I guess that at high frequencies there is a sort of dithering (Wikipedia). How I perceive this "dithering" is that at high frequencies, the sample rate isn't enough to "drow" the waveform with accuracy. This leads to zones where the amplitude is as the original and zones where the amplitude is reduced. If the waveform is a pure sinus, these zones are periodically. Well, music is not a sinus waveform but it consists of such. Some sounds that have more clear waveforms are more prone to such behavior. In addition, as going higher in frequency, this is getting more obvious.
Jim, I understand what you mean. The fact is that I can hear that at quite high volume, well above the noise floor of each medium. For example, at "Grease" soundtrack, the song "Born to hand jive" has a solo of percussions. On the LP the cymbals sound smooth. On the CD don't.