[comp.ai.philosophy] THE I OF THE BEHOLDER

mcdermott-drew@cs.yale.edu (Drew McDermott) (06/13/91)

  In article <1991Jun12.232457.2962@news.media.mit.edu> minsky@media-lab.media.mit.edu (Marvin Minsky) writes:
  >Conversants never do, in
  >fact, know that they are talking about the same things.  It is always
  >a matter of convention, convergence, and good fortune -- even in the
  >case of "mathematical truths".  When you and I both talk about "that
  >chair over there", our internal models differ substantially, but not
  >enough to make most practical interactions too difficult.  And the
  >cchir itself changes imperceptibly from one moment to the next as it
  >loses and gains atoms and suffers thermal agitations of its internal
  >degrees of freedom.  There is no chair, indeed, from a modern physical
  >point of view, only boundaries imposed by observers....

Okay, but what about this objection: There are no observers, indeed,
but only boundaries imposed by .... who??

Why do we grant such rock-solid existence to observers and not to
chairs?

Surely we're not genuflecting toward the almighty self here?

I think this is a genuine conundrum, but whatever solution we work out
for explaining why people are objectively real will also work for
chairs.  In any case it will not do to say that the reality of 
macroscopic objects is merely imposed by an observer, because the
observer is itself just another macroscopic object.

                                             -- Drew McDermott

minsky@media-lab.media.mit.edu (Marvin Minsky) (06/14/91)

In article <1991Jun13.163734.10165@cs.yale.edu> mcdermott-drew@cs.yale.edu (Drew McDermott) writes:
>
>Okay, but what about this objection: There are no observers, indeed,
>but only boundaries imposed by .... who??
>
>Why do we grant such rock-solid existence to observers and not to
>chairs?

Speak for yourself.  Each mind has many poorly defined parts -- I call
them agencies.  It is only naive people -- that is, people who have
read too much philosophy -- that grant rock-solid existence to their
"self".  It is a pre-Freudian idea that an observer makes clear
distinctions, "believes" propositions, etc.  

ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/14/91)

In article <1991Jun13.163734.10165@cs.yale.edu> mcdermott-drew@cs.yale.edu
(Drew McDermott) writes:
>  In article <1991Jun12.232457.2962@news.media.mit.edu>
>  minsky@media-lab.media.mit.edu (Marvin Minsky) writes:
>  >Conversants never do, in
>  >fact, know that they are talking about the same things.  It is always
>  >a matter of convention, convergence, and good fortune -- even in the
>  >case of "mathematical truths".  When you and I both talk about "that
>  >chair over there", our internal models differ substantially, but not
>  >enough to make most practical interactions too difficult.  And the
>  >cchir itself changes imperceptibly from one moment to the next as it
>  >loses and gains atoms and suffers thermal agitations of its internal
>  >degrees of freedom.  There is no chair, indeed, from a modern physical
>  >point of view, only boundaries imposed by observers....
>
>Okay, but what about this objection: There are no observers, indeed,
>but only boundaries imposed by .... who??
>
>Why do we grant such rock-solid existence to observers and not to
>chairs?
>
>Surely we're not genuflecting toward the almighty self here?
>
>I think this is a genuine conundrum, but whatever solution we work out
>for explaining why people are objectively real will also work for
>chairs.  In any case it will not do to say that the reality of
>macroscopic objects is merely imposed by an observer, because the
>observer is itself just another macroscopic object.
>
Poor Descartes!  All that business about separating out mind and body has
really been taking quite a beating lately.  Now, just when you thought it
was safe to go into the philosophy library, here comes Drew ready to take
on the COGITO!  By the time we are all done with him, all Descartes will
have left to his name will be a few fragments of analytic geometry
(pathetically Euclidean, at that)!

However, if we can overcome our fear of solipsism (or at least Chris
Hutchison's fear), perhaps the COGITO is not quite as arbitrary as Drew's
accusation makes it out to be.  Ultimately, it all boils down to this question
of whether or not "genuflecting toward the almighty self" is nothing more than
blind faith.  After all, what the COGITO is basically saying is that because I
am exercising my "mental powers" (whatever they may be), I can attribute to
myself a "rock-solid existence" which I cannot attribute to that chair I
observe over there.

