Just a hobbyist here, please correct me if I'm wrong, but I think I have a decent grasp of normal maps.
Basically, the baking process is just the hi-res model "firing" rays onto the low-res from it's geometry. Then when the ray hits the low-res it alters the normal on that part of the object. That alteration is reflected on the Normal Map that you unwrapped for. Normal map data is just telling the renderer how to work with light rays when the light rays hit each part of the low-res.
I find these images very helpful :
There are many many articles about this and I suggest reading everything you can find.