Searle's Chinese Room
ADRIAN MOHAMMED
I need as much information as possible about Searle's Chinese Room
David Longley
In article <01bc2ac8$7260bee0$4341...@adrian.carib-link.net>
adr...@carib-link.net "ADRIAN MOHAMMED" writes:
> I need as much information as possible about Searle's Chinese Room
>
It's been vacated. In fact, many say that it never was occupied.
--
David Longley
I3uddha
adr...@carib-link.net "ADRIAN MOHAMMED" writes:
>> I need as much information as possible about Searle's Chinese Room
Da...@longley.demon.co.uk (David Longley):
>It's been vacated. In fact, many say that it never was occupied.
I hear it's been turned into a diner. "Searle's Chinese Diner."
There's special instructions to help you understand the menu.
-Bear
Jim Balter
ADRIAN MOHAMMED wrote:
>
> I need as much information as possible about Searle's Chinese Room
It's a demonstration that you can't understand the philosophy of
the mind merely by virtue of being known as a philosopher of the mind.
--
<J Q B>
cwtu...@nospam.ucs.indiana.edu
In article <19970308233...@ladder01.news.aol.com>,
I3uddha <i3u...@aol.com> wrote:
>adr...@carib-link.net "ADRIAN MOHAMMED" writes:
>>> I need as much information as possible about Searle's Chinese Room
>
>>It's been vacated. In fact, many say that it never was occupied.
>
>I hear it's been turned into a diner. "Searle's Chinese Diner."
>There's special instructions to help you understand the menu.
Diner or not, you still need reservations. It's best to call in the morning.
Carl Turner
my real address: ctu...@ai.uwf.edu
Jim Washington
i3u...@aol.com (I3uddha) wrote:
>adr...@carib-link.net "ADRIAN MOHAMMED" writes:
>>> I need as much information as possible about Searle's Chinese Room
>
>Da...@longley.demon.co.uk (David Longley):
>>It's been vacated. In fact, many say that it never was occupied.
>
>I hear it's been turned into a diner. "Searle's Chinese Diner."
>There's special instructions to help you understand the menu.
Yes. They have an excellent moo shu pork. Curiously, you cannot
order it unless they ask you first whether you want it.
-- Jim Washington
Robin Faichney
Jim Balter wrote:
>
> ADRIAN MOHAMMED wrote:
> >
> > I need as much information as possible about Searle's Chinese Room
>
> the mind merely by virtue of being known as a philosopher of the mind.
No. It's a demonstration that you can't generate understanding
in systems that just aren't equipped for it.
Like trying to discuss philosophy with those who insist that
there is no such thing, merely (bad) science.
--
Robin Faichney
r.j.fa...@stirling.ac.uk (delete ".nospam" to reply to me)
http://www.stir.ac.uk/envsci/staff/rjf1/
Andrew Buzzell
despite being a restaruant - there is no sign as the restaurant itself is
unable to know for certain that it is a restaurant - or that it is at all
- of if in fact it is possible to be - or at least to think that you be
whether or not you be at all.
----------------------------------------------------------------------------
Andrew Buzzell
CogSci & Philosophy
Carleton Intellect Hatchery
Ottawa, Ontario, Canada, North America, Earth, Milky Way, Universe, Your Mind
============================================================================
Jose Fernando Camoes Mendonca Oliveira Silva
Andrew Buzzell wrote:
> despite being a restaruant - there is no sign as the restaurant itself is
> unable to know for certain that it is a restaurant - or that it is at all
> - of if in fact it is possible to be - or at least to think that you be
> whether or not you be at all.
But is the restaurant the tables or the food? Because, paraphrasing Searle,
if the tables are not "restaurant" and the "food" is not restaurant, then
how can the room be "restaurant"????
Parts and wholes: Marvin Minsky, The Society of Mind, Touchstone, pg 27.
May incomplete self-referencing systems never cross your path,
JC (0000-0033)
--
To err is human. To forgive is not my policy.
Jose F. Silva @ MIT,E56-345, Cambridge MA 02139, USA <cam...@mit.edu>
http://web.mit.edu/camoes/public/home.html (617)253-3198 Fax:258-7597
Jose Fernando Camoes Mendonca Oliveira Silva
> > ADRIAN MOHAMMED wrote:
> > > I need as much information as possible about Searle's Chinese Room
> Jim Balter wrote:
> > It's a demonstration that you can't understand the philosophy of
> > the mind merely by virtue of being known as a philosopher of the mind.
r.j.fa...@stirling.ac.uk.nospam wrote:
> No. It's a demonstration that you can't generate understanding
> in systems that just aren't equipped for it.
I tend to think that it is an example of assuming that the whole
cannot be more than the sum of its parts, and in that it is misleading
and outright wrong. It just resonates with the desires of those who
fing the idea that "intelligence" can be explained threatening.
For an example of why wholes are greater than the sum of their parts,
just look at the containment property of a box. None of its walls
alone shows any containment, but a box has that property.
Good computations,
Jim Balter
Robin Faichney wrote:
>
> Jim Balter wrote:
> >
> > ADRIAN MOHAMMED wrote:
> > >
> > > I need as much information as possible about Searle's Chinese Room
> >
> > the mind merely by virtue of being known as a philosopher of the mind.
>
> in systems that just aren't equipped for it.
But it doesn't demonstrate that, as has been readily shown
by David Chalmers in _The Conscious Mind_, and many others before
him. Rather, it demonstrates that John Searle is not a particularly
competent thinker, nor are the many people who hold him to have
shown something just because they happen to want to believe it.
--
<J Q B>
cgu...@symantec.com
It either demonstrates that Searle is not a particularly competent
thinker or that Searle is quite good at building deceiving arguments.
The argument is basically an argument against the validity of the
Turing Test. It argues that even if a machine could pass the Turing
Test, there is still no "understanding" anywhere in the machine;
none of the mechanical components understands anything, therefore
the appearance of "intelligence" is a mere illusion.
This argument has a premise that for a whole to have a property,
its parts must also have that property. This premise is false.
Followed to the extreme, for materialists, Searle's argument leads
to solipsism. That is: nobody's neurons understand anything either.
Therefore, the appearance of anybody's consciousness is an illusion.
Darn it! I thought I'd just convinced myself I was conscious...
Chris Guzik
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
Robin Faichney
Jose Fernando Camoes Mendonca Oliveira Silva wrote:
>
> > > ADRIAN MOHAMMED wrote:
> > > > I need as much information as possible about Searle's Chinese Room
>
> > > It's a demonstration that you can't understand the philosophy of
> > > the mind merely by virtue of being known as a philosopher of the mind.
>
> > No. It's a demonstration that you can't generate understanding
> > in systems that just aren't equipped for it.
>
> cannot be more than the sum of its parts, and in that it is misleading
> and outright wrong. It just resonates with the desires of those who
> fing the idea that "intelligence" can be explained threatening.
>
> For an example of why wholes are greater than the sum of their parts,
> just look at the containment property of a box. None of its walls
> alone shows any containment, but a box has that property.
I guess I bear some share of the responsibility for this thread
degenerating to such seriousness. However, I'm sufficiently
stung to take a further step in the same direction.
I think the concept of wholes *sometimes* being > sum(parts) is
perfectly obvious to anyone with the intelligence to operate a
newsreader. However:
If you define "understanding" in the objective sense of "behaving
as if understanding" then the Chinese Room system as a whole
obviously does understand. If you define it in the subjective
sense of "having the experience of understanding" then the system
obviously (?) does not understand. The people who favour the
systems response obviously use the objective sense of the word,
and Searle and his supporters the subjective sense. They are
therefore talking past each other. Why do people still insist
that one set is right and the other wrong? Sorry, but to me it
looks like UTTER STUPIDITY!!
Any other explanations? :-)
Daryl McCullough
Robin Faichney wrote:
> If you define "understanding" in the objective sense of "behaving
> as if understanding" then the Chinese Room system as a whole
> obviously does understand. If you define it in the subjective
> sense of "having the experience of understanding" then the system
> obviously (?) does not understand.
It's not obvious to me! What reason do you have for saying that?
The only reason I can think of is that (1) Searle doesn't have
the experience of understanding, and (2) the rulebook doesn't
have the experience of understnading, therefore, the system doesn't
have the experience of understanding. In other words, you are
making the assumption that if none of the parts have the experience
of understanding, then the whole doesn't have the experience of
understanding.
> The people who favour the systems response obviously use the
> objective sense of the word,
No, they do not! If all you care about is behavior, then there is
no *need* for the systems response. The systems response is the
claim that the system, as a whole, has the experience of understanding,
although the man inside the room doesn't.
> and Searle and his supporters the subjective sense. They are
> therefore talking past each other.
No, they are not talking past each other. The people who disagree
with Searle understand what he is claiming, but disagree that
Searle has proved his claims. Searle hasn't shown that there
is no experience of understanding in the Chinese Room.
Daryl McCullough
CoGenTex, Inc.
Ithaca, NY
r.j.fa...@stirling.ac.uk
In article <33313D...@cogentex.com>,
Daryl McCullough <da...@cogentex.com> wrote:
>
> Robin Faichney wrote:
>
> > If you define "understanding" in the objective sense of "behaving
> > as if understanding" then the Chinese Room system as a whole
> > obviously does understand. If you define it in the subjective
> > sense of "having the experience of understanding" then the system
> > obviously (?) does not understand.
>
> It's not obvious to me! What reason do you have for saying that?
> The only reason I can think of is that (1) Searle doesn't have
> the experience of understanding, and (2) the rulebook doesn't
> have the experience of understnading, therefore, the system doesn't
> have the experience of understanding. In other words, you are
> making the assumption that if none of the parts have the experience
> of understanding, then the whole doesn't have the experience of
> understanding.
No. This is obviously (?) a cultural thing. Some people take it
that the proposition that a system consisting of a
(non-understanding) person plus a rulebook (etc) could experience
anything is so implausible that the onus of proof is on those who
insist that it does. That group obviously includes myself. Now
I know there is at least one person in the other set: who believe
that that proposition is perfectly plausible, and so the onus is
on those who disbelieve it to show good reason. I've been out of
this forum for quite a while (though I've continued to ponder
such issues) otherwise I guess I wouldn't have displayed such
naivety.
I still don't understand why some people think Searle denies
emergence, though. Is it just that that's all they can come up
with (given that they don't expect scepticism with regard to the
experiential capacity of a big box with a slot in the side and a
blindly-rule-following person within)?
> The systems response is the
> claim that the system, as a whole, has the experience of understanding,
> although the man inside the room doesn't.
Is there any systems response supporter reading this who disagrees
with Daryl?
> The people who disagree
> with Searle understand what he is claiming, but disagree that
> Searle has proved his claims. Searle hasn't shown that there
> is no experience of understanding in the Chinese Room.
I really don't think he set out to prove anything -- just to
elucidate the issue. He, like me, didn't really expect anyone
to want to attribute experience to the system as a whole, so
didn't think he bore the burden of proof. IMHO -- it's a
while since I read the paper and I'm only guessing at what was
in his mind as he wrote it.
--
Robin Faichney
r.j.fa...@stirling.ac.uk
http://www.stir.ac.uk/envsci/staff/rjf1/
Neil Rickert
In <333007...@stirling.ac.uk.nospam> Robin Faichney <r.j.fa...@stirling.ac.uk.nospam> writes:
>I think the concept of wholes *sometimes* being > sum(parts) is
>perfectly obvious to anyone with the intelligence to operate a
>newsreader. However:
>If you define "understanding" in the objective sense of "behaving
>as if understanding" then the Chinese Room system as a whole
>obviously does understand.
That sounds ok, always presuming that a CR could be built that would
behave as if understanding.
> If you define it in the subjective
>sense of "having the experience of understanding" then the system
>obviously (?) does not understand.
There is nothing obvious about that.
In you define "understanding" as something which is attributed to the
room by skeptical philosophers who are giving opinions without having
actually observed the behavior of the room, then perhaps it is
obvious that it does not understand in that sense.
Daryl McCullough
r.j.fa...@stirling.ac.uk wrote:
> Some people take it that the proposition that a system
> consisting of a (non-understanding) person plus a rulebook
> (etc) could experience anything is so implausible that the
> onus of proof is on those who insist that it does. That
> group obviously includes myself.
The onus of proof always falls on the person making the claim.
If Searle claims to prove that Strong AI is false, then it
seems to me that he is the one obligated to prove that the
system doesn't understand. On the other hand, if someone is
claiming to prove that Strong AI is true, then it is up to
that person to prove that the system *does* understand. As it
is, it is a standoff---neither side can prove their claims.
My point is that the disagreement between Searle and his critics
is *not* (as you claimed) over the meaning of "understand". I
don't think that Searle has made his case, even if I grant his
meaning of "understand".
> I really don't think he set out to prove anything -- just to
> elucidate the issue.
I disagree. Searle has certainly made statements to the effect that
Strong AI is provably false. I thought that the Chinese Room was
supposed to be a proof of that claim. It certainly isn't clear
to me what exactly it is elucidating.
> He, like me, didn't really expect anyone
> to want to attribute experience to the system as a whole, so
> didn't think he bore the burden of proof.
As I said, the person making the claim has the burden of proof.
What the Chinese Room shows is that Strong AI has counterintuitive
consequences. Counterintuitive, not contradictory, not obviously
false. I was trained in math and physics, and after learning about
Godel's Theorem, quantum mechanics, relativity and black holes,
I no longer feel that being counterintuitive is any indication
that a theory is incorrect. I wouldn't *expect* a theory of
mind to be intuitive.
Gary Forbis
cgu...@symantec.com wrote:
> The argument is basically an argument against the validity of the
> Turing Test. It argues that even if a machine could pass the Turing
> Test, there is still no "understanding" anywhere in the machine;
> none of the mechanical components understands anything, therefore
> the appearance of "intelligence" is a mere illusion.
All these mushy words bother me. I wish Searle had never ventured
into this area. "Understanding," "intelligence," "thinking,"
"consciousness," etc.... all mushy.
Here's something. The Turing Test is a test of the examiner not
the examined. It determines the gullibility of the examiner and
nothing else. Now give me an objective test of "intelligence"
any day and I'll be much happier. Do the same for "consciousness"
and I'll be deliriously happy.
> This argument has a premise that for a whole to have a property,
> its parts must also have that property. This premise is false.
I agree that a whole may have properties its parts do not. I don't
think this is his premise.
> Followed to the extreme, for materialists, Searle's argument leads
> to solipsism. That is: nobody's neurons understand anything either.
> Therefore, the appearance of anybody's consciousness is an illusion.
I sometimes take this stance. I have conscioiusness. I do not know
if anone else does. Some of my behaviors are the result of conscious
activity and regularly so. I can sometimes predict others behaviors
by assuming they have functions fulfilled by my consciousness and
it is easier for me to just think of them as having consciousness so
I do so. (I don't alway know the functions my consciousness use
and refer to some of them as intuitions.)
> Darn it! I thought I'd just convinced myself I was conscious...
You don't have to rely upon your having neurons to prove to
yourself you have consciousness and the neurons need not be sufficient
for consciousness so your having them may be besides the point.
--
--gary
for...@accessone.com
Jose Fernando Camoes Mendonca Oliveira Silva
r.j.fa...@stirling.ac.uk.nospam wrote:
> I think the concept of wholes *sometimes* being > sum(parts) is
> perfectly obvious to anyone with the intelligence to operate a
> newsreader. However:
But apparently not to Searlites.
> If you define "understanding" in the objective sense of "behaving
> as if understanding" then the Chinese Room system as a whole
And as an outside observer how will I define it? You see, I haven't yet
been able to melt minds with any construct to understand the construct's
internal processes by direct observation. Thus, I requires an observable
difference, and if that difference does not exist, by use of Occam's
shaving instrument...!
> obviously does understand. If you define it in the subjective
> sense of "having the experience of understanding" then the system
That does not compute: can you define "having the experience of
understanding"?
The *subjective* experience is defined as something that cannot be observed
and so cannot be measured and so has no real effect that we can point out
but must nevertheless believe in? What is this, religion?
> obviously (?) does not understand.
^^^^^^^^^
So you have a measure which allows you to determine that *obviously*,
which will make the tables "a subjective" restaurant. You can actually
go into a process of "thinking" and SEE that something other than mere
operations is happening? How do you see that? Note that introspection
is not a scientifically accepted response, as your brain has incentives
to trick you into reinforcing former beliefs if you are trying to prove
them.
> The people who favour the
> systems response obviously use the objective sense of the word,
> and Searle and his supporters the subjective sense.
which conveniently lacks a measurable/observable construct and
therefore has no external validity test possibilities...
> They are
> therefore talking past each other. Why do people still insist
> that one set is right and the other wrong?
Because Searlites push their example to forward the cause against
IA as much as they can. And Ai people don't like the common folk
being misled by fallacious use of words like "subjective" and "inner"
in a field of science and technology (as is AI).
> Sorry, but to me it looks like UTTER STUPIDITY!!
Is it subjective stupidity, or objective stupidity? Oh, it's like
Zen, the correct answer is attainable only out of logic and in
contact with our inner selves. [Which propts the following:]
That is if we are not on Prozac, which changes your inner self. Is the
Prozac-generated personality (which it creates by changing the chemical
balance in the brain) an objective self (yes, it can be measured) as well
as a subjective self (Who knows? People on Prozac will be in the altered
state and only their unaltered selves could know...).
> Any other explanations? :-)
Yes. Searle didn't think his example through, and relies on emotion to
push it through. His followers patch the example with "subjective",
"groovny obvious" and other stuff like that, and then call
"UTTER STUPIDITY" when people *cannot* see that there is something
MAGICAL (now that somebody pointed out that quantic micro-tubes would be
more likely to happen in the sub-micron dimension of silicon chips than
across neurons, thus defeating Penrose's ideas) in a human being that can
never be captured by a model (that is what the chinese room would be: a
lookup table model of pure sintactic translation).
May the set of sets not containing themselves plague your thoughts :-)
JC (0000-0033)
--
I conceive a God that no human mind can grasp. Hence, I'm an atheist.
Mark Young
Gary Forbis wrote:
> Here's something. The Turing Test is a test of the examiner not
> the examined. It determines the gullibility of the examiner and
> nothing else. Now give me an objective test of "intelligence"
> any day and I'll be much happier. Do the same for "consciousness"
> and I'll be deliriously happy.
That's like saying a sporting match is a test of the skills of the home
team but not the visiting team. More people will be fooled by Eliza
than by the following program:
Program Hello(Input, Output); Begin Writeln("Hello, world!") End.
Unless you'd go so far as to say that every computer program would fare
exactly as well as every other, then there's *something* about the
computer program being measured.
The examiner engages in one round of the Immitation Game. At the end
they either get the right answer or the wrong answer. So at the end
we've got an on/off measure: the human is gullible or not. The
(putative) AI engages in multiple rounds of the game, and so we get a
reasonable measure of how good that thing is in pretending to be human.
Now perhaps the examiner could engage in multiple rounds of the game,
too (no rule against that), and then we could compare the examiners
against each other and tell which humans are more gullible than which.
We can compare computer programs against one another the same way to see
which are better at pretending to be human, but this doesn't tell us
anything like what we want to know. To get something interesting we
compare the performance of the computer against the performance of men
in the male-female version of the Immitation Game. If you like we can
use the same examiners (tho' random assignment from a selected pool
would be better) to play this series. Then we can see which computers
do better than which men. Turing's criterion is passed by a given
computer if it does better than the average man does.
Whether one person is more gullible than another might be an interesting
thing for us to know, but as long as the two sets of examiners are
equally gullible (in the aggregate), it has no effect on the results of
the test.
