RE: UDA revisited

From: Stathis Papaioannou <>
Date: Sun, 26 Nov 2006 21:16:51 +1100

Colin Hales writes:
> <snip>
> >> Also...paramecium is not noted for its
> >> scientific behaviour!
> >
> > The computer driving the paramecium shell might be difficult
> > to build, but in principle it would be the same sort of task as, say, a
> computer running an analogue clock or projecting a film
> > (i.e., originally filmed on a celluloid strip)
> > onto a screen. With sufficient attention to detail, it should
> > be impossible to distinguish the digital replica
> > from the original. If you don't believe the paramecium
> > replica can be made
> > indistinguishable from the original, which part of the paramecium is it
> that would be so hard to simulate? If you do manage to
> > simulate it, down to the quantum level if
> > necessary, then how could it possibly not behave like a real paramecium?
> >
> > Stathis Papaioannou
> With a Paramecium you might be able to get pretty close. Let's assume you
> have the nan-tech to build a replica paramecium coat for a small computer
> to inhabit that is running the model.
> The boundaries of the model would be where the distinction between the two
> might be able to be scientifically ascertained. Expose the paramecium to
> something that wasn't accounted for in the model. Hit 'em all with boiling
> water? Take them out of water? Shine bright lights? Put them in the dark?
> Starve them? Does it eat like a parmecium? Does it evade obstacles like a
> parmecium? Does it sleep like a paramecium? Does it die like a paramecium?
> Can it make a baby paramecium?
> Also: Consider my original definition:
> > A) that included a model of those aspects of the physics
> > participating in what the paramecium could have as experiences.
> > B) That included all the molecular pathways (cilia molecules, the lot)
> > C) that included a model of the response to the perceptual physics
> > D) That included a model of the environment of the paramecium
> A) is the killer. Modelling the physics that does the experience is not
> 'having the experience'.
You seem to be implying that there is some special physics involved in living
processes: isn't that skimming a little close to vitalism?. All I see is the chemistry
of large organic molecules, the fundamentals of which are well understood, even
if the level of complexity is beyond what modern chemists' computer models can
cope with. Classical chaos may make it impossible to perfectly model a living system
in that the behaviour of the model will deviate from the behaviour of the original
after a period of time, but the same is true if you try to model a game of pool.
As for "modelling the physics that does the experience" not being the same as having
the experience, I think your own argument showing that a colleague cannot behave
as if he is conscious by doing science without actually being conscious refutes this. If
you could model all the responses of a scientist to his environment on a computer in real
time and clothe this computer in the skin of a scientist, then this artificial scientist should
behave just like the real thing and therefore should have the same phenomenal
consciousness as the real thing.
> You tell it to respond 'as-if' it had them. What you do not do is model
> all possible paramecium experiences, only the ones you used to create the
> model. The experience and the behaviour in response are 2 different
> things. All you can observe is behaviour.
Of course you model all possible paramecium experiences: you wouldn't be doing your job
if you didn't. But that doesn't mean you have to program in each possible experience one
by one, any more than a word processing program needs to explicitly contain every possible
combination of characters a user might possibly input.
> You have to be really careful here. "Indistinguishable" by what criteria?
> If you have done tests and recorded X behaviours and there are actually
> X+Y behaviours, but you failed to observe them all, then you can be 100%
> sure you have made an indistinguishable paramecium, but you've really only
> made a 100% tested paramecium.
> However, if you made the paramecium with a chip that manufactured whatever
> mini-experiences constitute a paramecium's consciousness... you'll get
> closer..but even then I suspect they'll be different in some ways.
> Rodney Brooks (MIT AI Lab head) said "the world is its own best model".
> The only thing that is a paramecium is a paramecium. For our own purposes
> with experience chips embedded in a good computational model we could get
> functionally as close as suited our purposes. But that's as far as it
> goes, I think.
You might not make a perfect replica, but that isn't limited to living systems. It's
perhaps a practical impossibility to make a *perfect* replica of my car, but it may be
possible to make a replica to within a certain tolerance, even if that tolerance is the
degree to which the car changes from day to day just sitting in my driveway.
Stathis Papaioannou
Be one of the first to try Windows Live Mail.
 You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to
To unsubscribe from this group, send email to
For more options, visit this group at
Received on Sun Nov 26 2006 - 05:17:16 PST

This archive was generated by hypermail 2.3.0 : Fri Feb 16 2018 - 13:20:12 PST