As I see it, here is where solipsism comes in the door to prop up the COGITO:
The question, as Drew formulated it, is quite appropriate.  Why should the
class of observers be any better off than the class of chairs?  The answer
provided by solipsism is that they are not any better off.  However, there
is ONE observer who IS better off;  and that is Drew's "almighty self."  The
solipsistic argument goes that the self is the ONLY observer that needs to be
taken into account.  Any other observers do, indeed, have the same status as
chairs:  They are all products of the interpretative process which constitutes
observation.  The COGITO then goes one step further by basically asserting
that the "self" behind that interpretative process is essentially an
emergent property of the process.

This is a bit convoluted and kind of heady.  It is probably better discussed in
a congenial bar over a few beers.  However, I do not see it as a patently silly
point of view which may impede our attempts to make progress in artificial
intelligence.  Minsky is quite right that we should "get on with the work
of making machines that can solve problems and communicate with one another
as best they can."  As we learn more about the technology of situated automata,
it becomes more and more feasible to think of building those machines on a
foundation of solipsism.  Indeed, from a point of view of sound engineering,
there may be no other viable way in which to build them.

===============================================================================

Stephen W. Smoliar
Institute of Systems Science
National University of Singapore
Heng Mui Keng Terrace, Kent Ridge
SINGAPORE 0511

BITNET:  ISSSSM@NUSVM

"He was of Lord Essex's opinion, 'rather to go an hundred miles to speak with
one wise man, than five miles to see a fair town.'"--Boswell on Johnson

mcdermott-drew@cs.yale.edu (Drew McDermott) (06/14/91)

  In article <9106140011.AA14996@lilac.berkeley.edu> ISSSSM@NUSVM.BITNET (Stephen Smoliar) writes:
  >In article <1991Jun13.163734.10165@cs.yale.edu> mcdermott-drew@cs.yale.edu
  >(Drew McDermott) writes:
    >>
    >>Okay, but what about this objection: There are no observers, indeed,
    >>but only boundaries imposed by .... who??
    >>
    >>Why do we grant such rock-solid existence to observers and not to
    >>chairs?
    >>
    >>Surely we're not genuflecting toward the almighty self here?
    >>
    >>I think this is a genuine conundrum, but whatever solution we work out
    >>for explaining why people are objectively real will also work for
    >>chairs.  In any case it will not do to say that the reality of
    >>macroscopic objects is merely imposed by an observer, because the
    >>observer is itself just another macroscopic object.

  >However, if we can overcome our fear of solipsism (or at least Chris
  >Hutchison's fear), perhaps the COGITO is not quite as arbitrary as Drew's
  >accusation makes it out to be.  Ultimately, it all boils down to this question
  >of whether or not "genuflecting toward the almighty self" is nothing more than
  >blind faith.  After all, what the COGITO is basically saying is that because I
  >am exercising my "mental powers" (whatever they may be), I can attribute to
  >myself a "rock-solid existence" which I cannot attribute to that chair I
  >observe over there.

The problem is this: It's now becoming clear that the self is a
construct of the brain.  [Okay, clear to a handful of computationally
enlightend people.]  Hence the sentence "I can attribute existence to
myself," actualy means that "the brain can attribute existence to
its 'self' ".  But deciding that the brain itself exists is not
addressed by this formula.

When I question the existence of a macroscopic object, I am not
invoking skepticism.  I'm just following up Minsky's point that
macroscopic objects are not simple collections of microscopic parts
(atoms, e.g.).  A river is not a collection of water molecules,
because the particular molecules involved are always changing.  Let's
call it a "system" of molecules for want of a better term.  The puzzle
is to find the systems in a universe of constantly moving quarks and
leptons.  It won't do to say "Find an observer and let it find the
systems," because the observers are just systems themselves.

  >As I see it, here is where solipsism comes in the door to prop up the COGITO:
  >The question, as Drew formulated it, is quite appropriate.  Why should the
  >class of observers be any better off than the class of chairs?  The answer
  >provided by solipsism is that they are not any better off.  However, there
  >is ONE observer who IS better off;  and that is Drew's "almighty
  self."  

In Descartes' day it might have been plausible to suppose that at
least one observer was directly observed.  But we now realize that the
brain consists of many "observers" (if the term has any meaning) that
apparently conspire to advertise the presence of a single virtual
observer.  This entity, being at best a self-fulfilling hallucination,
is in no position to serve as the foundation of all ontology and
epistemology. 

  >This is a bit convoluted and kind of heady.  It is probably better discussed in
  >a congenial bar over a few beers.  

What isn't?

