Rickert on embedded computation (was re: science of consciousness.)

Oliver Sparrow ohgs at chatham.demon.co.uk
Wed Apr 29 03:27:02 EST 1998


 andersw+ at pitt.edu (Anders N Weinstein) wrote:

"" The main problem I find with this is: if your theory has it that what
"" goes on in Niagara Falls or the core of the sun or in a hurricane or in
"" alphabet soup when you boil it might be "computation", haven't you
"" robbed the idea of computation of any significant meaning?

Dogma: everything that is can be reduced to simulation; and all simulations
can be reduced to logical operations on an idealised computation widget.

Stigma to beat a dogma: the simulated Niagara may or may not be open to
instantiation on conceivable widgets, but given that it is, all that it can
offer is a Turing test on an interface. That is, an observer can be
incapable of telling a representation of a simulated Niagara from a similar
representation of the real one. At its most extreme, suppose that the
observer could be hooked up to nanomachines and whatnot that short
circuited the perceptual systems and fed in the corresponding data from the
simulated Niagara. That is, the observer would be embedded in one kind of
perceptual interface rather than another. Would the real Niagara be
'replaced' by its simulation? If the test for "replace" means, would an
individual confined to a chair be unable to tell the difference, then
perhaps so; but if the test is about generating electricity or serving as a
focus for tacky hotels, then no. 

All this begs the question of what 'computable' means. Is a system open to
exact analytical understanding? Is the understanding capable of
instantiation in a system which exactly emulates the system in question?
Are both of these systems using techniques which are connate with what we
mean by "compute"? Plants model the precession of the seasons with some
accuracy, but they do not follow operations that can be described as
computing; but are understandable; and are open both to simulation and
manipulation. 

Consider awareness in this context. Suppose I replace awareness with a
simulation of awareness. Er... but awareness is the interface and the thing
interfaced with: it is both the interpretation of views of Niagara and that
which notes these interpretations. So I substitute the same thing for
itself, and ask it if it can tell the difference? No, no, no: you change
the *plumbing*: John was running on a 80X86 and is now on DEC zeta, lucky
John. John has been recompiled: see John run! Can you feel the difference,
Sir? Of course Sir cannot, for Sir is the sum and fusion of what makes Sir
run, and how that feels is the consequence of what it is at the time.
Memory of what was is grounded in feeds which are created continually.

______________________________________

Oliver Sparrow



More information about the Neur-sci mailing list