From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!cs.columbia.edu!mail Tue Jan 21 09:27:03 EST 1992
Article 2877 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!cs.columbia.edu!mail
Newsgroups: comp.ai.philosophy
Date: Sat, 18 Jan 92 16:41:39 EST
>From: "John R. Kender" <jrk@division.cs.columbia.edu>
Subject: Do birds cause flight?
Message-Id: <CMM.0.90.2.695770899.jrk@division.cs.columbia.edu>

There is a component to this "*really* understands" debate that I
haven't yet seen explored, and I would be grateful for those of you
who are more trained in analytic philosophy to comment on it.

It appears to me that much of the problem with words like
"understand", Minsky's "live", and simpler works like "fly", "see",
"talk", or even "cook", is that they evolve over time as technology
advances.  Much of the current debate might be attributable to the
cultural lag in fitting old words with new meanings.  Indeed, this is
how I interpret the intention of Turing's paper: it is an exercise in
the redefinition of the word "think".  Similarly, Searle's arguments
appear to be, at least in part, an attempt to reaffirm the
contemporary meaning of "understand" in the face of technological
challenge.

Let me elaborate.  The definition of "fly" currently encompasses at
least three distinct concepts (my ODE lists several others), all
equally correct in usage.  The first and oldest is "to move through
the air using wings", presumably biological ones.  The second is
simply "to move through air", as airplanes (and rocks) do.  The third
is "to move through the air using a conveyance", as people do.  Thus,
birds, planes, and people, all "fly", now.  Before, only birds (and
rocks) did, and it would be interesting to see what Webster's
original dictionary had to say about it.  

I don't say "I flew to New York on an airplane"; I say, simply, "I
flew"; the qualifier is not necessary.  Conversely, I sometimes
specifically refer to the "flight of birds" rather than simply
"flight", more or less as I say "hard disk" rather than "disk" now
that there are floppies.  These qualified nouns I have heard to be
called "retronyms".  Likewise, I "see" people and objects on
television: "Did you see George Bush last night" is now commonly
interpreted to mean "on television."  In fact, I am forced to say
"*in person*" if I "really" saw him.  Likewise, I "talk" to
people--on the phone.  I "cook" my food--in a oven, even though it is
also correct to say the heats cooks it or the oven cooks it.  

In all these cases I am using an older, more generic word.  But the
word has evolved, through the long-term recognition of the
similarities of function and end result, to encompass artificial
meanings co-equally with the "natural" ones.  The NRA makes a big
point of this: guns don't shoot people, people do.

Now consider the word "fly", which is probably less threatening than
"understand" since it is an attribute which humans beings are
generally not possessed of.  The mechanisms of flight are clearly
known and universal enough that many ways of achieving it are
possible, both natural and artificial, and in all three senses of the
word: birds, airplanes, and people fly.  

But, it may have been the case a thousand years ago, that the
medieval scholastics found it necessary to attribute to birds the
property of "fly-tensionality", there being no theory of fluid
dynamics, and even Bernoulli not yet born.  (Actually, from what I
recall from when I studied them, they would have spoken of birds of
having a "flying principle".)  It is not hard to imagine one of them
indicating that artificial flight was impossible, by analogous
argument to Searle.  To wit, a man-powered machine could not fly
because the man in it himself could not--even if the machine were
made small and light enough that it could be carried inside the man's
hand.  Or, said another way, no collection of bodily movements or
artificial parts by and of themselves, no matter how elaborate or
numerous, are sufficient to cause flight since each individual one
could not: each individual movement or part does not have the "flying
principle".  It is birds that cause flight.

Historically what *did* occur, however, is that other things made by
man *did* move through the air, and their motion was eventually
referred to as flying, although at first by qualifiers such as
"heavier than air".  The human who piloted them, or even the
passengers, were also said to "fly".  It is only in the past five
years that a man, through his movements, has flown in a way similar
to birds, in the Gossamer Condor.  And, the scholastics would have
gotten one thing right: it is very unlikely that the apparatus would
be simple enough to fit in the hand.

Speculating further, what may happen to the word "understand", is
that initially the behavior of trivial artificial symbol systems will
be described by the same word, with qualifiers.  One sees it already
in a layman's semi-humorously describing MS-DOS and other complex
software as "understanding" commands; the humor is the qualifier.
There is no intention in their usage to imply that some homunculus is
trapped in their PC; in time, however, what we call "understanding"
may become the retronym "human understanding", like "hard disk" or
"flight of birds".  (The word "computer" has gone through exactly
that evolution: in the 1940s, it was someone who calculated figures.
The intermediate stage of qualifier is apparent in the meaning of
ACM: Association for Computing Machinery.  The "natural" sense of
"computer" is now archaic.)

Continued technology may lead to the word "understanding" to have the
three senses of "fly": people understand, computers understand, and
people "conveyed" by computers will also understand.  In this last
case, there is already some evidence of the last sense: people work
out projections using spreadsheets, but refer to this symbiotic act
by statements like "I projected the data," not "I projected the data
using the computer."  However, and this is my point, it is this
evolving middle sense of "understand"--computers understand like
airplanes fly-- that Turing and Searle are fighting over.

To some extent, all the rest is up to engineering.  It may or may not
be possible to construct a machine that "understands" under given
constraints, just as the Gossamer Condor is at one extreme, and the
apparent impossibility of nuclear-powered aircraft is at the other.
The history of aviation has long since won the linguistic war, and is
now profitably and non-controversially concerned with traditional
engineering matters of efficiency and scale.  Aviation, however, has
the advantage of the underlying theories of physics.  The underlying
theories of cognition, in my opinion, remain closer to the "flying
principles" of the scholastics, making it difficult for the common man to
straightforwardly use "understand" with the same culturally accepted
agreement as it affords to the various senses of "fly".

John


