When I talk about depth I mean front to back distance perception of the entire band within the space (or room) of original recording. Level changes will artificially change positioning within the depth of the soundstage, but not the overall depth of the original recording.
For example, if we accentuate say the vocals range, they would be more forward within the soundstage. While this could be construed as added depth, actually everything in that range is artificially pushed forward, therefore all that was accomplished was an offset on the position of that frequency range of information relative to the rest of the image. Having vocals 10 feet out from the drummer does not necessarily mean proper depth from a 2 channel playback system (although it could be possible given enough space and maybe a set of planer loudspeakers
Increased spacial information, as if you were there, allows the brain to better localize the space between and behind instruments and singers. If it exists in playback, the depth can be reconstructed in the brain, if not, then playback is more 2D. This tends to be the case in many 2 channel systems I've heard.
This would be far easier to explain in person in front of a system while making these changes. Trying to describe it in a few paragraphs is proving to be difficult.