It's spacially based. It makes more sense in 3D. It's just as compatible with echolocation as visual data. (The soundscape of a room tells you a LOT about your surroundings). I believe it's based within my visual system, just stripped of the superfluous visualisations. Interestingly, I can actually map mathematics into the same structures.
I'm doing a piss poor job of explaining it. Language lacks the nuance to describe it well, and I lack the skill to bend it into shape.