From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!garrot.DMI.USherb.CA!uxa.ecn.bgu.edu!mp.cs.niu.edu!rickert Tue Jun  9 10:06:00 EDT 1992
Article 6008 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!bonnie.concordia.ca!garrot.DMI.USherb.CA!uxa.ecn.bgu.edu!mp.cs.niu.edu!rickert
>From: rickert@mp.cs.niu.edu (Neil Rickert)
Newsgroups: comp.ai.philosophy
Subject: Re: Hypothesis: I am a [sensory] Transducer
Keywords: computation, transduction
Message-ID: <1992Jun1.154133.27532@mp.cs.niu.edu>
Date: 1 Jun 92 15:41:33 GMT
References: <1992May31.145204.16357@Princeton.EDU> <1992Jun1.012608.3756@mp.cs.niu.edu> <1992Jun1.023730.20079@Princeton.EDU>
Organization: Northern Illinois University
Lines: 107

In article <1992Jun1.023730.20079@Princeton.EDU> harnad@phoenix.Princeton.EDU (Stevan Harnad) writes:
>In article <1992Jun1.012608.3756@mp.cs.niu.edu> rickert@mp.cs.niu.edu (Neil Rickert) writes:
>
>>  Much of my response will be on the theme of what a computer is.  But let
>>me start by pointing out that a computer IS a transducer.
>
>Of course I know this, and of course I know that a computation refers
>to an implemented computation when it's running on a computer, and that
>implementation means transduction of a certain kind. Just substitute
>"TTT robot's sensory transducer" for "transducer" and maybe the message
>will get through.

  Let's imagine a person isolated in a room.  There is a radio, a telephone,
a TV.  There are windows, but the windows are really TV displays of the
outside world obtained through cameras.  There is a loadspeaker listening
to the sounds of the outside world.

  I claim (assuming nutrition, etc), that the person will manage quite
well.

  No imagine that the TVs are all digital TVs.  The radio is a completely
digitized radio.  The telephone system uses ISDN, so is completely digitized.
All information about the outside world that enters the room is digital.

  The digital nature of the data changes nothing.  Everything looks the
same to the occupant of the room.  The occupant will manage quite well
in this digital room.  But the digital data could just as well be input
to a computer.

>                  All the analogies with how you get the data on a
>digital or analog disk into a computer are irrelevant.

  No they are not.  We can get the direct digital data in the same
format that it enters the room.  There is nothing special about whether
it was a disk or some other form of data.

>                                                       Think only of a
>TTT robot getting about in the real world of objects. And when you try
>to substitute simulated transducers for its real sense organs, don't
>think of getting data from a disk into core, but think of getting the
>robot to hear and react as you do to the same sounds you are hearing,
>see and react as you do to the same sights you are seeing.

  The robot has all the same information in the digital input to the room.
In principle it can be programmed to extract all of the same sights and
sounds.  It does not have any of the human transducers.  It is using the
transducers external to the room, and the human is also using those
external transducers.  Inside the room the information from the external
transducers is converted back to analog format purely to feed to the
human's own transducers.  But this is a step required only because of
the nature of the human input facility, and would otherwise be completely
superfluous.

>                                                           The devices
>that transduce that real acoustic oscillation and that real EM
>radiation are the ones in question. And do not assume that they

  In the situation I describe, the devices transducing the real acoustic
oscillation and EM radiation are the devices external to the room, and 
not the devices which are part of the human anatomy.

>                                    And do not assume that they
>immediately crunch it into symbols either

  You shouldn't assume that the computer immediately crunches the data
into symbols either.  This is the basic fallacy.  If a Searle-type
argument had been proposed in the 1950s, people like von-Neumann would
have laughed it out of the room.  Most of the "crunching into symbols"
is no more than an artifact of the programming languages that we use
today.  In the old days the data would be somewhere in memory where it
would be accessed by dynamically modifying machine instructions on the
fly, and there would be no assignment to symbols.

  This whole idea of everything in a computer being symbolic only makes
sense if you say that 0 and 1 are symbols, and the symbol representing
the data is the data itself.  But we could equally well say that a
neural signal is symbolic, and the symbol is the number of ions required
for the ionization that creates that signal.

>                                          -- they could in principle
>keep it analog all the way to the effectors (by which I likewise mean
>robotic output devices, not the components of a computer).

  You keep emphasizing "analog" as if there were something special about
analog.  There is NOTHING special about analog.  It is the information
content that counts, not the manner of representation.

  As I mentioned in a previous article, analog has one effect.  You can
not make the Searle-style claim that an analog signal is syntactic or
symbolic.  There are two possible ways to interpret this:

	(a)  There really is something special about analog
or
	(b)  This clearly demonstrates that Searle is wrong in referring
	     to everything in the computer as symbolic and syntactic.

  You have apparently fallen under the spell of the gospel according to
Searle, and believe (a).  But everything we know tells us that (a) is
nonsense, that there is nothing special in the analog representation of
information, and that digital representation is often preferable due to
its superior resistance to signal degradation.

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940


