We can also eliminate them, with a little more difficulty, from our description of the computer. But the cognitive-science descriptions of the digital brain only seem to help us explain consciousness if we retain the intentional idioms and refuse to replace them.
Yet the real problem for cognitive science is not the problem of consciousness. Indeed, I am not sure that it is even a problem. Consciousness is a feature that we share with the higher animals, and I see it as an “emergent” feature, which is in place just as soon as behavior and the functional relations that govern it reach a certain level of complexity. The real problem, as I see it, is self-consciousness — the first-person awareness that distinguishes us from the other animals, and which enables us to identify ourselves, and to attribute mental predicates to ourselves, in the first-person case — the very same predicates that others attribute to us in the second- and third-person case.
Words like “I,” “choose,” “responsible,” and so on have no part in neuroscience, which can explain why an organism utters those words, but can give no material content to them. Indeed, one of the recurrent mistakes in neuroscience is the mistake of looking for the referents of such words — seeking the place in the brain where the “self” resides, or the material correlate of human freedom. The recent excitement over mirror neurons has its origins here — in the belief that these brain cells associated with imitative behavior might be the neural basis of our concept of self and of our ability to see others too as selves. But all such ideas disappear from the science of human behavior once we see human behavior as the product of a digitally organized nervous system.