   As we learn more about the technology of situated automata,
  >it becomes more and more feasible to think of building those machines on a
  >foundation of solipsism.  Indeed, from a point of view of sound engineering,
  >there may be no other viable way in which to build them.

I disagree completely with this idea.  The machines may someday arrive
at a solipsistic position, but if so, it will be false, as we can see
when we start to build them.

                                             -- Drew McDermott

minsky@media-lab.media.mit.edu (Marvin Minsky) (06/15/91)

Now I understand better what McDermott meant.  

I think I also understand better what Smoliar meant.  I rather liked
his use of the term "solipsism" to make a nice point about objective
truth being largely (or entirely) unavailable, so that our "knowledge"
is always to some degree really, only "belief" and that belief itself
is not ever clearly about "things" but is an internal repationship
between the person's "self-fulfilling hallucination conspiracy", as
McDermott put it, and other parts of that person's representational
data structures.  "I believe X" means, in that interpretation,
something sort of like "the imaginary person-self S that is me has
stored an X expression in the (imaginary) part of S's memory that is
used for expressions that are not to be easily changed."

ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/15/91)

In article <1991Jun14.153701.842@cs.yale.edu> mcdermott-drew@cs.yale.edu (Drew
McDermott) writes:
>
>When I question the existence of a macroscopic object, I am not
>invoking skepticism.  I'm just following up Minsky's point that
>macroscopic objects are not simple collections of microscopic parts
>(atoms, e.g.).  A river is not a collection of water molecules,
>because the particular molecules involved are always changing.  Let's
>call it a "system" of molecules for want of a better term.  The puzzle
>is to find the systems in a universe of constantly moving quarks and
>leptons.  It won't do to say "Find an observer and let it find the
>systems," because the observers are just systems themselves.
>
>  >As I see it, here is where solipsism comes in the door to prop up the
>  >COGITO:
>  >The question, as Drew formulated it, is quite appropriate.  Why should the
>  >class of observers be any better off than the class of chairs?  The answer
>  >provided by solipsism is that they are not any better off.  However, there
>  >is ONE observer who IS better off;  and that is Drew's "almighty
>  self."
>
>In Descartes' day it might have been plausible to suppose that at
>least one observer was directly observed.  But we now realize that the
>brain consists of many "observers" (if the term has any meaning) that
>apparently conspire to advertise the presence of a single virtual
>observer.  This entity, being at best a self-fulfilling hallucination,
>is in no position to serve as the foundation of all ontology and
>epistemology.
>
Why not?  Where are its points of weakness as far as support is concerned?  As
I understand your argument above, we cannot have either ontology or
epistemology until we have entities.  I am willing to join you in using
the word "systems" for those entities;  and I agree that finding those
systems is basically our "first problem."  Where I do NOT agree is with
your claim that observers are not qualified to find those systems by virtue
of being systems themselves.  Why should this disqualify them?  As I have been
trying to argue to Chris Hutchison, it may create a confusion which can only be
resolved by dialog among those observers;  but all this means is that one set
of hypotheses about what systems are may be displaced by another set.  This
would mean that any foundation we have for ontology and epistemology may not
be strictly static and is highly subjective, but to disqualify it as being a
foundation may be pushing the metaphor of a building's foundation too far.

===============================================================================

Stephen W. Smoliar
Institute of Systems Science
National University of Singapore
Heng Mui Keng Terrace, Kent Ridge
SINGAPORE 0511

BITNET:  ISSSSM@NUSVM

"He was of Lord Essex's opinion, 'rather to go an hundred miles to speak with
one wise man, than five miles to see a fair town.'"--Boswell on Johnson

G.Joly@cs.ucl.ac.uk (Gordon Joly) (06/17/91)

Drew McDermott writes:
 > 
 >   In article <1991Jun12.232457.2962@news.media.mit.edu> minsky@media-lab.media.mit.edu (Marvin Minsky) writes:
 >   >Conversants never do, in
 >   >fact, know that they are talking about the same things.  It is always
 >   >a matter of convention, convergence, and good fortune -- even in the
 >   >case of "mathematical truths".  When you and I both talk about "that
 >   >chair over there", our internal models differ substantially, but not
 >   >enough to make most practical interactions too difficult.  And the
 >   >cchir itself changes imperceptibly from one moment to the next as it
 >   >loses and gains atoms and suffers thermal agitations of its internal
 >   >degrees of freedom.  There is no chair, indeed, from a modern physical
 >   >point of view, only boundaries imposed by observers....
 > 
 > Okay, but what about this objection: There are no observers, indeed,
 > but only boundaries imposed by .... who??
 > 
 > Why do we grant such rock-solid existence to observers and not to
 > chairs?
 > 
 > Surely we're not genuflecting toward the almighty self here?
 > 
 > I think this is a genuine conundrum, but whatever solution we work out
 > for explaining why people are objectively real will also work for
 > chairs.  In any case it will not do to say that the reality of 
 > macroscopic objects is merely imposed by an observer, because the
 > observer is itself just another macroscopic object.
 > 
 >                                              -- Drew McDermott

