Well, because there are the problems with our perception of the world -- we perceive things (solid surfaces, e.g.) that aren't there, and fail to perceive things (atoms, radio waves, e.g.) that are there. This opens up that familiar gap between appearance and reality, and that in turn leads to the notion that reality is what's "outside", in some sense, while "appearance" is what's inside, in presumably the same sense -- in other words, it leads to just that awkward splitting of the world.
"Awkward", because once the world has been split in this way, it becomes a problem to knit it back together. Representational accounts of perception are formulated but are riven with difficulties -- are such "representations" dependent in some fundamental way on the external reality they're supposed to stand for, e.g.? Or are they simply internal mental structures that have at most a derivative, referential relationship to anything "external"? Such questions relate directly to the significant issue of how we can come to know anything of this external world.
(As an aside, I'll note that the disputes surrounding this issue seem to me to be a prime example of the kinds of confusions or at least unnecessary complications that result when the distinction between language-based consciousness and more general awareness is neglected. Thus Putnam, e.g., seems to be unnecessarily involved in some sort of metaphysics of "natural kinds" [but that Burge, with his social/cultural turn, avoids], which might lead you to think that the problems disappear if we simply reject such metaphysical entities -- but which would be wrong. In trying to understand or explain consciousness in general, the problems go beyond just such culturally-mediated structures as beliefs or concepts.)
But notice that, though we've split things in general into an internal and an external realm, consciousness itself retains its intuitive unity. What if, instead, we were to think of consciousness as a binary structure, as I've been arguing? Might this have the effect of moving the gap or split in the world inside consciousness? And would there be any benefit to so doing? Well, for one thing, the idea of so-called "natural kinds" seem immediately more plausible if we're speaking of "natural" or evolved structures of consciousness rather than of metaphysical structures. More importantly, I think, we're able not only to maintain the distinction between a mental structure and its meaning, but also to move both structure and meaning inside the general structure of consciousness -- because now we're speaking of two different mental structures. Thus, for example, when we see something red, in addition to perceiving red (the presentation of the quale "red" in the world-manifold component of consciousness), we also experience red (the "apprehension" or effect of this token by/on the actor/controller component) -- the content of the apprehension is just the quale or token itself, and the meaning of the quale or token is just the apprehension (i.e., its effect).
Such an internalizing of the split, of course, looks as though it leaves some epistemological concerns unresolved, assuming we continue to work within a representationalist framework -- multiplying internal components of consciousness doesn't appear to address the question of the relationship between either quale or token and the "external world". But this then becomes another reason to consider the epistemological inversion mentioned earlier, whereby the phenomenal world that consciousness presents is regarded as both the foundation and building material of all knowledge, and knowledge is seen as layered structures, of greater or lesser extent, built on top of, and out of, experience (and the environment is not considered as an object of knowledge in any sense). Wouldn't that make "truth" inherently, and disconcertingly, subjective? No, because, in speaking of "truth" or "knowledge" at all, we're speaking of consciousness in its special or language-based sense -- in this sense, truth admits of degrees, and both truth and knowledge become inherently social and practical (the latter being the feature that recovers a notion of objectivity). Here's a passage from Quine that makes a similar point -- it's toward the end of his "Two Dogmas of Empiricsm", in which he compares the objects of physical science with the gods of Homer, stating that the "epistemological footing of the two "differ only in degree and not in kind":
Both sorts of entities enter our conception only as cultural posits. The myth of physical objects is epistemologically superior to most in that it has proved more efficacious than other myths as a device for working a manageable structure into the flux of experience.