Abstract
I want to try to do something rather more speculative than the rest of you have done. I have been thinking recently about how one might explain some features of human reflective consciousness that seem to me to be very much in need of an explanation. I'm trying to see if these features could be understood as solutions to design problems, solutions arrived at by evolution, but also, in the individual, as a result of a process of unconscious self-design. I've been trying to think of this in the context of work in AI on the attempt to design intelligent robots – not "bed-ridden" expert systems, but systems that have to act in real time in the real world. If you want to think about something like this, you have to stray fairly far from experiments and hard empirical data; you have to get fairly speculative. Nevertheless the design efforts of people in AI do seem to bring home to conviction – if not to prove – various design constraints looming large and inescapable. If we can come to see why a system – or an organ or a behavior-pattern – must have certain features or a certain structure in order to do its task, this may help us ask the right questions, or at least keep us from dwelling on some of the wrong questions when we try to explain the machinery in the brain that is responsible for intelligent action.