As Rufus T. Firefly said, "any child of five can see that this is a
chair. Bring me a child of five..."

____

Gordon Joly                                       +44 71 387 7050 ext 3716
Internet: G.Joly@cs.ucl.ac.uk          UUCP: ...!{uunet,ukc}!ucl-cs!G.Joly
Computer Science, University College London, Gower Street, LONDON WC1E 6BT

                    Order is paramount in anarchy.

G.Joly@cs.ucl.ac.uk (Gordon Joly) (06/17/91)

Marvin Minsky writes:
 > Now I understand better what McDermott meant.  
 > 
 > I think I also understand better what Smoliar meant.  I rather liked
 > his use of the term "solipsism" to make a nice point about objective
 > truth being largely (or entirely) unavailable, so that our "knowledge"
 > is always to some degree really, only "belief" and that belief itself
 > is not ever clearly about "things" but is an internal repationship
 > between the person's "self-fulfilling hallucination conspiracy", as
 > McDermott put it, and other parts of that person's representational
 > data structures.  "I believe X" means, in that interpretation,
 > something sort of like "the imaginary person-self S that is me has
 > stored an X expression in the (imaginary) part of S's memory that is
 > used for expressions that are not to be easily changed."

A mere imagined hurricane in teacup.
Gordon.

____

Gordon Joly                                       +44 71 387 7050 ext 3716
Internet: G.Joly@cs.ucl.ac.uk          UUCP: ...!{uunet,ukc}!ucl-cs!G.Joly
Computer Science, University College London, Gower Street, LONDON WC1E 6BT

                    Order is paramount in anarchy.

dave@tygra.Michigan.COM (David Conrad) (06/19/91)

In article <1991Jun14.153701.842@cs.yale.edu> mcdermott-drew@cs.yale.edu (Drew McDermott) writes:
>
>The problem is this: It's now becoming clear that the self is a
>construct of the brain.  [Okay, clear to a handful of computationally
>enlightend people.]  Hence the sentence "I can attribute existence to
>myself," actualy means that "the brain can attribute existence to
>its 'self' ".  But deciding that the brain itself exists is not
>addressed by this formula.
No, it isn't, but so what?  What is most important, it seems to me, is that
an intelligent agent can achieve its goals without being able to prove that
anything, *not even the hardware on which it is implemented*, exists.  I can
think (I'm actually doing it right now!) without being able to *prove* that
my brain exists.  This is no obstacle.  The question for AI seems to be:
"Does an agent which 'believes' in objects, observers, etc. work 'better'
than a solipsistic one?  Does it 'learn' faster?  Can it do more?"  This,
I will admit, is out of my league.

>The puzzle
>is to find the systems in a universe of constantly moving quarks and
>leptons.  It won't do to say "Find an observer and let it find the
>systems," because the observers are just systems themselves.
>
Is this puzzle merely one of philosophy?  I cannot see why it is necessary
for us to solve the ultimate question of "Life, the Universe, and Everything"
before we can succeed with AI.  (The answer, by the way, is 42.)

>I disagree completely with this idea.  The machines may someday arrive
>at a solipsistic position, but if so, it will be false, as we can see
>when we start to build them.
>

If you have a disproof of Solipsism, I'm waiting with baited breath.  If you
simply mean that I know that the machine isn't the one true observer because
I know that I myself am also an observer, this is no more help to the machine
than you telling me that you really exist helps me to dispell my own
solipsistic feelings.

David R. Conrad
dave@michigan.com
-- 
=  CAT-TALK Conferencing Network, Computer Conferencing and File Archive  =
-  1-313-343-0800, 300/1200/2400/9600 baud, 8/N/1. New users use 'new'    - 
=  as a login id.  AVAILABLE VIA PC-PURSUIT!!! (City code "MIDET")        =
   E-MAIL Address: dave@Michigan.COM