Copy Link
Add to Bookmark
Report
AIList Digest Volume 2 Issue 059
AIList Digest Sunday, 20 May 1984 Volume 2 : Issue 59
Today's Topics:
Metaphysics - Perception, Recognition, Essence, and Identity
----------------------------------------------------------------------
Date: 15 May 84 23:33:31-PDT (Tue)
From: decvax!ittvax!wxlvax!rlw @ Ucb-Vax
Subject: A topic for discussion, phil/ai persons.
Article-I.D.: wxlvax.277
Here is a thought which a friend and I have been kicking around for a while
(the friend is a professor of philosophy at Penn):
It seems that it is IMPOSSIBLE to ever build a computer that can truly
perceive as a human being does, unless we radically change our ideas
about how perception is carried out.
The reason for this is that we humans have very little difficulty
identifying objects as the same across time, even when all the features of
that object change (including temporal and spatial ones). Computers,
on the other hand, are being built to identify objects by feature-sets. But
no set of features is ever enough to assure cross-time identification of
objects.
I accept that this idea may be completely wrong. As I said, it's just
something that we have been batting around. Now I would like to solicit
opinions of others. All ideas will be considered. All references to
literature will be appreciated. Feel free to reply by mail or on the net.
Just be aware that I don't log on very often, so if I don't answer for a
while, I'm not snubbing you.
--Alan Wexelblat (for himself and Izchak Miller)
(currently appearing at: ...decvax!ittvax!wlxvax!rlw Please put "For Alan" in
all mail headers.)
------------------------------
Date: 15 May 84 14:49:41-PDT (Tue)
From: ihnp4!houxm!hogpc!houti!ariel!norm @ Ucb-Vax
Subject: Re: A topic for discussion, phil/ai persons.
Article-I.D.: ariel.630
The computer needs to be able to distinguish between "metaphysically identical"
and "essentially the same". This distinction is at the root of an old (2500
years?) Greek ship problem:
Regarding Greeks ship problem: When a worn board is replaced by a new board,
the ship is changed, but it is the same ship. The difference leaves the
ship essentially the same but not identically the same. If all the boards
of a ship are replaced one by one until the ship is entirely redone with new
boards, it is still the same ship (essentially). Now, if all the old boards
that had been removed were put together again in their original configuration
so as to duplicate the new-board ship, would the new old-board ship be iden-
tically or essentially the same as the original old-board ship? Assume nailless
construction techniques were used thruout, and assume all boards always fit
perfectly the same way every time.
We now have two ships that are essentially the same as the original ship,
but, I maintain, neither ship is identical to the original ship. The original
ship's identity was not preserved, although its identity was left sufficiently
unchanged so as to preserve the ship's essence. The ship put together with
the previously-removed old boards is not identically the same as the original
old-board ship either, no matter how carefully it is put together. It too is
only essentially the same as the original ship.
A colleague suggested that 'essence' in this case was contextual, and I
tend to agree with him.
Actually, even if the Greeks left the original ship alone, the ship's identity
would change from one instant to the next. Even while remaining essentially
the same, the fact that the ship exists in the context of (and in relation to)
a changing universe is enough to vary the ship's identity from moment to mo-
ment. The constant changes in the ship's characteristics are admittedly very
subtle, and do not change the essential capacity/functionality/identity of the
ship. Minute changes in a ships identity have 'essentially' no impact. Only
a change sufficiently large (such as a small hole in the hull) have an
essential impact.
"Essence" has historically been considered metaphysical. In her "Introduction
to Objectivist Epistemology" (see your local bookstore) Ayn Rand identified
essence as epistemological rather than metaphysical. The implications of this
identification are profound, and more than I want to get into in this article.
Philosopher Leonard Peikoff's article "The Analytic-Synthetic Dichotomy", in
the back of the newer editions of Rand's Intro to Obj Epist, shows how crucial
the distinction between essence-as-metaphysical and essence-as-epistemological
really is.
Read Rand's book and see why the computer would have to make the same distinc-
tion. That distinction, however, has to be made on the CONCEPTUAL level. I
think Rand's discussion of concept-formation will probably convince you that
it will be quite some time before man-made machinery is up to that...
Norm Andrews, AT+T Information Systems (201)834-3685 vax135!ariel!norm
------------------------------
Date: 16 May 84 7:10:40-PDT (Wed)
From: hplabs!hao!seismo!rochester!rocksvax!sunybcs!gloria!rosen @ Ucb-Vax
Subject: Re: A topic for discussion, phil/ai persons.
Article-I.D.: gloria.176
Just a few quick comments,
1) The author seems to use perceive as visual perception. It can not
be a prerequisite for intelligence due to all the counter examples in
the human race. Not every human has sight, so we should be able to get
intelligence from various types of inputs.
2) Since humans CAN do it is the evidence that OTHER systems can do it.
3) The major assumption is that the only way a computer can identify objects
is by having static "feature-sets" that are from the object alone, without
having additional information, but why have that restriction? First,
all features don't change at once, your grandmother doesn't all-
of-a-sudden have the features of a desk. Second, the processor can/must
change with the enviornment as well as the object in question.
Third, the context plays a very important role in the recognition of
of an object. Functionality of the object is cruical. Remindings from
previous interactions with that object, and so on. The point is that
clearly a static list of what features objects must have and what features
are optional is not enough. Yet there is no reason to believe that
this is the only way computers can represent objects. The points
here come from many sources, and have their origin from such people
as Marvin Minsky and Roger Schank among others. There is a lot of
literature out there.