So here's something for you: this computer is better at pretending to
be a human than the average man is at pretending to be a woman. We know
the man isn't a woman, so he has to *think* what the woman would say.
We know the computer isn't a human, so it has to *what* what the human
would say? Whatever *what* is, it's doing better than thinking. Turing
said we might as well call it thinking.
...mark young
Jonathan A. Maxwell
Mark Young wrote:
] Gary Forbis wrote:
[discussion]
From what I understand of the Turing Test, the following
would not be allowed:
Examiner: Remember the number 3578
... later ...
Examiner: If you reverse the digits in that number
I gave you what do you get?
Is this really so, and _why_ ? How can something be said to be
intelligent without a memory .. also shouldn't the examiner be
allowed to ask what the previous examiner was talking about ?
--
thur Mail Address: LordA...@vt.edu or jmax...@vt.edu
n r
a JAMax "Though it be long, the work is complete and finished
h o w in my mind. I take out of the bag of my memory what
tan lle has previously been collected into it." --Mozart
Andrzej Pindor
In article <333007...@stirling.ac.uk.nospam>,
Robin Faichney <r.j.fa...@stirling.ac.uk.nospam> wrote:
>Jose Fernando Camoes Mendonca Oliveira Silva wrote:
>>
>> > > ADRIAN MOHAMMED wrote:
>> > > > I need as much information as possible about Searle's Chinese Room
>>
>> > Jim Balter wrote:
>> > > It's a demonstration that you can't understand the philosophy of
>> > > the mind merely by virtue of being known as a philosopher of the mind.
>>
>> r.j.fa...@stirling.ac.uk.nospam wrote:
>> > No. It's a demonstration that you can't generate understanding
>> > in systems that just aren't equipped for it.
>>
>> I tend to think that it is an example of assuming that the whole
>> cannot be more than the sum of its parts, and in that it is misleading
>> and outright wrong. It just resonates with the desires of those who
>> fing the idea that "intelligence" can be explained threatening.
>>
>> For an example of why wholes are greater than the sum of their parts,
>> just look at the containment property of a box. None of its walls
>> alone shows any containment, but a box has that property.
>
>I guess I bear some share of the responsibility for this thread
>degenerating to such seriousness. However, I'm sufficiently
>stung to take a further step in the same direction.
>
>perfectly obvious to anyone with the intelligence to operate a
>newsreader. However:
>
>as if understanding" then the Chinese Room system as a whole
>sense of "having the experience of understanding" then the system
>systems response obviously use the objective sense of the word,
>therefore talking past each other. Why do people still insist
>looks like UTTER STUPIDITY!!
>
It would be fine except that, as far as I can see, the very notion of
"subjective" (like in "subjective sense of having the experience of
understanding") bars anyone, Searle included, from being able to make
firm pronouncement about anything/anyone else having "subjective sense"
of anything. If such pronouncement could be made the sense in question would
not be subjective anymore, would it? It is sad that people make such a simple
logical error. The only thing we can do is to _speculate_ about the presence
of the subjective sense _on the basis_ of behavior.
>Any other explanations? :-)
>
There no point of any explanations. Searle's argument is flawed as far as
sound scientific methodology is concerned and hence has no value whatsoever.
Andrzej
--
Andrzej Pindor The foolish reject what they see and
University of Toronto not what they think; the wise reject
Information Commons what they think and not what they see.
andrzej...@utoronto.ca Huang Po
Gary Forbis
Mark Young wrote:
>
> Gary Forbis wrote:
> > the examined. It determines the gullibility of the examiner and
> > nothing else. Now give me an objective test of "intelligence"
> > any day and I'll be much happier. Do the same for "consciousness"
> > and I'll be deliriously happy.
>
> That's like saying a sporting match is a test of the skills of the home
> team but not the visiting team.
I will accept I overstated the case. A examiner is mapping her or his
expectations against the observed behaviors.
> So here's something for you: this computer is better at pretending to
> be a human than the average man is at pretending to be a woman. We know
> the man isn't a woman, so he has to *think* what the woman would say.
> We know the computer isn't a human, so it has to *what* what the human
> would say? Whatever *what* is, it's doing better than thinking. Turing
> said we might as well call it thinking.
The other day I was eating something but I didn't know what it was.
It tasted better than a jelly sandwich. I guess I might as well call it
a jelly sandwich.
Sometimes I say a computer is "thinking." I'll use this language even
though
I consider "thinking" a counscious activity and will use different
language
when I turn over similar activity to my subconsciousness--"I think I'll
clear my mind for awhile and let the problem jell (brew, or what have
you.)"
My mind seems to be able to work on tasks subconsciously simultaneous
with
conscious activity working on some other task. I believe most reserve
"thinking" for the conscious process only.
Now one might complain about homunculii and infinite regress but there
is
no reason. One doesn't have multiple simultaneous consciousmesses.
Maybe
my consciousness is the result of some multitasking process but it has
the feel of continuity. Denying continuity may lead to sychronization
problems; it at least complicates things, especially in explaining why
some processes may said to be conscious while others sharing the same
hardware are not. (Oh, a consciousness bit tagging a process makes the
process conscious or not.)
--
--gary
for...@accessone.com
Anders N Weinstein
In article <8588793...@dejanews.com>,
<r.j.fa...@stirling.ac.uk> wrote:
>> The systems response is the
>> claim that the system, as a whole, has the experience of understanding,
>> although the man inside the room doesn't.
>
>Is there any systems response supporter reading this who disagrees
>with Daryl?
One response might be to argue simply that the system understands
Chinese, and drop the question whether or not it has an "experience of
understanding". For it seems to be a reasonable position that
accompaniment by special occurrent subjective experiences is not
necessary -- and perhaps not even sufficient -- for understanding a
language.
As I recall, Searle's argument is based on the property of having
("intrinsic") intentionality, states characterized by intentional
content. I gather Searle does take it that really "understanding
Chinese" requires having certain requisite intentional states *while one
understands*. Should we say these states constitute the "experience of
understanding"?
In other work Searle argues that although intentional states are not
all conscious -- e.g. your beliefs are not -- all are essentially
capable of becoming conscious states. This is because he claims the
discrimination of aspects (differences in sense) in intentional content
is related to the experiential point of view of a conscious subject. So
he concludes the idea of an intentional state that could never become
conscious is incoherent.
I suppose Searle can be taken to have a mentalistic theory of
understanding according to which certain occurrent subjective
experiences -- namely conscious intentional states -- are required to
accompany the understanding use of Chinese words. The subjective
experiences are needed to provide, as it were, the mentalistic gold
backing the public linguistic currency.
I myself think the best viewpoint here is the expressivist view I find
in Wittgenstein, which I take to be a fully satisfactory middle ground
between behaviorism and mentalism. The behaviorist takes the overt
observable manifestations of understanding to be meaningless by
themselves, but concludes that is enough. The mentalist agrees that the
overt manifestations are devoid of meaning, but naturally recoils and
takes it that a special sort of inner entity -- a soul or mind or brain
or inner control system -- is needed to imbue the overt symbolic
activity with meaning via the backing of mental acts. But this process
is hidden invisibly behind the observable surface, which presents only
a meaningless facade to an outside observer.
But on the expressivist view we don't really need to have any occurent
conscious experiences *while* we understand or use language with
meaning. For example, if I say "I was at the bank", or respond
appropriately to that sentence, I *don't* have to undergo any occurrent
mental state in which I experience the word "bank" as meaning "river
bank", say. But I *do* have to have a *disposition*, a disposition to
explain the meaning one way rather than another, or to draw certain
conclusions or go on to say certain further things.
In this way it can be perfectly true that "I heard the word as meaning
river bank" but *not* true that my operating with the word was meaningful
becuase I backed it up by inner conscious states which occurred before
or while I uttered it.
The *big* difference between expressivisim and behaviorism is that the
expressivist rejects the conception of the visible surface of
linguistic activity as meaningless. This view rejects the idea that one
could adequately describe in purely non-intentional vocabulary the
surface that an understanding speaker presents to a similarly
understanding observer. It might be impossible to describe this as a
kind of meaningless or purely formal "input-output" behavior. The
capacity for playing Wittgenstein's "slab" language-game, for example,
is not reducible to any purely formal "input-output" behavior, and a
computer could not do it (though a robot might).
So one of the decisive moves in the conjuring trick is mobilizing the
scientistic notions of "input" and "output" in the description of the
process of human symbolic activity. Since both behaviorism and mentalism
accept this reduction in the idea of what "observable behavior" is,
they are both doomed to forever oscillate from one pole to another.
The alternative is to recognize that the use of a language by a living
human being can, on occasion, be itself directly expressive of
mentality or subjective intentional states. One should not think of
the subjective states Searle-wise (or Descartes-wise) as prior or inner
antecedent *causes* hidden behind the overt uses of words. Nor should
one think the observable uses of words are intrinsically meaningless
behavior, although of course one can always artificially redescribe
them that way if one likes. Rather the intentionality can sometimes be
a manifest feature *in* the human activity with words, *informing* them
in something like Aristotelean form in matter. In this way we reject
the idea of a brute behavioristic level of description common to
both of the above evils.
To really clear the way for such a view, I think we need much more than
the Turing test provides for. I think the expressivist conception also
requires something like a living body which itself serves as a natural
medium of expression. I do not think you will ever find suitable
support for this in an automatic computing machine alone -- to adapt a
remark of Wittgenstein's, its surface is as it were too smooth to
enable one to see its activity as expressive of a psychology or
subjective point of view.
On the other hand there is no reason at all you could not expect to be
able to ascribe psychological predicates to a suitable sort of robot.
But from this point of view, the idea of ascribing psychological
predicates to an automatic computing machine alone verges on the
absurd, just as absurd as ascribing psychological predicates to a brain
(something that I think is only sensible in very bizarre but barely
imaginable circumstances).
Robin Faichney
Daryl McCullough wrote:
>
> [...]
I don't want to argue about whether Searle's aim was
proof or merely elucidation. But nevertheless, for me
this is the point at which this discussion has the
potential to get interesting.
As noted in a recent post to this group, I was surprised
to find myself in agreement with Jim Balter that the
attribution of consciousness is emotionally motivated.
My analysis of the systems response supporters versus
Searle debate about the Chinese Room is this: Searle
shares my emotionally-based attitude that humans and
members of (at least some) other species should be
assumed to be conscious, while human artifacts such
computers, robots and Chinese Rooms should be assumed
not to be so. The SR people, meanwhile, wish to
attribute consciousness to the Chinese Room because
they have an emotional investment in AI, identifying
with its aims. There is no rational reason to do so.
Any comments? :-)
r.j.fa...@stirling.ac.uk
In article <E7EoC...@gpu.utcc.utoronto.ca>,
pin...@gpu.utcc.utoronto.ca (Andrzej Pindor) wrote:
>
> Searle's argument is flawed as far as
> sound scientific methodology is concerned and hence has no value whatsoever.
So, you're a logical positivist! Well, I'm not going to argue with
that -- I'll just note that, for most serious people, science is the
domain par excellence of *facts*, not values, for which we look
elsewhere.
But, to return to the issue, if Searle's argument is flawed because
it invokes subjectivity, isn't the assertion that the system as a
whole experiences understanding, equally flawed, for precisely the
same reason?
Daryl McCullough
Robin Faichney wrote:
> My analysis of the systems response supporters versus
> Searle debate about the Chinese Room is this: Searle
> shares my emotionally-based attitude that humans and
> members of (at least some) other species should be
> assumed to be conscious, while human artifacts such
> computers, robots and Chinese Rooms should be assumed
> not to be so. The SR people, meanwhile, wish to
> attribute consciousness to the Chinese Room because
> they have an emotional investment in AI, identifying
> with its aims. There is no rational reason to do so.
>
> Any comments? :-)
No, a question. Do you believe that it is a fair rephrasing of
your post to say
Searle rejects Strong AI because of his emotional
predisposition. There is no rational reason to do
so. Others accept Strong AI because of their emotional
predisposition. There is no rational reason for them
to do so, either.
Is that what you meant? Or were you instead saying (as your
post implied)
Although Searle has his emotional predispositions,
they don't get in the way of his rational thinking
(at least not in his analysis of the Chinese Room).
In contrast, the proponents of Strong AI are unable
to think rationally (about the Chinese Room), because
of their emotional predispositions.
If it was the former, I am in agreement. There is no *rational*
reason to care about the attribution of consciousness, understanding,
experience, etc. On the other hand, if the latter is what you
meant (which is the way it appeared to me), then my comment is
this: Can't you see where your *own* emotional predispositions
are getting in the way of rational discourse? Talk about the
pot calling the kettle black!
Neil Rickert
In <3332B4...@stirling.ac.uk.nospam> Robin Faichney <r.j.fa...@stirling.ac.uk.nospam> writes:
>My analysis of the systems response supporters versus
>Searle debate about the Chinese Room is this: Searle
>shares my emotionally-based attitude that humans and
>members of (at least some) other species should be
>assumed to be conscious, while human artifacts such
>computers, robots and Chinese Rooms should be assumed
>not to be so. The SR people, meanwhile, wish to
>attribute consciousness to the Chinese Room because
>they have an emotional investment in AI, identifying
>with its aims. There is no rational reason to do so.
>Any comments? :-)
Yes. I think you have it quite wrong. Searle made the claim that he
had proven the CR would not understand. All the SR people need to
do, is to demonstrate that Searle did not prove what he claimed.
Thus the SR people do not need to attribute consciousness to the CR.
They need only point out that, in his argument, Searle did not
consider the possibility that the understanding would be in the
system as a whole, rather than in the components that Searle
considered. That is sufficient to debunk Searle's claims.
andrew babian
In article <5gv5uf$o...@usenet.srv.cis.pitt.edu>, ande...@pitt.edu (Anders N Weinstein) says:
>
>On the other hand there is no reason at all you could not expect to be
>able to ascribe psychological predicates to a suitable sort of robot.
>But from this point of view, the idea of ascribing psychological
>predicates to an automatic computing machine alone verges on the
>absurd, just as absurd as ascribing psychological predicates to a brain
I used to think that way for a long time--that there could be no
disembodied mind. But now i'm not so sure. It is after all a matter
for the system designer. The problem is essentially that a mind
learns from testing the world, reaching out to shape it and being
effected by it, and does not merely make decisions about the input.
So, because of this learning requirement, there must be some kind of
interaction, and because knowledge is a manipulation of experiences.
The idea of a disembodied mind assumes that you could somehow "gift"
it experiences, and that it would still be a mind even if it could not
have any more. In additional there is an idea of a simulated world as
being enough to satisfy a mind (which isn't too unreasonable given
our ability to fantasize). And in any case, we only experience our
internal model of the world anyway. None of this really relates to the
oversimplified approaches we talk about now.
So, there are some concerns about embodiment of mind, but they wouldn't
bother a sufficiently knowledgeable designer.
r.j.fa...@stirling.ac.uk
In article <3333EB...@cogentex.com>,
> > Searle debate about the Chinese Room is this: Searle
> > shares my emotionally-based attitude that humans and
> > members of (at least some) other species should be
> > assumed to be conscious, while human artifacts such
> > computers, robots and Chinese Rooms should be assumed
> > not to be so. The SR people, meanwhile, wish to
> > attribute consciousness to the Chinese Room because
> > they have an emotional investment in AI, identifying
> > with its aims. There is no rational reason to do so.
> >
> > Any comments? :-)
>
> your post to say
>
> Searle rejects Strong AI because of his emotional
> predisposition. There is no rational reason to do
> so. Others accept Strong AI because of their emotional
> predisposition. There is no rational reason for them
> to do so, either.
Yes (with reservations -- see below)
> Is that what you meant? Or were you instead saying (as your
> post implied)
>
> Although Searle has his emotional predispositions,
> they don't get in the way of his rational thinking
> (at least not in his analysis of the Chinese Room).
> In contrast, the proponents of Strong AI are unable
> to think rationally (about the Chinese Room), because
> of their emotional predispositions.
No
> If it was the former, I am in agreement. There is no *rational*
> reason to care about the attribution of consciousness, understanding,
> experience, etc. On the other hand, if the latter is what you
> meant (which is the way it appeared to me), then my comment is
> this: Can't you see where your *own* emotional predispositions
> are getting in the way of rational discourse? Talk about the
> pot calling the kettle black!
I guess you missed where I said "Searle shares my emotionally-based
attitude".
I just love it when Strong AI-ers get irrational! :-)
As for the reservations mentioned above, in fact I believe it
quite rational to reject Strong AI on the basis it represents
the attempt to mix scientific and non-scientific concepts in
a way that cannot work. That is quite consistent with my
"admission" that my differentiation between naturally occuring
organisms and artifacts is emotionally-based. Searle can be
read as saying something along these lines, but he's not clear
about it even in his own mind, never mind in his writing.
Then again, for me, Searle's actual intentions are not of any
great interest. The Chinese Room is just a relatively good
kick-off point for discussions, as shown here, yet again.
Anders N Weinstein
In article <E7Got...@cruzio.com>, andrew babian <aba...@cruzio.com> wrote:
>>
>I used to think that way for a long time--that there could be no
>disembodied mind. But now i'm not so sure. It is after all a matter
>for the system designer. The problem is essentially that a mind
My own view is that the claim derives more from reflection on our
ordinary psychological concepts, in particular on what is necessary for
determinate norms that constitute meanings. E.g what is required for you to
have a concept of tables that could succeed in referring to artifacts
in the common world?
>learns from testing the world, reaching out to shape it and being
>effected by it, and does not merely make decisions about the input.
>So, because of this learning requirement, there must be some kind of
>interaction, and because knowledge is a manipulation of experiences.
>
>The idea of a disembodied mind assumes that you could somehow "gift"
>it experiences, and that it would still be a mind even if it could not
>have any more. In additional there is an idea of a simulated world as
Well it seems to me as soon as you separate off "experiences" and treat them
as *inputs* to "the mind", you are already well on your way down the
Cartesian garden path. For if experiences are just external inputs, then
the thinking subject ("the mind") is only able to test its hypotheses and
beliefs by reference to these inputs -- it is trapped behind a veil of
ideas and can never really see what is beyond them.
If you reject that conception of "experiences" then the problem disappears.
For example, my noticing that my wastebasket is empty is an "experience" of
mine. But to characterize it thus is to relate me to objects and a state
of affairs in the world, not to a proximal representation or "input".
>our ability to fantasize). And in any case, we only experience our
>internal model of the world anyway. None of this really relates to the
I think this is nonsense and the main source of all the problems. For
example, when I look and see that my wastebasket is empty I can be said
to experience the wastebasket in the world itself, not an "internal
model" or proxy. If I hallucinate a wastebasket, then I am simply not
experiencing any object at all, but only seeming to. But the immediate
objects of veridical perceptual experiences are things in the world,
not models or representations in the mind.
There might be a sense in which our brains have models which *they*
'use' or 'consult' in some fashion when we who are phenomenological
subjects act and think. But we who can see the world directly don't
always have to rely on proximal models or representations -- often we
can just look and see instead.
Andrzej Pindor
In article <8590244...@dejanews.com>,
<r.j.fa...@stirling.ac.uk> wrote:
>In article <E7EoC...@gpu.utcc.utoronto.ca>,
> pin...@gpu.utcc.utoronto.ca (Andrzej Pindor) wrote:
>>
>> Searle's argument is flawed as far as
>> sound scientific methodology is concerned and hence has no value whatsoever.
>
>So, you're a logical positivist! Well, I'm not going to argue with
>that -- I'll just note that, for most serious people, science is the
>domain par excellence of *facts*, not values, for which we look
>elsewhere.
>
I should have been more explicit. By saying "value" above I meant 'scientific
value', i.e. that one cannot trust an argument based on flawed methodology.
