can capture 20,000 cycles per second with 2 samples (very slightly more with the filter), lower frequencies are no problem because they end up having more than the required 2 samples per cycle. It has no effect on the accuracy of the reproduction. So at 10k we have 4 samples per cycle. 5k, 8 samples. By the time you get to bass frequencies, there's hundreds of samples per cycle if not more. It doesn't mean those frequencies are "better" because there's "more samples". Increasing the sample rate will increase the theoretical capture range of frequencies above human hearing, but we can't hear it. In some cases (certainly at 192 kHz) it makes for huge file sizes and a lot of data to process which normally has negative effects on system performance.