From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!usc!sdd.hp.com!mips!darwin.sura.net!cs.ucf.edu!news Tue Jun  9 10:05:59 EDT 1992
Article 6007 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!bonnie.concordia.ca!thunder.mcrcim.mcgill.edu!snorkelwacker.mit.edu!usc!sdd.hp.com!mips!darwin.sura.net!cs.ucf.edu!news
>From: clarke@acme.ucf.edu (Thomas Clarke)
Newsgroups: comp.ai.philosophy
Subject: Re: Hypothesis: I am a Transducer (Formerly "Virtual Grounding")
Message-ID: <1992Jun1.142749.8520@cs.ucf.edu>
Date: 1 Jun 92 14:27:49 GMT
References: <1992May31.145204.16357@Princeton.EDU>
Sender: news@cs.ucf.edu (News system)
Organization: University of Central Florida
Lines: 95

In article <1992May31.145204.16357@Princeton.EDU> harnad@phoenix.Princeton.EDU  
(Stevan Harnad) writes:
>          COMPUTATIONALISM = HOMUNCULARISM
>              (Or, I AM A TRANSDUCER)
> 
>  Although it is
> an over-simplification, consider my hypothesis to be that you ARE a
> transducer. If that hypothesis is correct, then there may be many
> different ways to implement you -- namely, all the different ways of
> implementing a transducer with your capabilities (TTT) -- but among
> those ways is definitely NOT one in which instead of a transducer there
> is a computer simulation of a transducer (a "virtual" transducer).

[Warning:  I am about to bring up quantum mechanics and chaos again
for those who find these topics uninteresting.]

I like to look at the problem of AI/consciousness from a physical point
of view.  I lean toward agreement with SH's conclusion that bare
computation cannot implement consciousness.  Unfortunately, this
raises the ugly question of where the consciousness resides.

>From the standpoint of classical phsyics, one can take the viewpoint of
Laplace that given initial conditions and perfect calculation, the
physical future can be perfectly predicted or simulated.  We have
learned that perfect calculation is captured by Turing machine
computation (Church-Turing thesis), so a modern day Laplacian would
ascribe perfect simulative capabilities to a suitably programmed
Turing machine.

Now if mind is physical, then from this viewpoint it can be 
Laplace-Turing simulated.  Since the L-T simlation is completely 
accurate, then everything about the mind, including conciousness, 
will be automatically captured by the L-T simulation.  The L-T simulation
of the mind may have to expand to include physical transduction,
perhaps even the entire universe, but given Laplace's classical view 
of physics, the Church-Turing thesis, and the physicality of
mind, the conclusion that consciousness can result from a computation
seems inevitable.

SH (and others) argue plausibly that computation alone cannot give 
consiousness.  If physicality is not abandoned, then either the C-T
thesis is false or Laplace's view is wrong.  C-T is only a hypothesis,
but nothing constructive is available that can exceed Turing machine
computation in power.  On the other hand, we know Laplace was wrong.

Laplace was wrong in two ways:  one the robustness of physical
calculation has been overthrown by chaos theory, and determinism
has been overthrown by quantum mechanics.  

Chaos is attractive in that it a purely "classical" phenomena
resulting only from the difficulties of non-linear calculation.
The problem is that Laplace is still right, it is only that the
initial conditions must be know to accuracy beyond that achievable
by any conceivable physical measurement.  Any error in the initial
conditions results ultimately in enormous error in the simulation.
But this is a problem only for mere mortals, not for Philosophical
Principals.

Quantum Mechanics is a horse of a different color.  It sets 
fundamental limits on what is simulable.  It is not possible
in principle to predict certain things.  Attempts to make quantum
mechanics classical and simulable by postulating an underlying reality 
lead to predictions that contradict experiment (e.g. Bell's inequality).
Worse (for physics) the standard attempts to interpret the math of
QM lead to says things like "the way function collapses upon
observation."  Observation?  by who?  Other interpretations lead
to equally mysterious things like:  
    many universes existing in parallel,  each containing a different 
version of every possible observation, or
    correlations between observations extending through all of space
acting at infinite speed, but having no locally observable
consequences.

Big difficulties for the Laplace-Turing simulation to capture all
quantum effects.  Some interpretations of quantum pheonomea require 
a conscious observer (?), others require simlation of every possible
world history, and yet others (equivalently) require taking into account 
all possilbe events in the entire universe.

Big opportunities for non-simulable, but physical phenomena to 
bring consciousness into the world in a way that cannot be simulated.
For example, the unobservable correlations traveling at infinite
speed sound a lot like a consciousness residing in the brain but 
having no observable behavioral effects.  

At any rate, to end this already too long post, there seems to be
ample room in the strange discoveries of 20th century physics for
conscious phenomena that Steven Harnad postulates cannot be simulated 
via a Turing machine.   

--
Thomas Clarke
Institute for Simulation and Training, University of Central FL
12424 Research Parkway, Suite 300, Orlando, FL 32826
(407)658-5030, FAX: (407)658-5059, clarke@acme.ucf.edu


