It all depends on the implementation at each sampling rate. The better the implementation, the less it will "sound like" anything at all. Of course any real-life comparison depends on many things that aren't strictly a matter of the sampling rate.
For a clear, well-informed discussion of the possible advantages of sampling at rates higher than 44.1 kHz, see Julian Dunn's AES paper "The benefits of 96 kHz sampling rate formats for those who cannot hear above 20 kHz," which is available for download on
http://www.nanophon.com/audio/antialia.pdf.
Please consider this, though: Better recording technology doesn't always make recordings sound "better." Sometimes it just makes them sound
more like the signals coming from the microphones, which sometimes it
not what people would rather hear. Plenty of people have fallen head over heels for various kinds of distortion. I can sympathize, but I prefer as an engineer to apply such distortion consciously--not to have it built into the equipment I'm using, where I can't raise or lower the effect when I want to.
I'm wary of recording devices that are said to make things sound "better," since if they have any effect on the sound quality at all, logically there will also have to be times when they make things sound different in a way that I would
not call better. There's no device or process that
always makes things sound better, and
never makes them sound worse!
--best regards