🚀 go-pugleaf

In article <1992Sep12.034215.16017@mp.cs.niu.edu> rickert@mp.cs.niu.edu (Neil Rickert) writes:

>However it is quite possible that you have it backwards.  Experience
>has shown true that the problem of  producing mechanized learning is
>quite hard.  Once we have the right kind of learning, consciousness may
>actually turn out to be much simpler than we think.

Well, learning is a significant technical problem for AI and psychology,
but it's not so much of a conceptual problem.  Learning is just a matter
of having the right sort of causal organization in a system.  Sure, due
to our empirical ignorance about brain function and the like, we don't
know just what causal chains it is that do the job, but that just
means we have to look harder.  Once we've uncovered those causal chains,
there won't be any further question "but why is this *learning*?".  The
answer to that question will be purely semantic: i.e., because those
causal chains lead to the successful adaption of behaviour in repose
to environmental stimulation, and that's all that it *means* for
something to be "learning".  Whereas with consciousness, even once we've
unconvered all the causal chains in question -- once AI and psychology
have done their work -- there will remain the further question of
"why do those causal chains give rise to consciousness"?.  So, learning
is a hard empirical problem, but consciousness is a deeper philosophical
problem.

-- 
Dave Chalmers                            (dave@cogsci.indiana.edu)      
Center for Research on Concepts and Cognition, Indiana University.
"It is not the least charm of a theory that it is refutable."