This idea of 2D vs. 3D two-channel listening has baffled me through years of system improvements and visits to shows. 3D is easy to rationalize for an omnidirectional speaker. But with a forward-firing speaker, the 3D effect seems to come from reproducing fine shades of volume through which we perceive precise distance between the microphone and the instrument, or subtle ambient cues at the far walls of the recording space through which we perceive an overall stage.
A system that is unresolving or lacking in dynamics can imply a deep stage by veiling and deadening. Everything sounds more or less recessed in depth. This may endear some listeners who prefer a relaxed system or who reliably take the same seats at concerts and like to feel the 20th row replicated in the living room.
A great system jumps forward by virtue of superior dynamics, while communicating depth cues photographically with the optics of a superior lens. It’s a 2D window to 3D. When an improvement in the system reveals subtle details at the far corners of the recording space, this is perceived as more and better 3D.
The effect is not determined simply by speakers, but by every link in the chain.