COMMENT ⊗   VALID 00004 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.xgp
C00013 00003	
C00014 00004	β
C00015 ENDMK
C⊗;
.xgp
.SQUISH
.font 0 fonts1;metm kst
.font 4 fonts1;metmbi kst
.tabfnt 1
.fill

.header Minsky      ∧    Reply to SEARLE   page 
On John Searle's paper: "Minds, Brains, and Programs"

		Marvin Minsky
		   M.I.T

The trouble is not within this technically brilliant
discussion but in an absurdity it assumes from the start --
that the everyday concepts of "intentionality" or of
"belief" are robust enough to permit such a precise
discussion.  Searle says without blushing:ε4 "I will simply
assert the following without argument.  The study of the
mind begins with such facts as that humans have beliefs,
while thermostats, telephones, and adding machines don't.
If you get a theory that denies this --- the theory is
false".ε0

This is so wrong.  The study of Mind is not the study of
Belief -- despite Searle's career commitment to that
position.  Instead, it is the attempt to discover powerful
and intelligible concepts -- be they old or new -- that help
explain why some machines or animals can do so much more
than others, e.g., manipulate processes that so effectively
control things in the world, perhaps via producing robustly
reliable "extensible" generalizations, internal models, or
whatever.

In the previous century, biological philosophers argued
about machines and life much as Searle argues about programs
and mind; one might have heard: ε4"The study of Biology
begins with such facts as that humans have →life_,
locomotives and telegraphs do not.  If you get a theory that
denies this --- the theory is false".  As all readers know,
we today understand much of "life" in terms of its
energetics and information-processing structures.  The
absolute, precise quality of "self-reproduction", for
example, survived quite recognizably in connection with DNA,
but had to be replaced by more complex ideas like encoding,
translation, and recombination, for explaining the
"reproduction" of sexual animals which do not, exactly,
"reproduce".

From our pre-scientific heritage we inherit many useful
idea-germs, like "dead-alive", "intellect-emotion",
"believe", know", "designate", "intend", and so forth.
Searle presumes these to be exclusively worthy of study,
while I see them as but flimsy suggestions left by our
failing ancestors.  They are useful in practical life, but
too crude to support powerful theories.  Now it would be
immodest (and unconvincing) for me to argue why my own new
ideas have more potential; instead I will explain the point
via a fantasy in which our future successors recount our own
plight:

.in 5
"The twentieth concept of 'Believe' proved inadequate until
replaced by a continuum in which, it turned out, stones
placed near zero, thermostats scored .072, and some men
achieved as high as 67.6.  According to modern concepts, It
appears theoretically possible for something to be believed
as intensely as 3600.  We were chagrined to learn that men
do not in fact believe very well nor, for that matter, are
they very proficient (on an absolute scale) at intending.
Still, they are comfortably separated from the thermostats."
.in 0

Of course, I'm joking, because I don't think a
one-dimensional continuum -- or indeed, any kind of scale,
will do.  Understanding how parts of a program or mind can
relate to other things outside that mind is complicated, not
because of some elusive aspect of the mythical "intention"
or "reference", but because different parts of the mind do
different things -- both with regard to externals and with
regard to each other.  The next fantasy explains why, and
perhaps also explains Searle's distress about the condition
of the person who incorporates into his mind -- "without
understanding" -- the hypothetical "squiggle-squoggle"
process that appears to understand Chinese:

.in 5
"The most pathetic error of our 20th Century precursors was
their unwitting adoption of the ancient formula 'A believes
that B means C' -- without questioning the conception of the
believer itself.  Because of this, they remained entrapped
(even Searle himself, until the 1980's) in what we today
call the 'single-agent fallacy' -- that inside each mind is
a single Knower, Believer, or Meaner of whom one can
predicate that 'A believes B'.  It is strange indeed how
long this philosophical obsession survived even after Freud
published his first clumsy portraits of our complex,
inconsistent and adversary mental constitutions.  To be
sure, we still find it useful to preserve in everyday life
much of the phenomenology that makes this seem so natural,
because the concept of an 'I' or a 'Self' seems still
indispensable for social purposes.  And it will probably
remain so, until our brain scientists become better able to
preserve, to rearrange, and to recombine (before or after
'death') those aspects of a brain's mind's parts that might
seem of value to our own successors."
.in 0

I wonder if Searle will see what I am getting at.  He talks
glibly of how we can "let the individual internalize all of
these elements of the system" and then complains that "there
isn't anything in the system that isn't in him".  We will
get nowhere if we keep seeking 'him' -- either somewhere in
the system or somehow in the whole system.  Perhaps Searle's
correspondents found it hard to refute his logic because
they too, fell into the same trap.  The point of whatever
Searle means by "Strong AI" is simply not, as he seems to
assume, to try to preserve and explain certain words like
"Self", "Mean", "Intend" and "Believe".  Real as these may
seem to us, they simply are not objective "things" that
confront us as ultimate theoretical problems or obstacles.
To be sure, they are objective as parts of our lexicons --
both in language and thought; nonetheless they are also only
clumsy, groping steps toward the better ideas I feel we are
approaching.  It is with discovering →those_ that concern
me, at least.

To an earlier, more "self-centered self" of my own, Searle's
arguments would have seemed the best of their kind.  But as
I write this, my phenomenology has some of the quality of a
double-exposure.  Searle's example about Chinese seems, to
the newer parts of my mind, much as though it were itself
written in Chinese.  I understand its syntax, I can parse
the sentences, and I can follow the technical deductions.
But the terms and assumptions themselves -- what the words
like 'intend' and 'mean' intend and mean -- escape the new
me.  They seem suspiciously like those very "formally
specified symbols" of his -- because their meanings, such as
they are, are incorporated into older parts of my mind that
are not in harmonious, agreeable contact with newer ones.  Yet
it is the newer ones that seem to be better able to untangle
the web of belief -- fittingly enough, through their
exploitation of new ideas and tools of "strong AI".
ew
β