Yes, science is the domain of facts, but it also uses logic to establish what
are facts and what are not and in search of a structure to these facts.
Scientific practice has established certain methodological principles which
help to avoid being tricked by our gullible minds into taking for facts things
which are not necessarily so. Searle demonstrates that philosphers routinely
flout these principles. Your statement "subjective understanding _obviously_
(my underlining) is not present" (or something to this effect) flouts
principles of scientific practice even more seriously since it is logically
incoherent (which I have pointed out and you have chosen to ignore).
>But, to return to the issue, if Searle's argument is flawed because
>it invokes subjectivity, isn't the assertion that the system as a
>whole experiences understanding, equally flawed, for precisely the
>same reason?
>
First of all I did not say that that the Searle's argument is flawed because
it invokes subjectivity. It is flawed because it does not specify criteria
which will be used to decide whether the understanding is found or not, _before_
opening the room and looking inside.
Yes, you are right that the assertion that the system as a whole _experiences
understanding_ is (equally) flawed. Now, who makes this assertion?
As far as I remeber, SR is that the system as a whole _understands_.
>--
>Robin Faichney
>r.j.fa...@stirling.ac.uk
>http://www.stir.ac.uk/envsci/staff/rjf1/
Andrzej
Jim Balter
r.j.fa...@stirling.ac.uk wrote:
>
> In article <33313D...@cogentex.com>,
> > > as if understanding" then the Chinese Room system as a whole
> > > obviously does understand. If you define it in the subjective
> > > sense of "having the experience of understanding" then the system
> > > obviously (?) does not understand.
> >
> > The only reason I can think of is that (1) Searle doesn't have
> > the experience of understanding, and (2) the rulebook doesn't
> > have the experience of understnading, therefore, the system doesn't
> > have the experience of understanding. In other words, you are
> > making the assumption that if none of the parts have the experience
> > of understanding, then the whole doesn't have the experience of
> > understanding.
>
> that the proposition that a system consisting of a
> (non-understanding) person plus a rulebook (etc) could experience
> anything is so implausible that the onus of proof is on those who
> insist that it does.
Excuse me, but this is idiotic. Searle took the onus upon himself.
He claimed to provide a proof that such a system does not understand.
You in turn take on the burden of proof as soon as you make the
positive claim that something is obviously true. Daryl pointed out
that you made an assumption, and that your conclusion doesn't
follow without it.
Beyond this, the proposition that any system understands is implausible,
yet some systems do. So yes, this is a cultural thing the culture
of willful ignorance and stupidity. (You introduced the word, so I
shall apply it where it fits.)
> > The people who disagree
> > with Searle understand what he is claiming, but disagree that
> > Searle has proved his claims. Searle hasn't shown that there
> > is no experience of understanding in the Chinese Room.
>
> I really don't think he set out to prove anything -- just to
> elucidate the issue. He, like me, didn't really expect anyone
> to want to attribute experience to the system as a whole, so
> didn't think he bore the burden of proof. IMHO -- it's a
> while since I read the paper and I'm only guessing at what was
> in his mind as he wrote it.
It show that it's been a while. Go back and read it and stop making
such stupid claims.
--
<J Q B>
Jim Balter
Robin Faichney wrote:
>
> Daryl McCullough wrote:
> >
> > [...]
>
> I don't want to argue about whether Searle's aim was
> proof or merely elucidation. But nevertheless, for me
> this is the point at which this discussion has the
> potential to get interesting.
>
> As noted in a recent post to this group, I was surprised
> to find myself in agreement with Jim Balter that the
> attribution of consciousness is emotionally motivated.
>
> Searle debate about the Chinese Room is this: Searle
> shares my emotionally-based attitude that humans and
> members of (at least some) other species should be
> assumed to be conscious, while human artifacts such
> computers, robots and Chinese Rooms should be assumed
> not to be so. The SR people, meanwhile, wish to
> attribute consciousness to the Chinese Room because
> they have an emotional investment in AI, identifying
> with its aims. There is no rational reason to do so.
>
> Any comments? :-)
Yes: that's not analysis, it is rank stupidity.
--
<J Q B>
Jim Balter
cgu...@symantec.com wrote:
> >> Jim Balter wrote:
> >But it doesn't demonstrate that, as has been readily shown
> >by David Chalmers in _The Conscious Mind_, and many others before
> >him. Rather, it demonstrates that John Searle is not a particularly
> >competent thinker, nor are the many people who hold him to have
> >shown something just because they happen to want to believe it.
> >
>
> It either demonstrates that Searle is not a particularly competent
> thinker or that Searle is quite good at building deceiving arguments.
Actually, it demonstrates both, unless Searle is a competent thinker
who *intends* his argument to deceive, but there is plenty of evidence
that he is more incompetent than nefarious. In any case, his
intellectual and philosophical methodology is certainly corrupt.
A typical example of what he uses in place of rigorous thinking is
to be found in his review of Chalmers' book, published in the March 1997
issue of The New York Review of Books as "Consciousness & the
Philosophers".
> The argument is basically an argument against the validity of the
> Turing Test. It argues that even if a machine could pass the Turing
> Test, there is still no "understanding" anywhere in the machine;
> none of the mechanical components understands anything, therefore
> the appearance of "intelligence" is a mere illusion.
>
> This argument has a premise that for a whole to have a property,
> its parts must also have that property. This premise is false.
Actually, I think the error is worse than that. Searle (and his
followers) have simply *mistaken* the mind of the Searle homunculus
to be the mind of the CR. This is quite clear from his ad reductio
premise, where he says (paraphrasing) "suppose my mind worked like a
computer". But in fact in the CR the workings of Searle's mind
are as unknown as ever; the experiment simply fails to test the premise.
As I said, this simply is not the product of a competent thinker.
A philosopher who is adept at manipulating words and phrases, yes,
but without the tools to check these manipulations for logical
accuracy. And it is a confusion between the two that gives his
followers some sense of confidence in their own views on the matter.
--
<J Q B>
Jim Balter
Jonathan A. Maxwell wrote:
>
> Mark Young wrote:
> ] Gary Forbis wrote:
>
>
> From what I understand of the Turing Test, the following
Fascinating that, in a discussion about "understanding", we have
such an interesting example of a use of the word.
Some piece of silliness that you got into your head from who-knows-where
does not constitute "understanding". By what process does
"I seem to recall that I heard somewhere" turn into "From what I
understand"?
> would not be allowed:
Who is going to disallow it?
> Examiner: Remember the number 3578
> ... later ...
> Examiner: If you reverse the digits in that number
> I gave you what do you get?
>
> Is this really so, and _why_ ?
You tell us.
> How can something be said to be
> intelligent without a memory ..
Yeah, Alan Turing was a real idiot.
> also shouldn't the examiner be
> allowed to ask what the previous examiner was talking about ?
No, that would have gone beyond the constraints of Alan Turing's kinky
lifestyle.
One way to begin to *understand* is to jettison the defective
intellectual methodologies that most people use in place of thinking.
--
<J Q B>
Jim Balter
Gary Forbis wrote:
>
> cgu...@symantec.com wrote:
>
> > The argument is basically an argument against the validity of the
> > Turing Test. It argues that even if a machine could pass the Turing
> > Test, there is still no "understanding" anywhere in the machine;
> > none of the mechanical components understands anything, therefore
> > the appearance of "intelligence" is a mere illusion.
>
> into this area. "Understanding," "intelligence," "thinking,"
> "consciousness," etc.... all mushy.
>
> the examined. It determines the gullibility of the examiner and
> nothing else. Now give me an objective test of "intelligence"
> any day and I'll be much happier. Do the same for "consciousness"
> and I'll be deliriously happy.
I remind you *again*, Gary, to reread Turing's paper, which proposed
a *comparison* between the judgements of machine imitating a woman
against the judgements of a man imitating a woman. For any given
judge, this is an *objective* test. Gullibility on the part of the
judge is neutral wrt man and machine.
> > This argument has a premise that for a whole to have a property,
> > its parts must also have that property. This premise is false.
>
> I agree that a whole may have properties its parts do not. I don't
> think this is his premise.
Please try to distinguish between statements and their inverses.
> > Followed to the extreme, for materialists, Searle's argument leads
> > to solipsism. That is: nobody's neurons understand anything either.
> > Therefore, the appearance of anybody's consciousness is an illusion.
>
> I sometimes take this stance. I have conscioiusness. I do not know
> if anone else does.
You talk about "mush", and then you go on with this gruel.
You have *what*? Certain neurophysiological states? A certain
brain organization? Subvocalizations? Interacting internal virtual
machines? What? What is it that you think you have? How about a
narrative generator and a narrative interpreter, where the narrative
generator generates narratives using symbols that the narrative
interpreter often leaves unresolved, like "consciousness"?
> You don't have to rely upon your having neurons to prove to
> yourself you have consciousness and the neurons need not be sufficient
> for consciousness so your having them may be besides the point.
Oh, you have a proof, do you? Why not share it? Or is "prove to
yourself" just a phrase applied to a certain sort of mush?
--
<J Q B>
Gary Forbis
Jim Balter wrote:
>
> Gary Forbis wrote:
> >
> > cgu...@symantec.com wrote:
> >
> > > The argument is basically an argument against the validity of the
> > > Turing Test. It argues that even if a machine could pass the Turing
> > > Test, there is still no "understanding" anywhere in the machine;
> > > none of the mechanical components understands anything, therefore
> > > the appearance of "intelligence" is a mere illusion.
> >
> > All these mushy words bother me. I wish Searle had never ventured
> > into this area. "Understanding," "intelligence," "thinking,"
> > "consciousness," etc.... all mushy.
> >
> > Here's something. The Turing Test is a test of the examiner not
> > the examined. It determines the gullibility of the examiner and
> > nothing else. Now give me an objective test of "intelligence"
> > any day and I'll be much happier. Do the same for "consciousness"
> > and I'll be deliriously happy.
>
> I remind you *again*, Gary, to reread Turing's paper, which proposed
> a *comparison* between the judgements of machine imitating a woman
> against the judgements of a man imitating a woman. For any given
> judge, this is an *objective* test. Gullibility on the part of the
> judge is neutral wrt man and machine.
Look, before one uses a test to identify members of a class one needs
to either assume or prove the test is sufficient for such
identification.
Most humans assume other humans have consciousness but machines do not.
They could be wrong but they make such an assumption. The test
specifically
disallows determining if the entity tested is a man or a machine and
relies upon behavior but behavior alone is not the normal way in which
membership is determined.
> > > This argument has a premise that for a whole to have a property,
> > > its parts must also have that property. This premise is false.
> >
> > I agree that a whole may have properties its parts do not. I don't
> > think this is his premise.
>
> Please try to distinguish between statements and their inverses.
I agree the asserted premise is false. I disagree that the argument
has the asserted premise. Further, I believe the argument is not
context free and before one can assert it is false one needs to show
that all possible interpretations are false, not that a specific
interpretation is false.
> > > Followed to the extreme, for materialists, Searle's argument leads
> > > to solipsism. That is: nobody's neurons understand anything either.
> > > Therefore, the appearance of anybody's consciousness is an illusion.
> >
> > I sometimes take this stance. I have conscioiusness. I do not know
> > if anone else does.
>
> You talk about "mush", and then you go on with this gruel.
> You have *what*? Certain neurophysiological states? A certain
> brain organization? Subvocalizations? Interacting internal virtual
> machines? What? What is it that you think you have? How about a
> narrative generator and a narrative interpreter, where the narrative
> generator generates narratives using symbols that the narrative
> interpreter often leaves unresolved, like "consciousness"?
"Consciousness" is not an unresolved sysbol. It resolves to a personal
property. It is unfortunate I cannot point to it for you. Maybe you
do not have a property that maps to social usage of "conscioiusness"
and leave it unresolved.
> > You don't have to rely upon your having neurons to prove to
> > yourself you have consciousness and the neurons need not be sufficient
> > for consciousness so your having them may be besides the point.
>
> Oh, you have a proof, do you? Why not share it? Or is "prove to
> yourself" just a phrase applied to a certain sort of mush?
Why don't you attempt to resolve the symbol "consciousness" to a
personal
property? You ask me to point to something in you I cannot observe but
merely assume (in my gullability) you have. When you resolve
"conscousness" to a personal property which provides a context in
which Searle's argument is true you will have "proved to yourself"
you do not have to rely upon your having neurons to do so.
The word is mushy because it has no publicly accessable referent. It
reslves to diferent referents in different people and I guess some just
leave it unresolved, relying upon some syntactic conventions to mask the
fact.
--
--gary
for...@accessone.com
Jim Balter
Robin Faichney wrote:
> If you define "understanding" in the objective sense of "behaving
> as if understanding" then the Chinese Room system as a whole
> obviously does understand. If you define it in the subjective
> sense of "having the experience of understanding" then the system
> obviously (?) does not understand.
Proclaiming this to be "obviously" true has nothing to do with Searle,
since Searle intends the absense of understanding in the Searle
homunculus to *be* the argument that the room lacks this "subjective
sense". But his argument is simply invalid; nothing interesting follows
from the lack of understanding of Chinese by the Searle homunculus.
David Chalmers gives a fairly succinct explanation of this failure
of logic in _The Conscious Mind_.
> The people who favour the
> systems response obviously use the objective sense of the word,
> and Searle and his supporters the subjective sense. They are
> therefore talking past each other. Why do people still insist
> that one set is right and the other wrong? Sorry, but to me it
> looks like UTTER STUPIDITY!!
What is utter stupidity is to so totally misunderstand the nature
of the argument after so much has been written about it.
The systems response is that lack of *subjective* understanding
by the Chinese room does not follow from lack of understanding the
Searle homunculus.
> Any other explanations? :-)
People who think the Searle argument is valid are incapable of
thinking clearly about the issues. This of course includes Searle.
This is not as controversial a claim as it may seem, since the errors in
Searle's "argument" are considered kid's stuff among logicians, despite
its popular standing.
--
<J Q B>
Jonathan A. Maxwell
Mark Young wrote:
] Jonathan A. Maxwell:
]
] There is no such restriction to the Turing Test. There may be such a
] restriction in the Loebner competition, but that's not the same thing.
Thanks Mark, that could be what I'm thinking of. I'll look into it.
And the other response:
Jim Balter wrote:
] Jonathan A. Maxwell wrote:
] >
] > [discussion]
] >
] > From what I understand of the Turing Test, the following
]
] Fascinating that, in a discussion about "understanding", we
] have such an interesting example of a use of the word. Some
] piece of silliness that you got into your head from
] who-knows-where does not constitute "understanding". By what
] process does "I seem to recall that I heard somewhere" turn
] into "From what I understand"?
Understand
4
c. To take or accept as a fact, without positive knowledge or
certainty; to get as an impression or idea; to believe. Chiefly
with obj. clause.
12
a. In parenthetic use (chiefly I understand): To believe or
assume, on account of information received or by inference.
That's from the OED. "From what I understand," in my experience
is a common phrase to mean the above. Were you grumpy the other
day?
] > would not be allowed:
]
] Who is going to disallow it?
I don't know, but if it's God then I pity the fool
who types it into the terminal!
] > Examiner: Remember the number 3578
] > ... later ...
] > Examiner: If you reverse the digits in that number
] > I gave you what do you get?
] >
] > Is this really so, and _why_ ?
]
] You tell us.
Because it's much harder to code a program to answer
any trick question a person could come up with than
to get it to seem like it can carry on a normal conversation.
But I was thinking of another test, it seems.
] > How can something be said to be
] > intelligent without a memory ..
]
] Yeah, Alan Turing was a real idiot.
It takes all kinds.
] > also shouldn't the examiner be allowed to ask what the
] > previous examiner was talking about ?
]
] No, that would have gone beyond the constraints of Alan
] Turing's kinky lifestyle.
I don't get it, was that sarcasm?
] One way to begin to *understand* is to jettison the defective
] intellectual methodologies that most people use in place of
] thinking.
Like not assuming a definition of 'understand'?? ;)
Robin Faichney
Andrzej Pindor wrote:
>
(RF: Sorry for causing that digression)
>Your statement "subjective understanding _obviously_
>(my underlining) is not present" (or something to this effect) flouts
>principles of scientific practice even more seriously since it is logically
>incoherent (which I have pointed out and you have chosen to ignore).
I'm not particularly wedded to that statement. In fact, I think I
had a "(?)" in the original. But your criticism is off-target anyway,
because I do not restrict myself to scientific methodology. In this
context, I see myself as a philosopher rather than a scientist. But
rather than pursue this any further, I'll happily drop that particular
assertion. Seems to me the discussion moved into more interesting
areas a few days ago.
> Yes, you are right that the assertion that the system as a whole _experiences
> understanding_ is (equally) flawed. Now, who makes this assertion?
> As far as I remeber, SR is that the system as a whole _understands_.
That's what I thought, but Daryl McCullough said otherwise, and noone
(else) in the several followups since has contradicted him, some
seeming in fact at least by implication to agree with him. It's a
pity you weren't "listening" when I explicitly asked, in my response
to the post in which he said that, whether anyone disagreed with him.
(But I am familiar with the unreliable nature of usenet.)
r.j.fa...@stirling.ac.uk
I hope noone is breathlessly awaiting my reply to any of
the ad hominems.
So that I know what I'm aiming at -- would the "SR people"
around here please try to reach some concensus as to
whether the system as a whole experiences understanding?
Of course, I would understand (!) if they'd rather play
down this issue.
While we're at it -- and I hope I'm not over-complicating
things by adding another question here -- if the Chinese
Room is *not* claimed to experience understanding, should
we or should we not interpret "the CR understands"
behaviourally, i.e. to mean no more nor less than "the CR
behaves like systems to which we generally attribute
understanding"?
--
Robin Faichney
r.j.fa...@stirling.ac.uk
http://www.stir.ac.uk/envsci/staff/rjf1/
-------------------==== Posted via Deja News ====-----------------------
scan...@delphi.com
John Searle's argues that syntax does not, can not lead to
semantics. This is the common delusion that leads to the
construction of unbreakable codes and ciphers. The National
Security Agency proceeds on the opposite assumption, that syntax
leads to semantics. History is on the side of NSA. If there is
information in the text it will evidence itself. Everyday, at Fort
Meade, mathematicians deduce semantics from syntax. Since the
Chinese writing enters she Chinese Room with no disguise of
randomness of information, we should soon understand it.
Since Searle's argument does not stand, we wonder why he presses
it. I think I recognize him as a latter day vitalist and a closet
dualist. He has a notion that "thinking" is associated with "life",
that DNA begets semantics. Searle would deny this, but if DNA is
not special then why argue?
Searle is officially a materialist and he would fight against an
accusation of dualism, but his "understanding" suggests a thinker.
A thinker who selects from data proffered by the brain,
recognizes the data (semantics), manipulates it, decides, and
returns the decision to the brain for execution. If there is no
thinker then why a Chinese Room?
Vitalism and thinking explain Searle.
Ray
Neil Rickert
In <8592977...@dejanews.com> r.j.fa...@stirling.ac.uk writes:
>I hope noone is breathlessly awaiting my reply to any of
>the ad hominems.
>So that I know what I'm aiming at -- would the "SR people"
>around here please try to reach some concensus as to
>whether the system as a whole experiences understanding?
Sure. Now, which system are you talking about? If you have an
actual working example of a Chinese Room, which performs up to specs,
I will be happy to try to examine how well it does and give my
opinion as to whether the system as a whole experiences
understanding.
If, however, your question is only hypothetical, then you should not
expect more than a hypothetical answer. My hypothetical answer is
that no system implemented in accordance with Searle's CR description
could possibly have behavior which would come close to suggesting
that we should consider an attribution of understanding.
>Of course, I would understand (!) if they'd rather play
>down this issue.
