This discussion has largely been focused on the electrical / physical aspects of sound, but may I diverge from that route for a moment?
While I largely agree that soundstage is dependent on the recording, when speaking specifically about headphones, it would be hard to imagine that the headphone type and distance and/or angle from the transducer to the ear wouldn't matter.
While I'm sure a number of acoustic variables may need consideration, if we look simply at 3 different headphone types, each with transducers that consistently move further from the ear (and ear drum), the perceived headstage significantly increases.
IEM -> Supra-aural -> Circum-aural.
At its most basic level, as the headphone transducers move further from the ear drum, headstage increases. Comparing IEMs (which sound like the music is playing in your head with very little L/R distance), to supra-aural (which sound like the music is at least outside your head) is IMO an obvious reflection of soundstage with these 2 headphone types.
Using a circumaural headphone moves the transducers even further from the ear, and also engage the pinna which have actually evolved to locate sound in 3 dimensions which may further create a more elaborate soundstage.
The famous K1000 are essentially speakers attached to the head and due to the great distance from the transducer to the ear, a large 3D soundstage is often reported. Other headphones like the AD700 have angled drivers set at a great distance from the ear and thus are often described as having a large soundstage. While it is difficult to determine exactly how far back the drivers are physically within the headphone enclosure, there seems to be a positive correlation between driver distance and the maximum perception of soundstage. It stands to reason that a hard panned sound would sound even further, if it were emenating from a source (i.e. driver) that was distally situated from the ear.
And lastly, it would seem overtly obvious, that psychacoustics and expectation bias also play a role in the perception of soundstage. It would seem rational to expect that if the K701 headphone begins to develop a reputation for having a large soundstage, those listening to the K701 will be expecting to hear one, whether real or imagined. And finally, the observation that pro-cablers have made about the resulting increase in soundstage when using "x" cable is further proof in my book that soundstage can be largely imagined.
