In the Cartesian tradition, the mind’s autonomy was of paramount importance. All we need to do, essentially, is get our internal workings functioning and in good order. It doesn’t really matter about the external world. Or, more correctly, it doesn’t matter until we had got our mental ship in order.
In epistemological externalism (as well as in reliabilism), on the contrary, what happens before the formation of a belief (or before cognitive operations as a whole) is what matters. The question is simple:
How did I acquire this particular belief?
So it's the required causal relation with the external world that determines whether or not a belief (or cognitive process) is justified. If it's acceptably justified, then we may have knowledge. We may not even be able to elaborate on the external causal processes that led to the formation of a belief. Though if we're in the right situation (vis-à-vis the world), then such processes which we may not be aware of will themselves somehow justify our beliefs.
Of course it's the case that we require causal relations with the external world when it comes to our perceptual knowledge. That is, in a purely empiricist manner, all we really need are reliable causal contacts with the world. Though, as Kant might have said:
There's much more to knowledge than mere causal interaction with the world.
Our minds and brains need to do something with all the incoming data – they need to synthesise it. That is the case even with basic perceptual beliefs.
Can we rely on perceptual information alone to make general statements about the world? That is, no single experience (or even a large group of experiences) alone will tell me that, say, that all swans are white.
And do we ever have a basic perceptual experience of someone not murdering someone else?
What about microscopic properties, objects or processes which are always beyond perceptual experience? What does pure experience tell us about these things?
What about numbers? They're supposed to be non-spatiotemporal abstract entities. How could I ever have perceptual experience of - or causal interaction with - numbers?
So not all beliefs and bits of knowledge are simply a question of reliable causal contacts with the concrete things that the beliefs and bits of knowledge are about. The traditional empiricist had no convincing answers to these epistemic problems. There are things about the mind-brain that work free of causal interactions. Or if there are causal processes in the mind-brain, then they're of such a kind that we don’t even know that they're happening. In that case, they have no epistemic relevance.
The internal world must have at least a little autonomy from the external world. Though perhaps we don’t necessarily need to think of it at all as the internal world. The mind-brain is indeed a part of the world. It is naturalistically acceptable. However, the mind-brain does things that other parts of the world don't do. It thinks about the world. It has representations and images of that world. It has intentionality – that is, directedness. It deals with meanings, concepts, truth, falsity and the rest.
Although the mind is part of the world, it's a special and unique part of it.
We causally interact with the world, and that in turn sets of causal processes in the mind-brain: it doesn’t follow that there'll be some kind of isomorphic and infinitely repeatable set of mind-brain relations to that world. The same causal processes bring about different beliefs in different people - or even the same person at different times. Something goes on in mind-brains that can't be accounted for in the same way that things that go in the world can be accounted for. This is not a mystical conclusion. It may simply be a fact about the mind-brain’s astonishing complexity and subtlety. Indeed it has been said that the mind-brain is the most complex thing in the entire universe.