Should I take this as your impersonal kind of ad hominem.
Neil Rickert
In <xLD6rK3....@delphi.com> scan...@delphi.com writes:
>John Searle's argues that syntax does not, can not lead to
>semantics. This is the common delusion that leads to the
>construction of unbreakable codes and ciphers. The National
>Security Agency proceeds on the opposite assumption, that syntax
>leads to semantics. History is on the side of NSA. If there is
>information in the text it will evidence itself.
I think you would have trouble sustaining this as an argument against
Searle. A response might be:
Sure, there is information in the text. It was placed there by the
humans who instituted the cipher, and it might be extracted by the
humans working for the NSA. But there is no semantics in the
computers doing the syntactic processing of the coded material.
> Everyday, at Fort
>Meade, mathematicians deduce semantics from syntax.
Sure. But these are human mathematicians. The machines deal only
with the syntactic operations.
> Since the
>Chinese writing enters she Chinese Room with no disguise of
>randomness of information, we should soon understand it.
That was not being questioned. The question was not whether we
understand it, but whether the CR understands it.
>Since Searle's argument does not stand, we wonder why he presses
>it.
I agree that Searle's argument does not stand. The argument does not
stand because of serious flaws in the argument, rather than because
some counter example has been produced. I suspect that what makes it
seems so attractive is that so many inadequate answers have been
given.
cgu...@symantec.com
In <333756...@netcom.com>, Jim Balter <j...@netcom.com> writes:
>cgu...@symantec.com wrote:
>
>> The argument is basically an argument against the validity of the
>> Turing Test. It argues that even if a machine could pass the Turing
>> Test, there is still no "understanding" anywhere in the machine;
>> none of the mechanical components understands anything, therefore
>> the appearance of "intelligence" is a mere illusion.
>>
>> its parts must also have that property. This premise is false.
>>
>
>followers) have simply *mistaken* the mind of the Searle homunculus
>to be the mind of the CR. This is quite clear from his ad reductio
>premise, where he says (paraphrasing) "suppose my mind worked like a
>computer". But in fact in the CR the workings of Searle's mind
>are as unknown as ever; the experiment simply fails to test the premise.
>As I said, this simply is not the product of a competent thinker.
>A philosopher who is adept at manipulating words and phrases, yes,
>but without the tools to check these manipulations for logical
>accuracy. And it is a confusion between the two that gives his
>followers some sense of confidence in their own views on the matter.
Maybe I'm giving Searle way too much credit, but it seems hard to
believe he could be guilty of such a horrible error as that.
I had always interpreted the CR argument as:
(1) Suppose we have a program P which is claimed to have understanding
of a language L.
(2) P can be implemented as a CR.
(3) This implementation consists of only (a) the homunculus, (b) the
input/output symbols, and (c) the rules by which input strings
are translated to output strings.
(4) Neither (a) the homunculus, (b) the symbols themselves, nor (c)
the translation rules, have any understanding of the language L.
(5) By (3) there is nothing else there to have understanding of L.
(6) Therefore, by (4) and (5), the CR does not understand L.
Hence my above claim that one of his premises is that "for a whole
to have a property, (I should have said: 'at least one of') its
parts must also have that property".
This false premise leads Searle to assert (5) and deny (or ignore?)
the possibility of emergent behavior.
This seems to me to be a variant of what Turing called the "argument
from consciousness" in the "Computing Machinery and Intelligence".
Then again, it's been a while since I've read Searle, so maybe
I'm getting my arguments crossed.
Chris Guzik
Andrzej Pindor
In article <5h6bjv$3...@usenet.srv.cis.pitt.edu>,
Anders N Weinstein <ande...@pitt.edu> wrote:
>In article <E7Got...@cruzio.com>, andrew babian <aba...@cruzio.com> wrote:
>>>
>>I used to think that way for a long time--that there could be no
>>disembodied mind. But now i'm not so sure. It is after all a matter
>>for the system designer. The problem is essentially that a mind
>
>My own view is that the claim derives more from reflection on our
>ordinary psychological concepts, in particular on what is necessary for
>determinate norms that constitute meanings. E.g what is required for you to
>have a concept of tables that could succeed in referring to artifacts
>in the common world?
>
>>learns from testing the world, reaching out to shape it and being
>>effected by it, and does not merely make decisions about the input.
>>So, because of this learning requirement, there must be some kind of
>>interaction, and because knowledge is a manipulation of experiences.
>>
>>The idea of a disembodied mind assumes that you could somehow "gift"
>>it experiences, and that it would still be a mind even if it could not
>>have any more. In additional there is an idea of a simulated world as
>
>Well it seems to me as soon as you separate off "experiences" and treat them
>as *inputs* to "the mind", you are already well on your way down the
So if experiences are not *input* to the mind, where do they come from? are
they generated internally? If not fully internally but in some correlation
with inputs, what is the remaining constutent of the experiences? Where does
it come from?
>Cartesian garden path. For if experiences are just external inputs, then
>the thinking subject ("the mind") is only able to test its hypotheses and
>beliefs by reference to these inputs -- it is trapped behind a veil of
>ideas and can never really see what is beyond them.
>
The last part of the above sentence does not square with the begining part -
where did 'ideas' come from? The mind is indeed trapped, but behind (veil
or not) inputs. Its only contact with the world is through these inputs. If
you disagree, please explain what are the other ways it makes contact with
the external world. Have you heard about virtual reality? Are you not aware
that the mind is unable to tell where do the inputs come from - from a real
world or from a computer generated world? You seem to refuse to accept
clear empirical evidence. Or do you believe in telepathy, and such which is
mediated by non-physical medium? Even most of those who believe in telepathy
would agree that it provides *input* to the mind, if by some unknown medium.
So how does mind "see" beyond inputs (ideas?)?
>If you reject that conception of "experiences" then the problem disappears.
This is a problem invented by yourself (and perhaps some other philosophers).
>For example, my noticing that my wastebasket is empty is an "experience" of
>mine. But to characterize it thus is to relate me to objects and a state
>of affairs in the world, not to a proximal representation or "input".
>
So is you noticing that your wastebasket is empty an "an experience" of yours,
because it is characterised as such? If it was not characterised as such
by a philospoher, it would not be that? If there were no philosophers, then
there would not be such problems, would there? If a philosopher has not
related you to objects and affairs of the world, would it all not be simpler,
just inputs to the mind? You seem to suggest that this other constituent of
experiences (besides input) is provided by philosophers relating objects
and affairs of the world. If I misread you, please tell me what is this
other constituent of experiences. If I read you correctly, then getting rid of
philosphers would seem to be the simplest way of solving the problem you
refer to. This is what I have been saying all along.
>>our ability to fantasize). And in any case, we only experience our
>>internal model of the world anyway. None of this really relates to the
>
>I think this is nonsense and the main source of all the problems. For
>example, when I look and see that my wastebasket is empty I can be said
^^^^^^^^^^^^^
You can, so what? Paper (screen) is patient, just saying something does not
make it true, as you keep thinking.
>to experience the wastebasket in the world itself, not an "internal
>model" or proxy. If I hallucinate a wastebasket, then I am simply not
>experiencing any object at all, but only seeming to. But the immediate
Again, just saying so does not make it true. Please explain the difference
between seeing a "real" wastebasket and 'hallucinating' one, due to being
in virtual reality environment, feeding corresponding signals directly
to your optic nerves. Difference to the mind, of course.
>objects of veridical perceptual experiences are things in the world,
>not models or representations in the mind.
>
And how does mind know whether it is dealing with "things in the world" or
with artificially generated inputs?
>There might be a sense in which our brains have models which *they*
>'use' or 'consult' in some fashion when we who are phenomenological
>subjects act and think. But we who can see the world directly don't
^^^^^^^^
And what is this "directly"? How does it come about? See the questions above
re. virtual reality environment.
>always have to rely on proximal models or representations -- often we
>can just look and see instead.
>
When we "look and see" we provide our minds/brains with physical inputs.
Do we have a way of 'seeing' beyond these inputs? If you still think so,
please elaborate.
andrew babian
In article <5h9157$5...@ux.cs.niu.edu>, ric...@cs.niu.edu (Neil Rickert) says:
>
>If, however, your question is only hypothetical, then you should not
>expect more than a hypothetical answer. My hypothetical answer is
>that no system implemented in accordance with Searle's CR description
>could possibly have behavior which would come close to suggesting
>that we should consider an attribution of understanding.
ok, a human person probably wouldn't be smart enough, or fast enough,
or bored enough to actually try to operate a chinese room. But what if,
in a hundred years or so when we have artificial people, on a lark one
of those people, (or a simulation of one of those people for time and
energy constraints) gets into a chinese room and starts flipping through
the rule and scrathing on the notepads, and works on a few questions.
After a couple of years, having only answered a handful of questions,
will he finally decide he has had enough and just ask for a chinese
dictionary?
Neil Rickert
This is still pretty hypothetical. If these artificial people can
properly be referred to as people, they will probably also be too
intolerant of boredom to be able to run a CR. But from my
perspective, the Chinese Room scenario misses the point anyway.
(1) It presupposes that a system which rigidly follows rules to
symbolic data could ever behave like a person. I am doubtful
that we shall ever have a sufficiently detailed set of rules
for this to be possible.
(2) Searle's scenario omits the well known fact that humans learn.
In the unlikely event that he could follow the rules rapidly
enough to generate the right behavior, we would expect Searle
to learn new ways of doing things himself, so that quite
possibly Searle would learn how to short circuit the prescribed
rules and would come to understand what he was processing.
Jim Balter
r.j.fa...@stirling.ac.uk wrote:
>
> I hope noone is breathlessly awaiting my reply to any of
> the ad hominems.
"Sorry, but to me it looks like UTTER STUPIDITY!!"
-- Robin Faichney
Your statements are stupid and idiotic, and I said so.
That is not ad hominem.
> I would understand (!) if they'd rather play
> down this issue.
And thus you make clear your stupid idiotic ad hominem agenda,
since *that* statement of yours is quite ad hominem.
To sum up:
I say that what is true about you follows from what you say.
You say that what SR defenders say follows from what is true about them.
The latter is ad hominem, the former is not.
--
<J Q B>
Jim Balter
cgu...@symantec.com wrote:
>
> In <333756...@netcom.com>, Jim Balter <j...@netcom.com> writes:
> >cgu...@symantec.com wrote:
> >
> >> The argument is basically an argument against the validity of the
> >> Turing Test. It argues that even if a machine could pass the Turing
> >> Test, there is still no "understanding" anywhere in the machine;
> >> none of the mechanical components understands anything, therefore
> >> the appearance of "intelligence" is a mere illusion.
> >>
> >> This argument has a premise that for a whole to have a property,
> >> its parts must also have that property. This premise is false.
> >>
> >
> >Actually, I think the error is worse than that. Searle (and his
> >followers) have simply *mistaken* the mind of the Searle homunculus
> >to be the mind of the CR. This is quite clear from his ad reductio
> >premise, where he says (paraphrasing) "suppose my mind worked like a
> >computer". But in fact in the CR the workings of Searle's mind
> >are as unknown as ever; the experiment simply fails to test the premise.
> >As I said, this simply is not the product of a competent thinker.
> >A philosopher who is adept at manipulating words and phrases, yes,
> >but without the tools to check these manipulations for logical
> >accuracy. And it is a confusion between the two that gives his
> >followers some sense of confidence in their own views on the matter.
>
> Maybe I'm giving Searle way too much credit, but it seems hard to
> believe he could be guilty of such a horrible error as that.
Why? Outside of popular consumers of consciousness fables,
what sort of standing do you think Searle has? And on what basis?
> I had always interpreted the CR argument as:
>
> (1) Suppose we have a program P which is claimed to have understanding
> of a language L.
But no one makes such a claim. Programs aren't the sorts of things that
can have understanding.
> (2) P can be implemented as a CR.
Implementations can have understanding. But the CR is not a program.
> (3) This implementation consists of only (a) the homunculus, (b) the
> input/output symbols, and (c) the rules by which input strings
> are translated to output strings.
(c) is a program.
> (4) Neither (a) the homunculus, (b) the symbols themselves, nor (c)
> the translation rules, have any understanding of the language L.
Searle has no concern with the symbols or translation rules.
To him they are just "bits of paper". He never addresses whether they
have understanding, he just treats the notion as absurd. Such a
dismissal in the context of his "demonstration" is certainly
"a horrible error". Without ever even considering anything other than
the homunculus as a source of understanding, he never even gets to
any interactions among them.
> (5) By (3) there is nothing else there to have understanding of L.
> (6) Therefore, by (4) and (5), the CR does not understand L.
>
> Hence my above claim that one of his premises is that "for a whole
> to have a property, (I should have said: 'at least one of') its
> parts must also have that property".
Isn't (a)+(b)+(c) a part of the whole? We don't need to get into
whether the whole is greater than the sum of its parts, just consider
the whole, or the sum of its parts. That's the SR. Searle cannot
comprehend it.
> This false premise leads Searle to assert (5) and deny (or ignore?)
> the possibility of emergent behavior.
Behavior isn't Searle's concern; he admits that the CR *behaves*
as though it it understands Chinese. His argument is, plain and simple,
that the Searle homunculus obviously doesn't understand Chinese (we all
concur), and therefore the premise (suppose my mind worked like a
computer) is contradicted (we don't all concur). The argument is indeed
in horrible error, since it conflates Searle's mind with the CR. He has
left out all other parts. He quite explicitly ridicules the notion that
they could be relevant components of a mind.
> This seems to me to be a variant of what Turing called the "argument
> from consciousness" in the "Computing Machinery and Intelligence".
>
> Then again, it's been a while since I've read Searle, so maybe
> I'm getting my arguments crossed.
No, but you are giving him "way too much credit". I suggest that
you go back and read the article and his responses as published
in _The Mind's I_ to see just how pathetic they are.
--
<J Q B>
Anders N Weinstein
In article <33374D...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>
>People who think the Searle argument is valid are incapable of
>thinking clearly about the issues. This of course includes Searle.
>This is not as controversial a claim as it may seem, since the errors in
>Searle's "argument" are considered kid's stuff among logicians, despite
>its popular standing.
Didn't Aaron Sloman's response freely concede the validity of Searle's
argument?
I thought Sloman argued that "Strong AI" as implicitly conceived by the
argument was nearly absurd and not in fact what any reasonable AI
person actually believed. When reformulated in a more reasonable way,
"Strong AI" needs to rely on a much "thicker" causal sense of
"implementation" which will block the undesirable conclusion.
I bring this up to suggest that the alleged errors in the argument
might not fall into the category of "kid stuff". Unless you are happy
to count Aaron Sloman among those "incapable of thinking clearly
about the issues."
Sloman finds an error in the Searle argument, to be sure, but certainly
seems to locate it in a rather different place than other respondants
do. Wherever such disagreement among smart people is found, I would say
we are not dealing just with "kid stuff".
andrew babian
In article <5h6bjv$3...@usenet.srv.cis.pitt.edu>, ande...@pitt.edu (Anders N Weinstein) says:
>
>In article <E7Got...@cruzio.com>, andrew babian <aba...@cruzio.com> wrote:
>>And in any case, we only experience our
>>internal model of the world anyway.
>
>I think this is nonsense and the main source of all the problems. For
surely not all of them.
>example, when I look and see that my wastebasket is empty I can be said
if youre going to make claims about what "can be said" then it a question
of the use of language, and i was using a form to emphasize the role
of brain processes in experience. We experience both memories and current
inputs, but the the current inputs are always mediated (if nothing
else through a projected image on the retina, if not the brain/mental
representation) and you wish to say that experience is only an
interaction with the world? That certainly limits ai discussion.
>to experience the wastebasket in the world itself, not an "internal
>model" or proxy. If I hallucinate a wastebasket, then I am simply not
>experiencing any object at all, but only seeming to. But the immediate
>objects of veridical perceptual experiences are things in the world,
>not models or representations in the mind.
>
>There might be a sense in which our brains have models which *they*
>'use' or 'consult' in some fashion when we who are phenomenological
>subjects act and think. But we who can see the world directly don't
"directly"--is that like how the rock perceives? or is it just a
manner of speaking.
>always have to rely on proximal models or representations -- often we
>can just look and see instead.
its just that 'just look and see' explains nothing relevant to the
question, as far as i can tell, and could even be misleading.
r.j.fa...@stirling.ac.uk
In article <3338B4...@netcom.com>,
Jim Balter <j...@netcom.com> wrote:
>
> r.j.fa...@stirling.ac.uk wrote:
> >
> > I hope noone is breathlessly awaiting my reply to any of
> > the ad hominems.
>
> "Sorry, but to me it looks like UTTER STUPIDITY!!"
> -- Robin Faichney
>
> Your statements are stupid and idiotic, and I said so.
> That is not ad hominem.
>
> > I would understand (!) if they'd rather play
> > down this issue.
My statements were not directed at any particular individual,
but were intended to stimulate responses. I have to admit,
however, I should have expected that such tactics would tend
to stimulate such responses as yours.
So, I still await a substantive reply to my questions, which
I will restate here for your convenience:
(a) Are SR supporters in this forum hopelessly split as to
whether the CR is supposed to experience understanding, or is
there yet hope of a consensus emerging?
(b) If the CR is *not* claimed to experience understanding,
is understanding attributed to it on a purely behavioural
basis, or is there some other consideration?
--
Robin Faichney
r.j.fa...@stirling.ac.uk
http://www.stir.ac.uk/envsci/staff/rjf1/
-------------------==== Posted via Deja News ====-----------------------
Anders N Weinstein
In article <E7MnI...@cruzio.com>, andrew babian <aba...@cruzio.com> wrote:
>>example, when I look and see that my wastebasket is empty I can be said
>
>if youre going to make claims about what "can be said" then it a question
>of the use of language, and i was using a form to emphasize the role
By some kind of semantic ascent, one can say that seeing and its objects
are what the proper use of the ordinary language of "seeing" characterizes.
The claim "I only see my mental models" seems to me to be an example of
a confused misuse of language that already has a use in such statements as
"I see my wastebasket" -- it trades on a misleading analogy between seeing
my wastebasket -- which we all already understand -- and some concept
of seeing a model (whatever that is supposed to be).
>of brain processes in experience. We experience both memories and current
But you can characterize their causal role perfectly well without
supposing that they furnish weird objects that are themselves seen.
>inputs, but the the current inputs are always mediated (if nothing
Again, this also looks to me like a very misleading use of language, although
of course it can be given a meaning as you suggest. We remember things
and we have experiences, sure, and wouldn't do so unless there were
inputs. But I don't know what you are talking about if you say I experience
an input.
>else through a projected image on the retina, if not the brain/mental
>representation) and you wish to say that experience is only an
>interaction with the world? That certainly limits ai discussion.
Not at all, in my view, but the ai discussion is not mainly about what
we experience. For example, surely you understand the sense in which
one can't see a retinal image without a special instrument. So the
image as an image is sort of irrelevant, since it is not normally an object
of perception by any subject, although it could become one if you, e.g.
cut out an eyeball and study it's optical properties. JJ Gibson is
good on this, I think.
>>to experience the wastebasket in the world itself, not an "internal
>>model" or proxy. If I hallucinate a wastebasket, then I am simply not
>>experiencing any object at all, but only seeming to. But the immediate
>>objects of veridical perceptual experiences are things in the world,
>>not models or representations in the mind.
>>
>>There might be a sense in which our brains have models which *they*
>>'use' or 'consult' in some fashion when we who are phenomenological
>>subjects act and think. But we who can see the world directly don't
>
>"directly"--is that like how the rock perceives? or is it just a
>manner of speaking.
I would hardly have thought that rocks perceive. By direct I mean
something like: without depending for its justification on an
inferential step from more some basic data or "input".
