I’m not sure if you’re asking what’s going on technically but as I understand it, instead of delivering you two glorious channels of PCM sound Atmos instead sends you every channel that would have fed into the master mix along with some information about gain and, crucially, 3D position. This means that the final mix happens in the room you’re listening in, with the information about where the speakers you’re listening with are. Headphones are a special, easy case.

Also! They can do the up/down with something called the Head-related Transfer Function. This is a bit of numerical magic: turns out we figure out where things are in 3D space using subtle phasing and tone differences. I’m not sure how individual they are to each person, but I bet they can run a quick training phase. I think some of the fancier apple headphones use accelerometers to keep the sounds in the same position relative to the space you’re in rather than you’re head, even.