------------------------------
Date: 16 May 84 9:50:24-PDT (Wed)
From: hplabs!hao!seismo!rochester!ritcv!ccieng5!ccieng2!bwm @ Ucb-Vax
Subject: Re: Essence
Article-I.D.: ccieng2.179
I don't think ANYONE is looking to build a computer that can understand
phiolosophy. If I can build something that acts the same as an IQ-80 person,
I would be happy. This involves a surprising amount of work, (like vision,
language, etc.) but could certainly be confused by two 'identical' ships
as could I. Just because A human can do something does not imply that our
immediate AI goals should include it. Rather, first lets worry about things
ALL humans can do.
Brad Miller
...[cbrma, rlgvax, ritcv]!ccieng5!ccieng2!bwm
------------------------------
Date: 17 May 84 7:04:41-PDT (Thu)
From: ihnp4!houxm!hocda!hou3c!burl!ulysses!unc!mcnc!ecsvax!emigh @
Ucb-Vax
Subject: Re: the Greek Ship problem
Article-I.D.: ecsvax.2511
This reminds me of the story of Lincoln's axe (sorry, I've forgotten the
source). A farmer was showing a visitor Lincoln's axe:
Visitor: Are you sure that's Lincoln's axe
Farmer: It's Lincoln's axe. Of course I've had to replace the handle
three times and the head once, but it's Lincoln's axe alright.
Adds another level of reality to the Greek Ship Problem.
Ted H. Emigh Genetics and Statistics, North Carolina State U, Raleigh NC
USENET: {akgua decvax duke ihnp4 unc}!mcnc!ecsvax!emigh
ARPA: ecsvax!emigh@Mcnc or decvax!mcnc!ecsvax!emigh@BERKELEY
------------------------------
Date: 16 May 84 15:20:19-PDT (Wed)
From: ihnp4!drutx!houxe!hogpc!houti!ariel!vax135!floyd!cmcl2!seismo!ro
chester!rocksvax!sunybcs!gloria!colonel @ Ucb-Vax
Subject: Re: the Greek Ship problem
Article-I.D.: gloria.178
This is a good example of the principle that it depends on who's
doing the perceiving. To a barnacle, it's a whole new ship.
Col. G. L. Sicherman
...seismo!rochester!rocksvax!sunybcs!gloria!colonel
------------------------------
Date: 16 May 84 15:17:06-PDT (Wed)
From: harpo!seismo!rochester!rocksvax!sunybcs!gloria!colonel @ Ucb-Vax
Subject: Re: Can computers perceive
Article-I.D.: gloria.177
If by "perception" you imply "recognition", then of course computers
cannot perceive as we can. You can recognize only what is meaningful
to you, and that probably won't be meaningful to a computer.
Col. G. L. Sicherman
...seismo!rochester!rocksvax!sunybcs!gloria!colonel
------------------------------
Date: 16 May 84 10:57:00-PDT (Wed)
From: pur-ee!uiucdcs!marcel @ Ucb-Vax
Subject: Re: A topic for discussion, phil/ai pers - (nf)
Article-I.D.: uiucdcs.32300026
The problem is one of identification. When we see one object matching a
description of another object we know about, we often assume that the object
we're seeing IS the object we know about -- especially when we expect the
description to be definite [1]. This is known as Leibniz's law of the
indiscernability of identicals. That's found its way into the definitions
of set theory [2]: two entities are "equal" iff every property of one is also
a property of the other. Wittgenstein [3] objected that this did not allow for
replication, ie the fact that we can distinguish two indistinguishable objects
when they are placed next to each other (identity "solo numero"). So, if we
don't like to make assumptions, either no two objects are ever the same object,
or else we have to follow Aristotle and say that every object has some property
setting it apart from all others. That's known as Essentialism, and is hotly
disputed [4]. The choices until now have been: breakdown of identification,
essentialism, or assumption. The latter is the most functional, but not nice
if you're after epistemic certainty.
Still, I see no insurmountable problems with making computers do the
same as ourselves: assume identity until given evidence to the contrary. That
we can't convince ourselves of that method's epistemic soundness does nothing
to its effectiveness. All one needs is a formal logic or set theory (open
sentences, such as predicates, are descriptions) with a definite description
operator [2,5]. Of course, that makes the logic non-monotonic, since a definite
description becomes meaningless when two objects match it. In other words, a
closed-world assumption is also involved, and the theory must go beyond first-
order logic. That's a technical problem, not necessarily an unsolvable one [6].
[1] see the chapter on SCHOLAR in Bobrow's "Representation and Understanding";
note the "uniqueness assumption".
[2] Introduced by Whitehead & Russell in their "Principia Mathematica".
[3] Wittgenstein's "Tractatus".
[4] WVO Quine, "From a logical point of view".
[5] WVO Quine, "Mathematical Logic".
[6] Doyle's Truth Maintenance System (Artif. Intel. 12) attacks the non-
monotonicity problem fairly well, though without a sound theoretical
basis. See also McDermott's attempt at formalization (Artif. Intel. 13
and JACM 29 (Jan '82)).
Marcel Schoppers
U of Illinois at Urbana-Champaign
uiucdcs!marcel
------------------------------
End of AIList Digest
********************
