The illusion of soundstage.

Pan pots used to any degree by any recording engineer are the secret…pan left, pan right, or no pan at all.

Head Related Transfer Function. :) The secret is how your head, ears, and hair affect incoming sounds from different angles. It’s more than just amplitude and simple phase. Your body adds comb filtering your brain picks up on to determine vertical and horizontal angles. Very cool science.

Having said that, to get good sound staging in any dimension you must have good room treatment in that direction. So, if you want good left to right soundstage, you should have good room treatment to the sides and behind the speakers. If you want height, good treatment above the speakers.

There’s a known speaker gimmick. Dipping the frequency response around 2.4 kHz will exaggerate the sense of imaging. You’ll see this in a number of "high end" speakers.

Most of the times I've heard this effect by the way it seemed like an accident, as opposed to a real attempt by the recording engineer to do something. Intimate jazz quartets for instance. 

The ambient information of the recording venue including reverberant decay, echo, first reflection and other subtle queues are captured in the recording thus two speakers can reproduce this information. Thus, depth, width and height dimensions, the 3D sphere or cube of the location as it were, should theoretically be reproduced (on a capable system). One might have to struggle a little bit to extract all the ambient information from the recording that’s there to extract but theoretically anyway it’s there.

So what i really was looking for was how, physically, sound waves can appear to be originating from outside the width of the transducers. i can imagine with an electrostat... but in phase or not, how does the wave(?) appear to come from  a different place. i am a visual thinker and want some sort of representation in that sense. thanks for all of the replies btw

Look up head related transfer functions.  Short explanation is you never hear a flat response.  Sounds that reach your ears from different directions have comb filtering effects, which your brain interprets as direction.  If you can fake those effects through sigmal processing you can fake the direction.