while technology is very advanced it still has a lot of limits. Watching what is happening in someone's brain without damaging them is quite difficult. From the article "Currently, people who volunteer for such experiments have to be able to fall asleep while wearing an electroencephalogram (EEG) cap inside of a noisy MRI machine—no easy feat." So it would seem that studying internal brain activity during healthy sleep is indeed a challenge.
Imagine some years into future we will find that sleep does not work the way we thought because all data was from selected few unnormal enough to be able to sleep inside mri machine.
and EEG is noisy as shit (SNR wise) and fMRI has a 2-second lag and MEG's ill-posed spatially and repeatability is hard and imaging only gets you a few hundred microns into the grey matter and you can't do it in humans and "single units" usually aren't and you can't do that in humans either anyway and and and ...
