(Note: I think this is pretty much in line with what R. Scott Bakker is saying at his excellent blog Three Pound Brain, not that he'd sign off on it. What I'm saying below may well be either obvious or wrong. It feels to me, though, like a pretty extreme form of speculation and possibly a decent heuristic.)
Definitions work by exclusion. You draw the line around the category with what it isn't.
Imagine a scene, any scene. There's a zoom-in mode where you look specifically at a thing and everything else gets blurry and there's a zoom-out mode where you see everything, you think. But it looks to me like the zoom-out mode came very late in the evolutionary game. The ability to have two modes certainly did. Is the zoom-out built from the zoom-in? If so, it seems unlikely that the zoom-out would escape the structure of the zoom-in.
Zoom-out mode makes it look like everything fits together seamlessly. The frame is a nothingness on the periphery you only consider for thought experiments. (The evolution of vision was not driven by awareness of a periphery.) But when you're looking, you're just looking...at a complete picture.
What if the building blocks of consciousness are the organization of world configurations by what they're not? Dog and not-dog, tree and not-tree. You need to leverage each "object" against everything else. But it's not even an object, it's a non-object through and through, built up from other non-objects. What you get is not a picture of reality; it seems a stretch to even call it an approximation. What you get is functional (you can find food) but it's also a picture that has no direct connection to what we imagine it to depict. If the foundation is layers upon layers of not-thats, it starts to make sense why the world we perceive is so dreamlike and nutty, why it's so unreal.
Could this be mainly a visual process that other senses don't use? Is consciousness as we know it the capture of most or all sensory information by vision-based processes?