>>always have to rely on proximal models or representations -- often we
>>can just look and see instead.
>
>its just that 'just look and see' explains nothing relevant to the
>question, as far as i can tell, and could even be misleading.
But you have to understand that it is not at all trying to explain, it
is just *describing* in everyday terms. Whatever goes on in the brain
when we see, it must somehow constitute or make up a case of or
otherwise fall under the ordinary concept of "looking and seeing", I
would say, since I think these ordinary concepts already have a use
which is relatively autonomous with respect to brain science.
I would suggest this is something like the way the everyday concept of
a solid table, call it solid-cs, is autonomous with respect to
materials science. I.e. we can find out well enough whether the table
is solid-cs by e.g. seeing if we can stand on it without its breaking.
This is true even if we don't know anything about how solidity-cs is
implemented.
Similarly, as masters of the English language, we already know what
seeing, recognizing, remembering, etc are (if you like, "seeing-cs",
etc) and this even if we don't know how these events are made possible
by neural processes. From that vantage point we can remind folks that
"You can only see your mental model" is an error, for example, one can
see one's wastebasket. At best it is an extremely misleading way of
talking about the neural mechanisms which are necessary to make our
seeing possible. But those mechanisms are not seen.
"Our disease is one of wanting to explain" -- Wittgenstein.
Burt Voorhees
Jim Balter <j...@netcom.com> writes:
>cgu...@symantec.com wrote:
>> >> Jim Balter wrote:
>> >But it doesn't demonstrate that, as has been readily shown
>> >by David Chalmers in _The Conscious Mind_, and many others before
>> >him. Rather, it demonstrates that John Searle is not a particularly
>> >competent thinker, nor are the many people who hold him to have
>> >shown something just because they happen to want to believe it.
>> >
>>
Actually, ad homonim attacks don't do much to advance the argument.
I have found that Searle is quite competent as a thinker, and far
more honest than some of the other characters around. Now if you
really want superficially clever deception, Dennett is your man.
Consciousness Explained is almost a text book in retorical trickery,
all designed to mask the fact that the entire thesis is based on
the fallacy of assuming the consequence. Which, come to think of
is, is just what the Systems Response people do, too.
>A typical example of what he uses in place of rigorous thinking is
>to be found in his review of Chalmers' book, published in the March 1997
>issue of The New York Review of Books as "Consciousness & the
>Philosophers".
>
And what is it that is used? What do you mean by rigorous thinking?
(One can't really demand mathematical proof from a philosopher...)
>Actually, I think the error is worse than that. Searle (and his
>followers) have simply *mistaken* the mind of the Searle homunculus
>to be the mind of the CR. This is quite clear from his ad reductio
>premise, where he says (paraphrasing) "suppose my mind worked like a
>computer". But in fact in the CR the workings of Searle's mind
>are as unknown as ever; the experiment simply fails to test the premise.
CR-o-philes on the other hand, need to explain exactly where in the
room the subjective understanding takes place, if not in Searle's
brain. No good just waving ones hands and saying "Gee, it's all
so complicated in there, there's got to be some understanding
somewhere." But that is all they have to fall back on.
bv
Neil Rickert
>I have found that Searle is quite competent as a thinker, and far
>more honest than some of the other characters around. Now if you
>really want superficially clever deception, Dennett is your man.
>Consciousness Explained is almost a text book in retorical trickery,
>all designed to mask the fact that the entire thesis is based on
>the fallacy of assuming the consequence.
Excuse me, but when boiled down all logical arguments are
tautological. That is to say, every logical proof is non-ampliative,
and in some sense can demonstrate a conclusion only by assuming that
conclusion. So presumably you are really saying that you agree with
Searle when he assumed what he was trying to prove, but you did not
agree with Dennett's choice of assumptions.
> Which, come to think of
>is, is just what the Systems Response people do, too.
And that just goes to show that you did not understand the Systems
Reply.
>CR-o-philes on the other hand, need to explain exactly where in the
>room the subjective understanding takes place, if not in Searle's
>brain.
No they don't. They only need to demonstrate that Searle's proof is
full of holes. If Searle could actually construct a physical
implementation of his Chinese Room, and if Searle could demonstrate
that his implementation actually had the intended behavior, then it
might be appropriate to as the SR people to explain where the
understanding is. But since Searle cannot construct his
implementation, his whole argument must rely on rhetorical trickery.
> No good just waving ones hands and saying "Gee, it's all
>so complicated in there, there's got to be some understanding
>somewhere." But that is all they have to fall back on.
The SR folks are not required to show that there would be
understanding in there. They need only show that Searle's disproof
fails.
Neil Rickert
In <5hcifk$6...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>No they don't. They only need to demonstrate that Searle's proof is
>>full of holes. If Searle could actually construct a physical
>>implementation of his Chinese Room, and if Searle could demonstrate
>>that his implementation actually had the intended behavior, then it
>>might be appropriate to as[k] the SR people to explain where the
>>understanding is. But since Searle cannot construct his
>>implementation, his whole argument must rely on rhetorical trickery.
>Don't understand this point at all -- why do you say *Searle* must
>construct an implementation?
I guess I don't understand the point either. For I have reread what
I said, and I cannot see where I said that Searle must construct an
implementation. In any case, if you want to read that into what I
wrote, you should remember that it was in the context of whether the
onus would fall on SR supporters to indicate where exactly the
understanding is situated.
> This seems to be a rather strong requirement to
>place on someone aiming to produce a philosophical argument -- either
>he actually implement something or shut up?
No, I did not suggest that at all. At most I was suggesting that
either Searle must produce a realization of his CR, or there is no
onus on his opponents to identify the locus of understanding.
Part of the problem, is that Searle assumed that computation is
arbitrarily realizable. But it is very doubtful that the required
computation could be realized in Searle's CR model, if for no other
reason that the computing power is grossly inadequate.
>I would say the problem is not exactly the "logic" of his argument --
>it cannot just be dismissed as a crude logical howler.
Perhaps not. But at least it is a subtle logical howler. Searle has
assumed that the understanding must be in the cpu, and that is not a
position supported by any AI advocates that I have heard of. The cpu
is commonly taken as mindless.
>is rather that we need a more precise and explicit understanding of the
>commitments involved in the crucial material terms: "computer",
>"program", "syntactic", "instantiation" and the like.
Or perhaps such explicitness is not possible.
>My own view is that Searle's argument might be perfectly valid given
>his definitions for these terms -- but they may well be the wrong ones
>to use in formulating the mostly inexplicit concepts and
>presuppositions of real AI.
I can accept that as a reasonable view.
>Of course, as I understand it, there is no important reason for anyone
>working in AI to want a *computer* or *program* to think or have
>intentionality. I always wonder: why couldn't AI just content itself
>with the aim of building a *robot* that could think or have
>intentionality? A robot might have properties no computer as such could
>ever have.
I agree with you there. So did those who proposed the Robot Reply
to Searle, a reply which Searle dismissed.
Jim Balter
Burt Voorhees wrote:
>
> Jim Balter <j...@netcom.com> writes:
>
> >cgu...@symantec.com wrote:
>
> >> >> Jim Balter wrote:
>
> >> >But it doesn't demonstrate that, as has been readily shown
> >> >by David Chalmers in _The Conscious Mind_, and many others before
> >> >him. Rather, it demonstrates that John Searle is not a particularly
> >> >competent thinker, nor are the many people who hold him to have
> >> >shown something just because they happen to want to believe it.
> >> >
> >>
> Actually, ad homonim attacks don't do much to advance the argument.
I didn't offer it to advance an argument. The argument is in work
I cited.
> I have found that Searle is quite competent as a thinker, and far
> more honest than some of the other characters around.
ad hominem
> Now if you
> really want superficially clever deception, Dennett is your man.
ad hominem
> Consciousness Explained is almost a text book in retorical trickery,
> all designed to mask the fact that the entire thesis is based on
> the fallacy of assuming the consequence. Which, come to think of
> is, is just what the Systems Response people do, too.
ad hominem
> CR-o-philes on the other hand, need to explain exactly where in the
> room the subjective understanding takes place, if not in Searle's
> brain. No good just waving ones hands and saying "Gee, it's all
> so complicated in there, there's got to be some understanding
> somewhere." But that is all they have to fall back on.
ad hominem
Try advancing an argument some day.
--
<J Q B>
Anders N Weinstein
In article <E7M6z...@gpu.utcc.utoronto.ca>,
Andrzej Pindor <pin...@gpu.utcc.utoronto.ca> wrote:
>In article <5h6bjv$3...@usenet.srv.cis.pitt.edu>,
>Anders N Weinstein <ande...@pitt.edu> wrote:
>>
>>ordinary psychological concepts, in particular on what is necessary for
>>determinate norms that constitute meanings. E.g what is required for you to
>>have a concept of tables that could succeed in referring to artifacts
>>in the common world?
>>
>Exactly. So what is required?
I would say it requires an organized system of human social practices in
which tables play a role. The pre-existing social system reproduces itself
via the individual when a new individual is initiated into the skills
of operating within it. So that the skill or know-how that is your
understanding of tables is internally or conceptually related
to this system of cultural norms and its history of cultivation in you.
Moreover, I think entering fully into such a system requires also the
ability to operate with the requisite language, for example the word
"table", in conjunction with other words in assertion, inferences and
action. It might also require that you in some fashion hold others
responsible to a common framework of norms, for example you don't let
other people call any old thing a table.
One could go on reflecting here, I just want to indicate a bit of what's
involved. No doubt it is somehow implemented in your brain, but we can
characterize the social roles and the process of normalizing modes of
behavior in abstraction from whether you have a brain.
But the distinctively cognitive element consists largely in being able
to *say* "that's a table" and form arguments on that basis, and to go on
to enjoy "inner" states whose content you can express with those token.
You might try Wilfrid Sellars, "Some Reflection on Language Games" in
his _Science, Perception, and Reality for one story about distinctively
linguistic behavioral norms. A fuller social behaviorist story might be
found in Heidegger, or Hubert Dreyfus' rather clearer exposition of
Heidegger in his book _Being-in-the-World_.
I should make clear that I am in my own way a kind of behaviorist about
the mind, insofar as I think understanding is mainly a matter of
skills, capacities or know-how, and that rational, self-conscious
understanding is conditioned by skills with public linguistic symbols,
including "I". On the other hand I don't think these behaviors can
really be adequately described in mechanistic terms.
Challenging someone to a duel is a behavior, but its significance is,
as the saying goes, socially constituted, by its role in a system of
cultural norms. An intellectualist view has it that these norms, or the
rationality (or its lack) we can find in the particular act, must
somehow be represented explicitly in the individual mind and so, on
materialist principles, in your brain. To me, that's as may be, it
might be true, but it need not, it might also be false, since what
matters at the autonomous sociological level is the know-how and the
capacity for this sort of conduct itself, not how it happens to be
implemented in the brain.
And I think of persons as self-conscious conceptual thinkers as entites
that are properly explained as constituted at the relatively autonomous
social level. If you think of words and more generically symbol systems
like blueprints and scores on the model of tools and other cultural
artifacts you should find this obvious; now just think of silent
thought as a kind of secondary, inner operating with those same
cultural tools.
>>Well it seems to me as soon as you separate off "experiences" and treat them
>>as *inputs* to "the mind", you are already well on your way down the
>
>So if experiences are not *input* to the mind, where do they come from? are
I guess I could say they come from the environment.
>they generated internally? If not fully internally but in some correlation
No, that would be hallucinating.
>with inputs, what is the remaining constutent of the experiences? Where does
>it come from?
You could I guess say the experience itself is the input. For example,
the experience I referred to as my becoming conscious of my wastebasket
as empty is itself a possible input to further reasoning I might do.
Of course, to forestall your probable objection, my having this
experience certainly requires on the presence of light and the
stimulation of the retina and the operation of the visual system. It
might itself be a state of my brain as far as I am concerned. Roughly,
I take it that all these neural processes have two aspects, just as the
process of writing out a check has two aspects, a material one and a
socially significant one. Under one aspect, all these neural operations
are physical events in accordance with the laws of physics. But under
another aspect, some of them might also constitute, for example, my
becoming directly or non-inferentially aware of my wastebasket as
empty. So that all these merely causal prerequisites are not themselves
*inferential* bases for my subsequent beliefs.
In particular, I need not say that under the second aspect the
epistemic event is caused by some more basic "input" to my awareness --
what is input to my awareness is just the information about the distal
objects itself.
Another way to put might be to say is that in the obtaining of
knowledge in perception we have "fact" causation rather than
event-event causation. That is, in the case of veridical perception,
you can say I see that p because of the manifest fact that p. In this
case the cause is not really something conceptually distinct from the
effect. At this level you do not really have two independent but
related events, a proximal representation and a distal reality. Rather,
there is only one event with both subjective and objective poles -- it
is subjective, since it is *my* becoming aware of the world, and it is
objective, in that the information or intentional content of the
experience is a fact *about* objects in the distal world, say the
wastebasket.
>>Cartesian garden path. For if experiences are just external inputs, then
>>the thinking subject ("the mind") is only able to test its hypotheses and
>>beliefs by reference to these inputs -- it is trapped behind a veil of
>>ideas and can never really see what is beyond them.
>>
>The last part of the above sentence does not square with the begining part -
>where did 'ideas' come from? The mind is indeed trapped, but behind (veil
The phrase "veil of ideas" is traditional in philosophy to describe the
hopeless condition arrived at by Berkeley and Hume. These empiricists
tended to think of ideas in somewhat the manner of proximal images before
the gaze of a conscious subject (although the subject tends, on this view,
to rather disappear).
> The mind is indeed trapped, but behind (veil
>or not) inputs. Its only contact with the world is through these inputs. If
I disagree. It's contact with the world can be through information carried
in patterns of (what you call) "inputs". Information is in a sense an
abstraction. For example, when I detect the information that this
wastebasket (here I'm pointing in my office to W1) is empty, what I detect is
in a way an abstraction -- the proposition that this wastebasket [W1]
is empty. But it is something that I think might also be taken to be
immanent in the patterned radiation reaching my retina in some way.
In this way I do not in any way *infer* this content from
representations of patterns on my retina -- I certainly have no
conscious representations of patterns on my retina, so I could not
perform any such inference. Rather I am now able to *extract* the
information the information about the world that is already (I know not
how) coded there.
>you disagree, please explain what are the other ways it makes contact with
>the external world. Have you heard about virtual reality? Are you not aware
I have heard of virtual reality. I gather most current systems are
rather low quality and would not fool anyone. To really get a total
body simulation of sufficient quality would take an awful lot of
computational power. I believe as a matter of engineering fact that it
is very hard to reproduce all the relevant complexity of our
interactions with a real environment with anything short of the real
environment itself.
But in any case I am talking about perception of the real reality, for
example my office. If you are in a virtual reality then you are not
really perceiving anything, you only seem to be. I don't recall
saying that one could always tell the difference.
>that the mind is unable to tell where do the inputs come from - from a real
Again, a conscious subject just is not in the position of being presented
with "inputs" and having to wonder where they come from. As Sartre put it,
"All consciousness is consciousness of an object that is not itself" (or
something like that)
That is to say, (perceptual) consciousness is exhausted by such things
as my becoming conscious of facts about my office furniture -- or, if
it is deceptive, merely seeming to be conscious of putative facts about
putative furniture. But there are no ideas or images or any other kind of
"input" to be found in normal perceptual consciousness.
>world or from a computer generated world? You seem to refuse to accept
>clear empirical evidence. Or do you believe in telepathy, and such which is
>mediated by non-physical medium? Even most of those who believe in telepathy
No. In any case I don't think telepathy would be perception, it would be
something like the implantation of reliable hunches from one knows not
where, a bit like blindsight.
>would agree that it provides *input* to the mind, if by some unknown medium.
>So how does mind "see" beyond inputs (ideas?)?
We don't need the concept of inputs to conscious experience, unless you
think of facts or information or propositions or similar abstractions
as the inputs. So there really is no problem here.
>>For example, my noticing that my wastebasket is empty is an "experience" of
>>mine. But to characterize it thus is to relate me to objects and a state
>>of affairs in the world, not to a proximal representation or "input".
>>
>So is you noticing that your wastebasket is empty an "an experience" of yours,
>because it is characterised as such? If it was not characterised as such
>by a philospoher, it would not be that? If there were no philosophers, then
To characterize it thus one need not be a philosopher. Someone who says
"Weinstein just noticed that his wastebasket was empty [and, say,
wondered how that happened because he had just filled it]" has characterized
a visual experience in a way that relates it to things in the world.
It is hard to understand how perception could be an avenue to knowledge
if its content was never related to the world.
>there would not be such problems, would there? If a philosopher has not
In a sense, since common sense language does not find any problem. I am
suggesting some philosophers can help others back to common-sense
idioms by recognizing that the characterization of perceptual
experience is world-involving. By the way, some philosophers have
labelled this view "perceptual externalism"; there is a whole section
of references concerning this topic in the bibliography on David
Chalmers' web page.
I do not myself think common sense uses your notion of "inputs" to the
mind. It does employ the conception of seeing as non-inferential
detection of information concerning the environment.
>just inputs to the mind? You seem to suggest that this other constituent of
>experiences (besides input) is provided by philosophers relating objects
But I do not think in terms of two constituents, one of which is
"input" and another one mysteriously added by "philosophers". I think
only in terms of one component, information or intentional content.
This is employed in characterizations such as "sees" or "notices" or
"detects" that p.
I can ask someone -- can you see this wastebasket, this color, this
book, and the like. What if I also ask them -- can you see or describe
the sensory "input" from which you inferred to the existence of the
wastebasket? Can you tell me what steps you took in this inference,
why you think it is a good one? In general they need not be able to
answer in any way for they did not and could not perform any such
inference -- and remember, I am only talking about the level of
consciousness, I have no interest in what might go on outside your
awareness in your parts, since these could not be inferences you
perform.
Possibly they can be provoked to construct an answer in terms of
features they do not normally "see" or notice at all, what Gibson
called the "visual field". But he argued well that that queer
experience was an epiphenomenon that had little to do with seeing.
You can check out John Searle's book _Intentionality_
for a theory that is in some ways along the right lines -- it
characterizes conscious mental states solely in terms of an abstraction
like intentional content (while conceding these states are realized in
neural matter). On the whole I think Searle's view fails because of its
commitment to internalism about intentional content, but his general
conception of intentionality has many good points.
>>>our ability to fantasize). And in any case, we only experience our
>>>internal model of the world anyway. None of this really relates to the
>>
>>I think this is nonsense and the main source of all the problems. For
>>example, when I look and see that my wastebasket is empty I can be said
> ^^^^^^^^^^^^^
>You can, so what? Paper (screen) is patient, just saying something does not
>make it true, as you keep thinking.
All right, rephrase as "I can *rightly* be said...".
>>to experience the wastebasket in the world itself, not an "internal
>>model" or proxy. If I hallucinate a wastebasket, then I am simply not
>>experiencing any object at all, but only seeming to. But the immediate
>
>Again, just saying so does not make it true. Please explain the difference
>between seeing a "real" wastebasket and 'hallucinating' one, due to being
>in virtual reality environment, feeding corresponding signals directly
>to your optic nerves. Difference to the mind, of course.
Well sure, if you are really seeing a wastebasket then there really is
a thing you see, for example, and your experience is in some way
responsive to *it*, the real wastebasket which anyone else can see and
touch. In a virtual reality you are not in touch with any real
wastebasket, for example, your experience is not caused by a real
wastebasket but induced by a computer from something other than a
wastebasket.
There are problem cases about what has been called "prosthetic vision"
-- what if you are in a virtual reality being fed information derived
from a real wastebasket? But in some sense I take the difference to be
primitive, I am not offering a causal theory of perception. But I don't
really see it as problematic. There might be no difference "to the mind".
