If imaging is the left-to-right placement of sounds, layering is the front-to-back depth — the sense that the music is arranged in receding planes, some performers nearer and some further away, rather than pasted flat onto a single wall. It's the Z-axis of the soundstage, and it's what people are pointing at when they call a presentation three-dimensional or holographic.
It depends heavily on the depth of the stage. A headphone that renders only width and height gives you a flat panorama; one with real depth lets sounds stack believably behind one another — a lead vocal forward, the backing vocals set behind it, cymbals slightly back and above, the bass anchored low, ambience wrapping the edges. Height contributes too: tones placed a little higher and further back are easier for the brain to separate than when everything sits on one plane, so a stage with some vertical dimension tends to read as better-layered.
Layering pairs naturally with separation, but they're distinct ideas worth not collapsing. Separation is about distinctness — the gaps between elements, often a lateral and tonal thing. Layering is specifically about ordered depth — whether you can hear how the mix is stacked from front to back. A headphone can separate instruments cleanly across a flat plane (good separation, modest layering), or it can place them at convincingly different distances (good layering). The great staging headphones — the Sennheiser HD800 is the perennial reference — manage both, which is what creates that uncanny private-world effect.
It's especially hard for IEMs, which bypass much of the outer-ear and room interaction that helps speakers build spatial depth, and which feed hard left/right channel information straight to the eardrum. Newer tunings have coaxed surprising depth out of in-ears, but open over-ear designs with angled drivers still hold the advantage, and a crossfeed feature can help the illusion along.
Layering also leans on low-level detail and decay. Reverb tails, room cues, and small microdynamic shifts are what the brain uses to infer distance; if those cues are masked by bass bloom or chopped off by poor treble extension, the image flattens. So layering lives somewhere between spatial performance, resolution, tonal balance, and transient behavior — which is also a reminder that, like soundstage, it's largely psychoacoustic. A dip in the lower-to-upper-midrange contrast can read as depth, so precise claims about exactly how far back something sits deserve the same grain of salt as precise claims about width.
A common reviewer mistake is treating layering as just wide soundstage. Width helps, but a huge, vague stage can have weak layering, while a smaller, intimate one can layer beautifully if the front-to-back relationships are clear. The newcomer's test is an orchestral or live recording: ask not how wide it sounds but whether you can hear the rows — the section three feet behind the one in front — and whether the reverb has somewhere to go.