>>objects of veridical perceptual experiences are things in the world,
>>not models or representations in the mind.
>>
>And how does mind know whether it is dealing with "things in the world" or
>with artificially generated inputs?
This is not usually a significant question since there is no reason to
suspect one has been placed without one's knowledge in a virtual
reality game, which are not presently good enough to fool one anyway.
But really, I never said a person could necessarily know the difference
in all cases. There's still a difference that is crucial to the way
we characterize a person's epistemic standing.
>When we "look and see" we provide our minds/brains with physical inputs.
Possibly. But we ourselves never see those inputs. When we look/see we
often manage to provide *ourselves* with information about the
environment. We might at about the same time provide our brains with
its inputs, but the inputs to the brain are certainly not inputs to our
conscious awareness.
Note I take it to be very important here to think phenomenologically,
that is to concentrate on how things are from the subjective point of
view of the conscious subject whose mode of awareness seems to provide
a vista opening out directly onto the distal world -- and which, in
veridical cases, really *does* provide such a vista.
I don't really give a damn about the details that happen in the retina
or the brain, I am talking about the metaphysics of the
phenomenological level or about the nature of the subjective experience
of a conscious subject; I say, get the characterization of *that* level
right, as, e.g., Sartre more nearly does, and then all the traditional
philosophical problems of empiricism concerning the possibility of
knowledge or intentionality answering to something outside the mind will
just vanish.
That is why I think we don't need more empirical science of the wholly
unconscious (and therefore non-mental) goings on in our visual systems
to help us explain how this sort of intentionality is *conceptually*
possible. We do need that science to help explain how it is *causally*
possible that we should have the perceptual capacities we manifestly
do, and such sub-cognitive science regarding its implemention level is
a perfectly good thing in its own right. The latter level is not very
philosophically exciting to me since it seems to be one level down from
the level I find most interesting, the phenomenological or, if you
like, the intentional systems level. And there does seem to me to be an
ever present danger of confusing the two levels, of treating that sort
of cognitive science as the real scientific account of our thinking being.
As to "mind" in a way I would say there is no such thing.
Anders N Weinstein
In article <5hcdo8$9...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>No they don't. They only need to demonstrate that Searle's proof is
>full of holes. If Searle could actually construct a physical
>implementation of his Chinese Room, and if Searle could demonstrate
>that his implementation actually had the intended behavior, then it
>might be appropriate to as[k] the SR people to explain where the
>understanding is. But since Searle cannot construct his
>implementation, his whole argument must rely on rhetorical trickery.
Don't understand this point at all -- why do you say *Searle* must
construct an implementation? He is not himself doing AI or computer
science, after all. This seems to be a rather strong requirement to
place on someone aiming to produce a philosophical argument -- either
he actually implement something or shut up? That would rule out the
possibility of any philosophical debate at all.
Searle is trying to critique the assumptions of those other who believe
in the logical possibility of "Strong AI" as he defines it. All he
needs for his argument is a certain conception of what it is to be a
computer.
In fact he takes it that he is developing logical consequences of a
certain definition of what it is to be a computer (or, perhaps, a
computer program instantiation), viz., to be an agency that manipulates
symbols in accordance with formal or non-semantic rules.
I would say the problem is not exactly the "logic" of his argument --
it cannot just be dismissed as a crude logical howler. The whole issue
is rather that we need a more precise and explicit understanding of the
commitments involved in the crucial material terms: "computer",
"program", "syntactic", "instantiation" and the like.
Both sides in the debate over the Chinese Room can equally be charged
with throwing these terms around at a merely intuitive level, when we
all could benefit from some serious work on getting clearer about
what they mean.
My own view is that Searle's argument might be perfectly valid given
his definitions for these terms -- but they may well be the wrong ones
to use in formulating the mostly inexplicit concepts and
presuppositions of real AI.
Of course, as I understand it, there is no important reason for anyone
Chris Hooley
On 26 Mar 1997 18:07:36 -0600, ric...@cs.niu.edu (Neil Rickert)
wrote:
> (Burt Voorhees) writes:
>
>>I have found that Searle is quite competent as a thinker, and far
>>more honest than some of the other characters around. Now if you
>>really want superficially clever deception, Dennett is your man.
>>Consciousness Explained is almost a text book in retorical trickery,
>>all designed to mask the fact that the entire thesis is based on
>>the fallacy of assuming the consequence.
NR
>Excuse me, but when boiled down all logical arguments are
>tautological. That is to say, every logical proof is non-ampliative,
>and in some sense can demonstrate a conclusion only by assuming that
>conclusion.
Snipped . . .
Could you refer me to the proof on this?
Or, better yet, explain it? :-)
Are there links to Goedel, Varela, G. Spencer Brown or others?
Respectfully,
Chris
andrew babian
In article <5hcrrb$a...@ux.cs.niu.edu>, ric...@cs.niu.edu (Neil Rickert) says:
>Perhaps not. But at least it is a subtle logical howler. Searle has
>assumed that the understanding must be in the cpu, and that is not a
>position supported by any AI advocates that I have heard of. The cpu
>is commonly taken as mindless.
what exactly is the cpu (or searle in the CR) supposed to correspond to?
That's using a pretty primitive view of the mind to say there is some
discrete object that this analogy might apply to. A more sophisticated
understanding of mind would hold that this "cpu" is just an abstraction
for many different kinds of brain processes and structures. It may be
that CR as a model is so flawed as to even defy sensible analysis, and it
certainly isn't a basis for anything i would do. But i would hold out the
hope that if we learned enough from development on some other lines
we could actually consider a CR model, even though we knew better.
Andrzej Pindor
In article <3337B0...@stirling.ac.uk.nospam>,
Robin Faichney <r.j.fa...@stirling.ac.uk.nospam> wrote:
>Andrzej Pindor wrote:
>>
>
>(RF: Sorry for causing that digression)
>
>>Your statement "subjective understanding _obviously_
>>(my underlining) is not present" (or something to this effect) flouts
>>principles of scientific practice even more seriously since it is logically
>>incoherent (which I have pointed out and you have chosen to ignore).
>
>I'm not particularly wedded to that statement. In fact, I think I
>had a "(?)" in the original. But your criticism is off-target anyway,
>because I do not restrict myself to scientific methodology. In this
>context, I see myself as a philosopher rather than a scientist. But
Scientific methodology was established as a set of practices which minimize
chances of a researcher (scientist or not) being tricked by his/her mind
(wishful thinking, etc). Are you saying that you (and philosophers in general)
are not at all concerned about it? Of course, if philosophers are
after producing stories without any regard if these stories relate to the
real world or are products of their mind then what you say is understandable.
However, I do not think moste people interested in AI would concern themselves
with fantasies.
It seems clear, btw., that Searle has no appreciation that mind plays tricks
on us (philosophers too) and that if we want to have any trust in what we
seem to find out about the world we have to do our utmost to avoid dangers of
such delusions.
>rather than pursue this any further, I'll happily drop that particular
>assertion. Seems to me the discussion moved into more interesting
>areas a few days ago.
>
>> Yes, you are right that the assertion that the system as a whole _experiences
>> understanding_ is (equally) flawed. Now, who makes this assertion?
>> As far as I remeber, SR is that the system as a whole _understands_.
>
>That's what I thought, but Daryl McCullough said otherwise, and noone
>(else) in the several followups since has contradicted him, some
>seeming in fact at least by implication to agree with him. It's a
>pity you weren't "listening" when I explicitly asked, in my response
>to the post in which he said that, whether anyone disagreed with him.
>(But I am familiar with the unreliable nature of usenet.)
>
Has Daryl McCullough said that in his view CR "experiences understanding"?
I'd be very surprised. Can you provide a quote? "Experiencing understandingg",
being a subjective thing, cannot be ascertained by anyone outside, can it?
>--
>Robin Faichney
Burt Voorhees
>In <5hc7s6$9...@aurora.cs.athabascau.ca> bu...@cs.athabascau.ca (Burt Voorhees)
writes:
>>I have found that Searle is quite competent as a thinker, and far
>>more honest than some of the other characters around. Now if you
>>really want superficially clever deception, Dennett is your man.
>>Consciousness Explained is almost a text book in retorical trickery,
>>all designed to mask the fact that the entire thesis is based on
>>the fallacy of assuming the consequence.
>Excuse me, but when boiled down all logical arguments are
>tautological. That is to say, every logical proof is non-ampliative,
>and in some sense can demonstrate a conclusion only by assuming that
>conclusion. So presumably you are really saying that you agree with
>Searle when he assumed what he was trying to prove, but you did not
>agree with Dennett's choice of assumptions.
Sorry, not true. If I say: All men are mortal; Socrates is a man;
therefore Socrates is mortal this is not assuming the consequence,
it is asserting that a certain individual belongs to a certain class
and therefore must have the defining qualities of that class. What
I'm saying is that whether or not I agree with Searle (I think it is
a pretty tight argument, but not absolute) I find from reading his
other works that although he sometimes makes what I consider to be
mistakes, they are honest mistakes. And I find that he is honestly
defending his point of view, and seeking to discover what is true.
Dennett, on the other hand is guilty in my opinion of tremendous
intellectual dishonesty. I could give you a list of the various
tricks he uses to try to con people into buying his point of view,
including Slippery Slope, False Appeal to Authority, Misleading
Language, Sleight of Hand, Smoke and Mirrors, Muddying the Water,
Argument from Ignorance, and or course, Assuming the Consequence
(check out pp. 214,215 in his book: he assumes the consequence,
uses misleading language to obscure this, and then appeals to
something which he has claimed cannot be accepted as evidence).
>> Which, come to think of
>>is, is just what the Systems Response people do, too.
>And that just goes to show that you did not understand the Systems
>Reply.
Can you substantiate that?
>>CR-o-philes on the other hand, need to explain exactly where in the
>>room the subjective understanding takes place, if not in Searle's
>>brain.
>No they don't. They only need to demonstrate that Searle's proof is
>full of holes. If Searle could actually construct a physical
>implementation of his Chinese Room, and if Searle could demonstrate
>that his implementation actually had the intended behavior, then it
>might be appropriate to as the SR people to explain where the
>understanding is. But since Searle cannot construct his
>implementation, his whole argument must rely on rhetorical trickery.
Sorry again, Searle is presenting a philosophical thought experiment,
not an engineering claim. You are just Begging the Question.
>> No good just waving ones hands and saying "Gee, it's all
>>so complicated in there, there's got to be some understanding
>>somewhere." But that is all they have to fall back on.
>The SR folks are not required to show that there would be
>understanding in there. They need only show that Searle's disproof
>fails.
You do not understand the nature of Searle's argument. It poses
precisely the challenge of showing where in the room there is
understanding, or as another poster has said, falling back on
the assertion that the only legitimate criterion for assuming
that there is understanding is behavioral. You need to do your
philosophical homework.
bv
Jim Balter
Anders N Weinstein wrote:
>
> In article <33374D...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
> >
> >People who think the Searle argument is valid are incapable of
> >thinking clearly about the issues. This of course includes Searle.
> >This is not as controversial a claim as it may seem, since the errors in
> >Searle's "argument" are considered kid's stuff among logicians, despite
> >its popular standing.
>
> Didn't Aaron Sloman's response freely concede the validity of Searle's
> argument?
No, you bloody idiot. We've been through this. Aaron argued
that Searle's *conclusion*, that "Strong AI is false", was true,
if "Strong AI" is Aaron's "Strong strong AI". He never said that
Searle's argument was *valid*.
> I thought Sloman argued that "Strong AI" as implicitly conceived by the
> argument was nearly absurd and not in fact what any reasonable AI
> person actually believed.
Which has nothing to do with whether Searle's argument is valid,
you stupid fucking bloody idiot.
Years and years and years of energy wasted on this stupid
newgroups by people who are absolute NINCOMPOOPS, passing for
philosphers and such.
--
<J Q B>
Anders N Weinstein
In article <5hfp5u$g...@aurora.cs.athabascau.ca>,
Burt Voorhees <bu...@cs.athabascau.ca> wrote:
>You do not understand the nature of Searle's argument. It poses
>precisely the challenge of showing where in the room there is
Why must understanding be located anywhere? Can't it be a
non-localizable feature, like, say, a personality trait? (where is
one's irritability, where is one temper, etc?)
I guess the understanding is where the person is, but there is no
reason it needs to be localized any more precisely than that. I don't
think Searle's argument really depends on this demand.
Anders N Weinstein
In article <333B93...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>Anders N Weinstein wrote:
>>
>> Didn't Aaron Sloman's response freely concede the validity of Searle's
>> argument?
>
>No, you bloody idiot. We've been through this. Aaron argued
>that Searle's *conclusion*, that "Strong AI is false", was true,
>if "Strong AI" is Aaron's "Strong strong AI". He never said that
>Searle's argument was *valid*.
I was recalling the point in his 1985 paper where Sloman wrote of
Searle's reasoning:
His logic is unassailable, if his premisses are true, in particular the
premiss that nothing understands Chinese in this situation.
Note btw that "unassailable logic" in the sense of validity of
deductive reasoning does not depend at all on whether the premises are
in fact true. A deductively valid argument can still have a false conclusion.
Sloman's diagnosis of the problem rests on pointing out an ambiguity in
the interpretation of the crucial thesis of "Strong AI". Sloman
*speculated* that Searle may have slid unwittingly between two
interpretations, but does not go so far as to charge Searle with
committing a fallacy of equivocation anywhere in his argument.
Rather Sloman says explicitly that Searle's argument succeeds at
refuting the strong interpretation Searle was apparently using ("Strong
strong AI"). But Sloman suggested that this is not really very
exciting, since the thesis on that interpretation is too strong to be
plausible. In fact, he claims no one has ever published a defense of
it and that on suitable reflection it can be seen to be absurd.
Sloman suggests we ought instead to consider different, more plausible,
interpretations of the thesis (varieties of "Weak Strong AI") than the
one Searle used. Against these, Searle's argument will fail, since a
human being hand-simulating a program inside the Chinese room will
typically *not* constitute the right sort of program instantiation.
This is because the program will not really be "in control" in such setup.
Note that if Sloman's diagnosis is the right one, then (as he notes)
most of Searle's opponents, including I presume every one of the
original BBS commentators, were also guilty of failing to identify the
crucial ambiguity.
Note also that Sloman himself did not push the Systems Reply at all, he
did *not* make any claim that the room as a whole would understand in
the hypothetical example. Apart from expressing some qualms about
Searle's reliance on imagined introspection, Sloman did not dispute the
possibility that there would be no understanding in such an unreliable
instantiation as Searle envisages.
Sloman's view seems to entail, for example, that if you were to replace
an intelligent control system implemented in (reliable) computing
machinery piece by piece by one requiring by an (unreliable) human
hand-simulator, it is reasonable to say that the understanding could
disappear.
>> I thought Sloman argued that "Strong AI" as implicitly conceived by the
>> argument was nearly absurd and not in fact what any reasonable AI
>> person actually believed.
>
>Which has nothing to do with whether Searle's argument is valid,
My point exactly: Sloman's response did not take issue with the
validity of Searle's argumentation. He disputed whether Searle's
interpretation of the key thesis of Strong AI was in fact a reasonable
one to consider. To me, this is not the same kind of objection as
pointing out an error in logic.
Maybe this is a pointless terminological quibble. But to my mind, the
articulation of a subtle ambiguity in the interpretation of a key
thesis, such as Aaron Sloman presented, is not at all the same as the
diagnosis of a simple error in deductive logic.
Sven-O. Berg
ric...@cs.niu.edu (Neil Rickert) wrote in article
<5hcrrb$a...@ux.cs.niu.edu>:
>In <5hcifk$6...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>Of course, as I understand it, there is no important reason for anyone
>>working in AI to want a *computer* or *program* to think or have
>>intentionality. I always wonder: why couldn't AI just content itself
>>with the aim of building a *robot* that could think or have
>>intentionality? A robot might have properties no computer as such could
>>ever have.
>I agree with you there. So did those who proposed the Robot Reply
>to Searle, a reply which Searle dismissed.
AFAIC, this would be a far more interesting thing to discuss.
I think you may liken computer reading inputs from the keyboard to a
blind and deaf person. In a way, a blind and deaf person elaborating
on colors and music doesn't know what on earth he's talking about.
Some might object to even this, but at least to me, it seems
plausible.
OTOH, what if a machine is equipped with a video camera, being
able to map raw pixel input to natural language, and interpret them
sensibly? Isn't this what the Robot Reply is about? What did Searle
say about this objection, and what do the others here think?
Curious,
Sven
--
write to soberg (at) internet (dot) de if you wish to reply by email
Sven-O. Berg
cho...@idnsi.net (Chris Hooley) wrote in article
<333ab0f4...@news.gte.net>:
>On 26 Mar 1997 18:07:36 -0600, ric...@cs.niu.edu (Neil Rickert)
>wrote:
>>Excuse me, but when boiled down all logical arguments are
>>tautological. That is to say, every logical proof is non-ampliative,
>>and in some sense can demonstrate a conclusion only by assuming that
>>conclusion.
>Snipped . . .
>Could you refer me to the proof on this?
>Or, better yet, explain it? :-)
Neil is right here, due to the fact that this is how a ``valid
conclusion'' is defined:
By definition, you'd contradict yourself when asserting the truth of
the assumptions AND deny the truth of the conclusion, that is: you
end up with asserting p & ~p. Now if neither the assumptions nor the
conclusion are contradictory themselves, then either p or ~p must
have been part of *both* premises and conclusion simultaneously.
I guess you might find this definition in every basic textbook on
formal logic; I remember dimly that we had an excellent one named
``Beginning Logic'' by J.E. Lemmon, but this one is from 1967, so
I don't know whether it's still available.
Burt Voorhees
You're right about that. An imprecise use of language on my part,
attributing something like spatial location to "understanding".
What I whould have said was that the CR poses the question of
how there could be any understanding at all, if not located in
Searle inside the room, and if one claims that there is; i.e.,
that the room as a whole does understand then is this based on
only a third person attribution, or is the much stronger assertion
being made that the room itself has first person experiences.
Searle would deny the latter, and claim that what is demonstrated
by his thought experiment is that the third person attribution of
understanding is insufficient to guarantee its actual first person
presence. That is, that first person experience must be taken as
a distinct ontological catagory rather than something which can be
reduced to things availiable to external observations.
bv
Neil Rickert
In <5hgsef$t4t$1...@news.nacamar.de> i...@fed.up.with.spam (Sven-O. Berg) writes:
>ric...@cs.niu.edu (Neil Rickert) wrote in article <5hcrrb$a...@ux.cs.niu.edu>:
>>I agree with you there. So did those who proposed the Robot Reply
>>to Searle, a reply which Searle dismissed.
>AFAIC, this would be a far more interesting thing to discuss.
>I think you may liken computer reading inputs from the keyboard to a
>blind and deaf person.
I don't think this is quite the same thing. The blind and deaf
person still has a wide variety of other inputs from reality.
Presumably spatial perception is possible due to sensory signals
throughout the body.
> In a way, a blind and deaf person elaborating
>on colors and music doesn't know what on earth he's talking about.
>Some might object to even this, but at least to me, it seems
>plausible.
I think I would agree. A blind person could talk about the physics
of light and wavelength, and the physics of soundwaves. But color
and music are not just the physics.
>OTOH, what if a machine is equipped with a video camera, being
>able to map raw pixel input to natural language, and interpret them
>sensibly? Isn't this what the Robot Reply is about?
Yes, that is the point of the Robot reply.
> What did Searle
>say about this objection, and what do the others here think?
Searle dismissed the RR out of hand, with little comment. It
apparently seemed just obvious to Searle that the Robot would not be
in any better position than the Chinese Room. There are others in AI
who object to any suggestion that having a perceptual system might be
a prerequisite for intelligence. Personally, I think that perception
is critically important.
Jim Balter
Anders N Weinstein wrote:
>
> In article <333B93...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
> >Anders N Weinstein wrote:
> >>
> >> Didn't Aaron Sloman's response freely concede the validity of Searle's
> >> argument?
> >
> >No, you bloody idiot. We've been through this. Aaron argued
> >that Searle's *conclusion*, that "Strong AI is false", was true,
> >if "Strong AI" is Aaron's "Strong strong AI". He never said that
> >Searle's argument was *valid*.
>
> I was recalling the point in his 1985 paper where Sloman wrote of
> Searle's reasoning:
>
> His logic is unassailable, if his premisses are true, in particular the
> premiss that nothing understands Chinese in this situation.
All Sloman has said here is that circular arguments are valid.
So stop trying to misuse him as an authority.
--
<J Q B>
Jim Balter
Jim Balter wrote:
>
> Anders N Weinstein wrote:
> >
> > In article <333B93...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
> > >Anders N Weinstein wrote:
> > >>
> > >> Didn't Aaron Sloman's response freely concede the validity of Searle's
> > >> argument?
> > >
> > >No, you bloody idiot. We've been through this. Aaron argued
> > >that Searle's *conclusion*, that "Strong AI is false", was true,
> > >if "Strong AI" is Aaron's "Strong strong AI". He never said that
> > >Searle's argument was *valid*.
> >
> > I was recalling the point in his 1985 paper where Sloman wrote of
> > Searle's reasoning:
> >
> > His logic is unassailable, if his premisses are true, in particular the
> > premiss that nothing understands Chinese in this situation.
>
> All Sloman has said here is that circular arguments are valid.
> So stop trying to misuse him as an authority.
To be more explicit, here is an extract from Sloman's paper:
Searle claims that provided Pc is expressed in a programming language
he understands, he can take the place of the computer and produce an
instantiation of Pc which will fool others outside the room into
thinking that something inside the room understands Chinese (if Pc is
a good program). But, since he doesn't understand Chinese, and (he
claims) the total system consisting of him executing the program
doesn't understand Chinese, it follows that the alleged mental state
does not exist. So C(x) (that is B(x) instantiating Pc) is not
sufficient for M(x) in this case.
His logic is unassailable, if his premisses are true, in particular
the premiss that nothing understands Chinese in this situation.
The "unassailability" here is the circularity of the argument. The "(he
claims)" is a matter of taking (not SR) *as a premise*, so of course
*given that premise* the SR does not rebut Searle. Weinstein has failed
to appreciate Aaron's wry expository style. When that premise might or
might not be justifiable is something that Sloman examines throughout
his paper.
Sloman claims that Searle has indeed refuted something, and he points
out just what that is:
So, if the strong AI thesis is defined as claiming that ANY
instantiation of the right program will suffice for the production of
understanding, then that thesis implies that the process of
understanding Chinese exists undetected in a thunder cloud, along with
many other mental processes.
This shows that the extreme strong AI thesis is absurd, and I do not
believe that any of Searle's opponents would have bothered to reply if
they had appreciated that he was attacking this thesis, though I have
met one or two intelligent people who have been tempted, though only
for a short time, to try to defend it, along with the consequence that
thunder-clouds have thoughts.
Since the claim that thunder clouds have thoughts is not the claim that
most people disagreeing with Searle are trying to defend, Searle's
argument, and Sloman's explication of it, is no challenge to them, which
is just the point that Sloman is making. David Chalmers, in his paper
"A Computational Foundation for the Study of Cognition", at
http://ling.ucsc.edu/~chalmers/papers/computation.html argues that the
issue of thunder clouds understanding Chinese or walls instantiating
Wordstar programs does not apply to an appropriate notion of
computation, based upon the idea of implementation. The views of
Chalmers and Sloman are quite consistent in this regard, and Sloman has
discussed the implementation issue fairly recently in this forum.
For an indication of how thoroughly Sloman rejects Searle's argument as
applied to non-absurd forms of Strong AI, here's another extract:
One form of reply would assert that there would be a process of
understanding, despite Searle's inability to introspect it. (He takes
it as obvious that he would not experience understanding. Actually,
since the whole experiment is hypothetical we don't even know for sure
that he wouldn't, as a result of achieving fluency with the program,
come to feel he understood Chinese as well as English. But let that
pass.) Since many do not find behaviourist criteria adequate,
something more than behavioural evidence would have to be invoked. I
have attempted this in [Sloman 1985] which sketches some of the design
requirements for a machine to understand symbols in the way we do, and
then attempts to show that a substantial subset is already satisfied
by simple computers. The paper also argues against attempting to use
ordinary concepts, like 'understanding', to draw global distinctions
in the space of possible behaving systems.
Here Sloman says that (reordering without changing meaning) "simple
computers" "already satisf[y]" "a substantial subset" "of the design
requirements for a machine to understand symbols in the way we do",
a very far cry from viewing Searle's argument as valid.
Some other words from Sloman's paper in re Searle:
"John Searle's attack on the Strong AI thesis, and the published
replies, are all based on a failure to distinguish two interpretations
of that thesis"
"Searle attacks strong strong AI, while most of his opponents
defend weak strong AI."
"I believe that although his main conclusions are quite wrong, he raises
issues which are potentially of considerable importance for AI, and that
his arguments should therefore not simply be ignored by those who
disagree." [Note that people like Chalmers have addressed these issues
head on. Now it is 1997, Searle is still quite wrong, and it is time to
move on.]
"I do not believe that many of those who thought they were disagreeing
with Searle would in fact wish to defend the strongest strong AI
thesis. However, close examination suggests that that is the one Searle
thought he was attacking, though he, like almost everyone else, failed
to be clear about this."
"There is an extreme version of the Strong AI thesis against which his
attack is successful. But this leaves open the status of a weaker
version, which is probably the one his commentators thought they were
defending (insofar as they were clear about what they were defending).
This paper is mainly concerned with clarifying the problems. A weak
version of the Strong AI thesis is defended at length in [Sloman 1985]."
"I don't claim that Searle understood enough to have any very precise
notion of computation in mind, and it is likely that he slid unwittingly
between the different versions I shall distinguish. But for the purposes
of his argument, and his more rhetorical flourishes, he seemed to use a
very limited characterisation."
"But Searle also claims that there are certain neurological conditions N
such that
N1. N(x) is necessary for M(x)
where N refers to a collection of causal powers to be found in brains,
but which do not come into existence merely because any purely formal
pattern of processing occurs.
He does not actually produce any argument for N1, and he is totally
vague about the precise nature of N, apart from allowing that in
principle N might be satisfied by some electronic artefact, provided
that it was sufficiently similar in its powers to the brain. Neither
does he give any reasons why N should be necessary for M."
"Consider again the 'Strong Calculator Thesis'. If a physical process
just happened to 'instantiate' some algorithms for numerical
calculation, but did so purely by chance, like the leaves in the wind,
would this be calculation? Does the strong calculator thesis apply only
to behaviour produced by some appropriate causal mechanism? If so, this
may give us a clue as to what, if anything, is correct in Searle's
argument."
"Can a person function as a computer?
We can now return to Searle's Chinese-understanding experiment. He has
assumed that for the Strong AI thesis it does not matter how the
behaviour is produced, so long as it instantiates an appropriate
program. So it doesn't matter whether the instructions are followed by a
computer or a person or even whether the corresponding processes are
produced by some purely random process. But we now see that it is
important to distinguish different cases. In particular, the behaviour
in a computer does not merely instantiate a program; rather a physical
representation of the program causes the behaviour to occur, with the
______
aid of circuitry (and software in some cases) carefully designed to
ensure that there is a reliable connection between program and
behaviour. Any engineer designing an intelligent system would require
something like this. Evolution likewise."
"What's Right in Searle's Argument
I conclude that Searle is right to suggest that, as far as our ordinary
concepts of calculating and understanding are concerned, not just any
production of patterns fitting an (appropriate) program specification is
enough to produce understanding. He is right to suggest that the
processes must be produced by a system with the right causal powers.
This is a requirement not only for mental processes, but for any
computational process which is used as a reliable basis for taking
decisions or controlling anything. The mere fact that a pattern of
leaves or water molecules happens to solve a mathematical problem, does
not make either the leaves or the water into a calculator.
However, if this is the upshot of Searle's argument, I doubt whether he
or any of his opponents would have bothered arguing about it, had they
seen this clearly."
[JQB again] Note that this is *all* that Sloman says is right about
Searle's argument. One has to read his paper carefully and
intelligently, and without trying to find in it what you want to find,
to understand all of what Sloman is saying. For the most part, he uses
Searle's argument as a springboard for discussing different notions of
"Strong AI" and what characteristics a system must have to actually live
up to it; this is very similar to Chalmers' computation paper cited
above. He challenges Searle's *premise* that no understanding takes
place in the room, challenges the appropriateness of using ordinary
words like "understanding" to draw dividing lines among behaving
systems, challenges the claim that the Searle homunculus wouldn't
necessarily understand Chinese if the experiment were carried out and
challenges Searle's "introspection" as a basis for the claim, and even
challenges the implication that, just because the Searle homunculus
wouldn't understand Chinese, neither would a computer in his place; in
fact, he suggests that a hypnotized person might better understand
Chinese as a consequence of following the rules than a fully conscious
Searle. Basically, he leaves Searle's arguments in shreds, rather than
"freely conced[ing] the validity of Searle's argument". Sloman makes
quite a different argument than Searle does, a much better informed and
reasoned and more sophisticated one.
For those who want to read the paper themselves (and judge how both
Weinstein and I have misrepresented it), they can find it at
ftp://ftp.cs.bham.ac.uk/pub/groups/cog_affect/searle85.text
And more recent thoughts from Sloman can be found at
http://www.cs.bham.ac.uk/~axs/misc/strong.ai.searle
One doesn't have to agree with everything Sloman says (I don't) to at
least recognize a level of rigor and expertise lacking in Searle's
writings, or Weinstein's, in which he writes
"My point exactly: Sloman's response did not take issue with the
validity of Searle's argumentation. He disputed whether Searle's
interpretation of the key thesis of Strong AI was in fact a reasonable
one to consider. To me, this is not the same kind of objection as
pointing out an error in logic.
Maybe this is a pointless terminological quibble. But to my mind, the
articulation of a subtle ambiguity in the interpretation of a key
thesis, such as Aaron Sloman presented, is not at all the same as the
diagnosis of a simple error in deductive logic."
Where Weinstein shows what a MARTIAN he is (see Weinstein's previous
suggestion that Martians who accepted affirming the consequent as an
inference rule would not be recognized by us as thinking), since I never
claimed that "Sloman diagnosed a simple error in deductive logic", and
so to point out that he didn't is irrelevant; in fact, Sloman had bigger
fish to fry, and so nothing follows from the fact that Sloman did not
come out and say "Searle made a silly logic error".
Weinstein also shows how DISHONEST he is, since his point was NOT
"exactly" the negative that "Sloman's response did not take issue with
the validity of Searle's argumentation", but rather the positive
implication in "Didn't Aaron Sloman's response freely concede the
validity of Searle's argument?" But in fact Sloman conceded no such
thing, beyond conceding that Searle's argument is circular, which in
fact is a subtle way to point out that Searle' *logic* is kid stuff, and
not even worth addressing. But Sloman accepts neither Searle's premises
nor his conclusions as those conclusions apply to anything other
than "Strong strong AI".
--
<J Q B>
Jim Balter
In reviewing some of Aaron's postings, I found this choice bit when he
was responding out of a mistaken belief that I had claimed that his
argument at most showed that a CPU does not understand, when in fact I
was talking about Searle's argument:
> [Jim]
> > As Neil
> > has pointed out, at most it refutes the claim that a CPU chip
> > understands Chinese, a rather different claim from any form of Strong
> > AI.
>
> No: my argument does NOT refute this. For a CPU chip is far more
> than a formal syntactic structure.
>
> In fact I've argued (in my IJCAI 1985 paper) that there is a limited
> kind of understanding that is actually to be found in a CPU without
> any AI programs needed. E.g. computers work *because* they are built
> on CPUs that have a minimal understanding of addresses and
> instructions and counters, and a few other things.
>
> That was a point Searle never grasped, I think. But I would expect
> you and Neil to understand it well. So I am surprised you think I
> would try to refute it.
So here we have Aaron rather vigorously DENYING that he has shown that a
CPU chip does NOT understand Chinese. Weinstein has some balls claiming
that Aaron "freely concedes the validity of Searle's argument".
Continuing:
> The claim that there's no understanding at all in such a mechanism,
> which I would expect to come from various philosophers (including
> Anders) is probably based on the all too tempting notion that we
> have a very clearly delimited dichotomous concept of understanding
> such that for all X either X understands or it doesn't (what Minsky
> calls "dumb-bell" concepts).
I doubt that either Minsky or Sloman were unware of the ambiguity of
that term.
In that posting Sloman refers to another of his papers, which looks like
excellent reading for anyone who actually wants to understand these
issues and how they might be solved rather than takings Searle's "it
can't be done" approach or Weinstein's "problem? what problem?"
approach:
http://www.cs.bham.ac.uk/~axs/cog_affect/Sloman.turing90.text
And here's more, in which Aaron responds to Weinstein:
> From: A.Sl...@cs.bham.ac.uk (Aaron Sloman)
> Date: 1997/01/22
> Message-Id: <5c3rql$5...@percy.cs.bham.ac.uk>
>
> ande...@pitt.edu (Anders N Weinstein) writes:
>
> > Date: 19 Jan 1997 22:44:35 GMT
> > Organization: University of Pittsburgh
> >
> > In article <5btr12$9...@usenet.bham.ac.uk>,
> > Aaron Sloman <a...@cs.bham.ac.uk> wrote:
> > >
> > >The claim that there's no understanding at all in such a mechanism,
> > >which I would expect to come from various philosophers (including
> > >Anders) is probably based on the all too tempting notion that we
> >
> > Right you are. I would agree with Searle that computers are not the
> > right sort of subjects to "understand" anything, even minimally. As
> > he says, they're just not in that line of work.
>
> Sounds like mere linguistic dogmatism to me.
>
> > On the other hand, there are metaphorical uses of the words: We could
> > say that the immune system "understands" or can "recognize" the
> > difference between pathogens and normal tissue, can get confused
> > about this, attack normal cells by mistake, or whatever. No one thinks
> ^^^^^^
> > such metaphors attribute any sort of mentality to the immune system.
>
> How can you possibly make such a sweeing generalisation?
>
> I happen to think exactly that. I don't regard these things as
> metaphors to be contasted with something literal.
>
> When we talk about the visual system "making inferences", "analyzing
> sensory data", and the like we are talking in this same useful
> but non-literal way.
>
> When I talk like that I mean what I say to be taken literally. If
> you go searching around for metaphorical interpretations, you'll
> never understand what I mean.
>
> > The mistake is to confuse this handy but not
> > literally true way of speaking with the literal sense in which a
> > person can analyze some data represented in a graph or table.
>
> This is just dogmatic linguistic imperialism.
>
> > I think the primary uses apply to the activities of whole persons or
> > organisms. A lot depends on the subjects' having a life to lead in
> > which the activities play a role. Using this as a basis we describe the
> > concepts of sub-personal events and process by analogy.
>
> Speak for yourself. I don't.
>
> > ...
> > It need have nothing to do with dichotomous concepts. There can be many
> > degrees of understanding. A beginner or child or chimp or a dog might
> > be said to have an incomplete or otherwise different understanding of
> > many things.
>
> But you are still assuming a sharp dichotomy between things that can
> *literally* understand anything, whether completely or incompletely
> and those that cannot.
>
> > >Whereas in fact (I claim) our normal notion of understanding, like
> > >our notions of consciousness, perception, experience, desire,
> > >emotion, freedom etc. is a cluster-concept involving a large number
> > >of components that can coexist in different subsets.
> >
> > Agreed.
> >
> > >A CPU just happens to have a very small subset.
> >
> > Why not say that the stomach understands proteins?
>
> Why not indeed. I don't know enough about how the stomach works, but
> I suspect that your digestive system is at least as sophisticated at
> taking in and processing and using information as an ant or wasp.
> (I recently suggested to a teacher that that might be a good way to
> teach kids about the digestive system.)
>
> In both cases, the understanding is of a very limited sort.
>
> > A player piano
> > understands music?
>
> Well, you know as well as I do how little it understands, so what
> point are you making with this example.
>
> It's the old debate about thermostats. Lots of philosophers
> completely failed to grasp the point McCarthy and Dennett were
> making in saying that a thermostat has beliefs and desires of a
> sort. (A very primitive sort).
>
> > These things just happen to have a small subset of
> > the full concept...
>
> Yes.
>
> Just as a circle as a limiting case of what we mean by "ellipse",
> and in the mathematical sense, is an ellipse, even though it does
> not have two well defined major and minor axes, nor two foci, etc.
>
> If anyone argues that it is not REALLY an ellipse, that calling it
> an ellipse is just a metaphor, etc. then I'll just lose interest in
> the discussion. I don't care about such verbal quarrels.
>
> What's important is not whether the limiting or reduced cases really
> are ellipses or really, literally, understand, but exactly what that
> sort of case is like, how it differs from other cases, what sorts of
> intermediate cases there are etc. Linguistic debates about whether
> descriptions are literal or not strike me as boring and pointless.
Finally, I found something from Aaron where he touches upon an
error that I have mentioned, in which Searle conflates the
Searle homunculus with the room as a whole and dismisses any
"understanding" that might flow from any other portion of the room:
> What I think is interesting about the Chinese room argument is that
> it introduces a nice example of two systems being in control of the
> same processes: there's Searle's collection of beliefs, desires,
> etc. as he reads the instructions and carries them out. And then
> there's the control exercised by the set of instructions. These are
> very different.
That Aaron does not dwell on this error in his 1985 paper does not mean
that he was not aware of it, although he did refer one of the
implications, which he reiterates following the above:
> It is arguable that in this case the program does not have adequate
> control. E.g. any time Searle gets bored he can stop, or if he feels
> mischievous he can produce "wrong" intermediate states or wrong
> outputs, etc. Even if he doesn't actually do that, the fact that he
> could do so reduces the control of the program over what's going on.
This is very similar to what he wrote in the 1985 paper. Perhaps Sloman
didn't mention there that this follows from the Searle control system
being very different from the program that Searle follows because this
just seemed too obvious, or because it was only a side issue in a long
paper, or because he just hadn't thought of it that way. But that was
1985 and this is 1997, and in the mean time much has been published on
the subject, including Chalmers' polpular book in which he makes a
similar point, and some people have learned things in between
while others apparently have not.
With material from Aaron like this, Anders has some balls trying to use
him as support for Searle's argument.
--
<J Q B>
Sven-O. Berg
Woops, I made a mistake: It shouldn't read
>Now if neither the assumptions nor the
>conclusion are contradictory themselves,
but:
:Now if neither the assumptions nor the negated
:conclusion are contradictory themselves,
Sorry for the confusion.
BTW, as you (Chris) asked for links: Check out Wittgenstein's
``Tractatus logico-philosophicus''...
Sven
>I guess you might find this definition in every basic textbook on
>formal logic; I remember dimly that we had an excellent one named
>``Beginning Logic'' by J.E. Lemmon, but this one is from 1967, so
>I don't know whether it's still available.
>Sven
>--
>write to soberg (at) internet (dot) de if you wish to reply by email
--
Write to soberg (at) internet (dot) de if you wish to reply by email.
Joseph O'Rourke
In article <333756...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>A typical example of what he uses in place of rigorous thinking is
>to be found in his review of Chalmers' book, published in the March 1997
>Philosophers".
I wonder if you'd care to expand on this? Can you point to any
specific flaws in his reasoning in that NYRB article? I also
found it wanting (and even wrote a letter to the Editors), but
I do not feel it was straightforwardly fallacious.
r.j.fa...@stirling.ac.uk
In article <E7q23...@gpu.utcc.utoronto.ca>,
pin...@gpu.utcc.utoronto.ca (Andrzej Pindor) wrote:
>
> Has Daryl McCullough said that in his view CR "experiences understanding"?
> I'd be very surprised. Can you provide a quote? "Experiencing understandingg",
> being a subjective thing, cannot be ascertained by anyone outside, can it?
I entirely agree. However:
>Daryl McCullough <da...@cogentex.com>
>Date: 1997/03/20
>Message-Id: <33313D...@cogentex.com>
>Newsgroups: comp.ai.philosophy
>
>The systems response is the
>claim that the system, as a whole, has the experience of understanding,
>although the man inside the room doesn't.
Seemed odd, to me, at the time.
Robin Faichney
I'm stooping to following-up my own post, because noone else
has done so and I want to modify what I'm saying in the hope
that someone will feel capable of responding.
On Wed, 26 Mar 1997 I wrote:
>
> (a) Are SR supporters in this forum hopelessly split as to
> whether the CR is supposed to experience understanding, or is
> there yet hope of a consensus emerging?
OK, I can live without an answer to this one. Only Daryl
McCullough has said explicitly that the SR position is that
the system as a whole experiences understanding, so I'll
take it that this is the minority position, and disregard
it from now on. The consensus, I assume, is that Searle
fails to show that the system does not experience
understanding.
> (b) If the CR is *not* claimed to experience understanding,
> is understanding attributed to it on a purely behavioural
> basis, or is there some other consideration?
This was intended merely to lead in to a discussion on the
attribution of consciousness. Which is not to say that any
attempt to answer it would not be interesting in itself, but
I've given up hope of that now, so I'll say a little more
and we'll see if anyone is willing to take me on.
My first serious contribution to this thread was inspired by
the discovery that Jim Balter and myself are in agreement
that the main motivation for attributing consciousness is
emotional.
When I last regularly contributed to this group, very roughly
2 years ago, one of the points I tried to make was that those
who see science as having the capacity to design and build
machines with not only intelligence but consciousness should
give some thought to the psychology of its attribution. Now,
while that most definitely *is* worth thinking about, and I'm
glad to see that Jim for one has been doing so, my hope was
that ultimately that might lead to a deeper appreciation of
the philosophical problem.
The trouble is, as Andrzej Pindor wrote in article
<E7EoC...@gpu.utcc.utoronto.ca>:
>the very notion of
>"subjective" (like in "subjective sense of having the experience of
>understanding") bars anyone, Searle included, from being able to make
>firm pronouncement about anything/anyone else having "subjective sense"
>of anything.
My point is that in fact, in this context, there is *nothing
but* the psychology of attribution. (Where "psychology" is
very loosely interpreted to include sociology, anthropology,
linguistics, etc.) So to the extent that AI aims at a machine
that *really* experiences anything, it is fundamentally
misdirected, and in fact the only possible valid aim it can
have (in this context) is to produce a machine that people
believe to be conscious. Of course, various other serious
problems now arise, e.g. which people, how do we test their
belief and eliminate the possibility that it is motivated
by some factor other than just the machine's behaviour, and
so on. But leaving these aside for now, can I *please* have
some reaction to my main conclusion:
In so far as it is concerned with conscious machines, what
AI is *really* about is not producing such machines, but
producing machines that people can be convinced are
conscious, the means by which that is accomplished being
ultimately irrelevant.
One interesting aspect of this is that it brings the Turing
Test right back to center stage (if it ever left it).
(I'm on holiday for a few days now, so any response to this
may not be responded to until 7 April, though I may look in
to check email and this group before then.)
--
Robin Faichney
r.j.fa...@stirling.ac.uk (delete ".nospam" to reply to me)
http://www.stir.ac.uk/envsci/staff/rjf1/
Jim Balter
r.j.fa...@stirling.ac.uk wrote:
>
> In article <E7q23...@gpu.utcc.utoronto.ca>,
> pin...@gpu.utcc.utoronto.ca (Andrzej Pindor) wrote:
> >
> > Has Daryl McCullough said that in his view CR "experiences understanding"?
> > I'd be very surprised. Can you provide a quote? "Experiencing understandingg",
> > being a subjective thing, cannot be ascertained by anyone outside, can it?
>
> I entirely agree. However:
>
> >Daryl McCullough <da...@cogentex.com>
> >Date: 1997/03/20
> >Message-Id: <33313D...@cogentex.com>
> >Newsgroups: comp.ai.philosophy
> >
> >The systems response is the
> >claim that the system, as a whole, has the experience of understanding,
> >although the man inside the room doesn't.
>
> Seemed odd, to me, at the time.
From Searle's "Minds, Brains and Programs" in _The Mind's I_:
"But Newell and Simon (1963) write that the kind of cognition they
claim for computers is exactly the same as for human beings.
I like the straightforwardness of this claim, and it is the sort of
claim that I will be considering."
....
"1. The Systems Reply (Berkeley). ``... understanding is not being
ascribed to the mere individual; rather it is being ascribed to this
whole system of which he is a part.''"
The ad reductio premise of the CR which Searle wishes to assume in order
to demolish is the Newell and Simon claim. *Whatever* we mean by
"understand" in humans, the program in the CR is ex hypothesi
"the right program" to produce it in the CR. If humans "experience
understanding", whatever that might mean, then ex hypothesi the CR
"experiences understanding".
But, and here seems to be a major source of misunderstanding, the SR
does not make the positive assertion that there is such a program
that produces "understanding" or "experiencing understanding".
Rather, it says that *if* the Simon and Newell functionalist
thesis is correct, that there is such a program, then the understanding
is to be ascribed to the room as a whole, not to the Searle homunculus.
It would do no good for the SR to simply assert that the room
understands; that would not satisfy the burden of rebuttal, since Searle
claims to have just proven that no such thing is possible. Rather, the
SR rebuts Searle's argument, not necessarily his conclusion, by pointing
out that he is ascribing the claimed understanding to the wrong thing,
the Searle homunculus, and thus Searle pointing out that the
Searle homunculus doesn't understand Chinese does not disprove
Simon and Newell's thesis, which is what Searle is attempting to do.
However, the history of the discussion of the CR indicates that very
few of the correspondents understand the burden of rebuttal or the
proper structure of logical argument in general.
--
<J Q B>
Andrzej Pindor
In article <8593709...@dejanews.com>, <r.j.fa...@stirling.ac.uk> wrote:
.............
>So, I still await a substantive reply to my questions, which
>I will restate here for your convenience:
>
>(a) Are SR supporters in this forum hopelessly split as to
>whether the CR is supposed to experience understanding, or is
>there yet hope of a consensus emerging?
>
(or is supposed to) are hopelessly confused. As everyone here should know,
we cannot with certainity say that anyone, except us, "experiences
understanding" (or anything else). Being convinced that we, ourselves, do,
we extrapolate this to others, which are "like us". If they are different, we
are at loss - do dolphins experience understanding? CR contraption is so
different from "us", that making pronouncements of such nature is just silly.
The only thing we can say is that something "shows understanding". I am puzzled
why you fail to see this and persist in creating strawmen?
>(b) If the CR is *not* claimed to experience understanding,
>is understanding attributed to it on a purely behavioural
>basis, or is there some other consideration?
>
On what basis do attribute understanding to a student you are examining?
Surely, you have done so, haven't you?
Andrzej Pindor
In article <5ha3ep$n...@usenet.srv.cis.pitt.edu>,
Anders N Weinstein <ande...@pitt.edu> wrote:
>In article <E7M6z...@gpu.utcc.utoronto.ca>,
>Andrzej Pindor <pin...@gpu.utcc.utoronto.ca> wrote:
>>In article <5h6bjv$3...@usenet.srv.cis.pitt.edu>,
>>Anders N Weinstein <ande...@pitt.edu> wrote:
>>>
>>>ordinary psychological concepts, in particular on what is necessary for
>>>determinate norms that constitute meanings. E.g what is required for you to
>>>have a concept of tables that could succeed in referring to artifacts
>>>in the common world?
>>>
>>Exactly. So what is required?
>
>I would say it requires an organized system of human social practices in
>which tables play a role. The pre-existing social system reproduces itself
>via the individual when a new individual is initiated into the skills
>of operating within it. So that the skill or know-how that is your
>understanding of tables is internally or conceptually related
>to this system of cultural norms and its history of cultivation in you.
>
object with human society. This is a crucial thing I have been pointing out
to you several times. The long-winded answers of yours come down to this fact.
I do not see why it is necessary to write down tens and tens of lines if it
all can be succinctly expressed by the word "interaction" - what makes
an object a table is its interaction with humans. Perhaps philosophers are
paid by number of words (:-)), but most of these words do not advance
the issue. They even obscure it.
..................
>>>Well it seems to me as soon as you separate off "experiences" and treat them
>>>as *inputs* to "the mind", you are already well on your way down the
>>
>>So if experiences are not *input* to the mind, where do they come from? are
>
>I guess I could say they come from the environment.
>
>>they generated internally? If not fully internally but in some correlation
>
>No, that would be hallucinating.
>
>>with inputs, what is the remaining constutent of the experiences? Where does
>>it come from?
>
>You could I guess say the experience itself is the input. For example,
>the experience I referred to as my becoming conscious of my wastebasket
>as empty is itself a possible input to further reasoning I might do.
>
>Of course, to forestall your probable objection, my having this
>experience certainly requires on the presence of light and the
>stimulation of the retina and the operation of the visual system. It
>might itself be a state of my brain as far as I am concerned. Roughly,
>I take it that all these neural processes have two aspects, just as the
>process of writing out a check has two aspects, a material one and a
>socially significant one. Under one aspect, all these neural operations
>are physical events in accordance with the laws of physics. But under
>another aspect, some of them might also constitute, for example, my
>becoming directly or non-inferentially aware of my wastebasket as
>empty. So that all these merely causal prerequisites are not themselves
>*inferential* bases for my subsequent beliefs.
>
of an event (or an object) from effects of its interaction with a wider
system (human society). I have pointed out to you that interactions within
a system are also a subject of physical science. Are you listening?
>In particular, I need not say that under the second aspect the
>epistemic event is caused by some more basic "input" to my awareness --
>what is input to my awareness is just the information about the distal
>objects itself.
>
Who says about "more basic input"? You do not seem to be listening. I was
talking about effects of system-wide interactions. If a tulip bulb is
worth a fortune today and is nearly worthless tomorrow, this is not and
effect of a change in the tulip bulb itself (at some 'more basic' level),
but an effectof the changes in the system which result in the said tulip
bulb having a different effect on (interaction with) this system. I realise
that this way of looking at things may be very foreign to you, but
hopefully you can still learn a new perspective, can't you? The advantage of
this perspective is that it does away with all this vague stuff which you are
producing and which have no verifiable content.
..........
>> The mind is indeed trapped, but behind (veil
>>or not) inputs. Its only contact with the world is through these inputs. If
>
>I disagree. It's contact with the world can be through information carried
>in patterns of (what you call) "inputs". Information is in a sense an
I do not understand - first you say that you disagree and the next sentence
seem to be in agreement with what I said.
>abstraction. For example, when I detect the information that this
>wastebasket (here I'm pointing in my office to W1) is empty, what I detect is
>in a way an abstraction -- the proposition that this wastebasket [W1]
>is empty. But it is something that I think might also be taken to be
>immanent in the patterned radiation reaching my retina in some way.
>
Are you saying that the patterned radiation reaching your retina has
information in it about wastebasket?? Obviously the discussions with you are
a waste of time - we discussed this very point before and I have provided
you with arguments why this cannot be so to which you have not provided
an answer. Now you are back to the same claim. I'll try once more, but will
not attempt this again.
I'll take a simpler example, so as to minimize a danger of you sidetracking
the issue. Consider looking at a white surface and seeing a letter "A"
written on it. From what you have said it seems that in your view light
reflected from the surface and falling on your retina will contain in its
physical patterns an information about letter "A", right?
You will agree that even for other human cultures the same pattern may have
another meaning, right? Even more so for other possible minds, Martian, etc,
right? There is no reason that there could not be a virtual infinity of
other minds, each of which would see something different in the pattern of
light. Are you then saying that all this possible interpretations are
contained in the pattern of light? As far aa I am concerned this much more
inconceivable than assuming that letter "A" is an interpretation of the
physical pattern of light provided by the particular human mind, and
resulting from the structure of the brain, including all changes to it
arising from the of social interactions (learning the alphabet, etc), much
in the same way as a given input to a computer (a sequence of 0's and 1's)
can rtesult in a different output, depending on what program is runing in
the computer. You are not hopefully saying that a given sequence of 0's and
1's contains in it all posible outputs which it can lead to with all possible
programs which can take this input, are you?
Now, unless you can provide a counterargument making sense, please do not
repeat the same claim. As far as above goes, it does not hold water.
>In this way I do not in any way *infer* this content from
>representations of patterns on my retina -- I certainly have no
>conscious representations of patterns on my retina, so I could not
>perform any such inference. Rather I am now able to *extract* the
>information the information about the world that is already (I know not
>how) coded there.
>
As far as I can see, the problem here is that you never seem to consider
what is this "I" above you are talking of, objectively speaking. This
is why I )and others) see as dualist, because you refuse to make a connections
of this "I" with the physical system which a human is. Hiding behind
philosophical jargon like "epistemic subject" or like is ducking the issue.
>>you disagree, please explain what are the other ways it makes contact with
>>the external world. Have you heard about virtual reality? Are you not aware
>
>I have heard of virtual reality. I gather most current systems are
>rather low quality and would not fool anyone. To really get a total
>body simulation of sufficient quality would take an awful lot of
>computational power. I believe as a matter of engineering fact that it
>is very hard to reproduce all the relevant complexity of our
>interactions with a real environment with anything short of the real
>environment itself.
>
Now you are hiding behind technical difficulties. If you are not saying
that such thing is in principle impossible, why raise the issue?
Consider only an audio input. You can have a full illusion of being in
front of a waterfall (with your eyes closed) even if the audio inputs which
make you to.think so would be produced by a couple of well placed
loudspeakers in an empty room.
>But in any case I am talking about perception of the real reality, for
>example my office. If you are in a virtual reality then you are not
>really perceiving anything, you only seem to be. I don't recall
>saying that one could always tell the difference.
>
But if you can't tell the difference, what sense does it make to claim that
we see (or hear) directly the objects of the world?? Our brain/mind responds
then only to the physical inputs, and not the "real" objects of the world.
So what sense does your claims make? It looks like they are quasi-religious
beliefs, and are not subject to logical or empirical verification. If you
disagree, please explain what logical sense do your claims above make?
>>that the mind is unable to tell where do the inputs come from - from a real
>
>Again, a conscious subject just is not in the position of being presented
>with "inputs" and having to wonder where they come from. As Sartre put it,
>"All consciousness is consciousness of an object that is not itself" (or
>something like that)
>
>That is to say, (perceptual) consciousness is exhausted by such things
>as my becoming conscious of facts about my office furniture -- or, if
>it is deceptive, merely seeming to be conscious of putative facts about
>putative furniture. But there are no ideas or images or any other kind of
>"input" to be found in normal perceptual consciousness.
>
And how does the above answer the basic question: if we can be conscious of
a waterfall even if we only receive sounds from a couple of loudspeakers in
an otherwise empty room, what sense does it make to claim that we are
"directly conscious of the world"?
..............
>Well sure, if you are really seeing a wastebasket then there really is
>a thing you see, for example, and your experience is in some way
>responsive to *it*, the real wastebasket which anyone else can see and
>touch. In a virtual reality you are not in touch with any real
>wastebasket, for example, your experience is not caused by a real
>wastebasket but induced by a computer from something other than a
>wastebasket.
>
You seem to overlook the fact that you are relying on the presence of
an "external" arbiter who knows whether there is a 'real" wastebasket or
not. Where (and/or who) is this arbiter to tell us that there really is
a wastebasket there and not something very different which affects all our
senses in a way which we interpret as a wastebasket in a analogous (but very
limited) way to the way the sounds form the loudspeakers in the example above
persuaded us that we are in front of a waterfall ? You are clearly assuming
that there exists, what is termed by some people "a God's eye view". Apart
form a religious stance I do not see a justification for the existence of
such a "view". Perhaps you should read Lakoff's "Women, Firde and Dangerous
Things".
.............
>>When we "look and see" we provide our minds/brains with physical inputs.
>
>Possibly. But we ourselves never see those inputs. When we look/see we
>often manage to provide *ourselves* with information about the
>environment. We might at about the same time provide our brains with
>its inputs, but the inputs to the brain are certainly not inputs to our
>conscious awareness.
>
Besides the fact that "we" and "see" above are the terms you seem to take for
granted. i.e. without any reflection what do they in fact mean to you (and
a gurantee that they mean roughly the same to others, note the problem
with your use of the word "objective") I migh agree, but so what?
As I can see the inputs to the conscious awareness come from other parts of
the brain (or are produced by other brain processes) from the physical
inputs to the brain. If you disagree, you must be separating the conscious
awareness from the physical brain, which would be dualism you deny.
>Note I take it to be very important here to think phenomenologically,
>that is to concentrate on how things are from the subjective point of
>view of the conscious subject whose mode of awareness seems to provide
>a vista opening out directly onto the distal world -- and which, in
>veridical cases, really *does* provide such a vista.
>
Fine, but if you do this you will be unable to understand/explain
a relationship of this "conscious subject" to the physical brain. By refusing
to see such relationship you are falling into the trap of believing in
"direct seeing" instead of realising that this "seeing" may come from
transformations of physical inputs by subconscious processes in the brain,
on the basis of the previous brain history and physical architecture of
the brain. I know that you "dislike" such interpretation, but have tried
to reflect why? Perhas this dislike, so intense that you refuse to give
a fair consideration to such views, prevents you from getting a fresh
perspective on the subject of your interest.
>I don't really give a damn about the details that happen in the retina
^^^^^
>or the brain, I am talking about the metaphysics of the
Well, see above.
>phenomenological level or about the nature of the subjective experience
>of a conscious subject; I say, get the characterization of *that* level
>right, as, e.g., Sartre more nearly does, and then all the traditional
>philosophical problems of empiricism concerning the possibility of
>knowledge or intentionality answering to something outside the mind will
>just vanish.
>
If you are refering to the quote from Sarte you have provided, I do not see
how it would do what you expect. I'd rather think that Wittgenstein provided
a way to dispense with a majority of philosophical problems, like your
confusion concernig various meanings of the word "is" which I have pointed
out in my posting and you have conveniently ignored.
>That is why I think we don't need more empirical science of the wholly
>unconscious (and therefore non-mental) goings on in our visual systems
>to help us explain how this sort of intentionality is *conceptually*
>possible. We do need that science to help explain how it is *causally*
>possible that we should have the perceptual capacities we manifestly
>do, and such sub-cognitive science regarding its implemention level is
>a perfectly good thing in its own right. The latter level is not very
>philosophically exciting to me since it seems to be one level down from
>the level I find most interesting, the phenomenological or, if you
>like, the intentional systems level. And there does seem to me to be an
>ever present danger of confusing the two levels, of treating that sort
>of cognitive science as the real scientific account of our thinking being.
>
We obviously differ in what we understand by "scientific account". You seem
that it is a language game which you "like". For me it is something which
can be empirically verified..
>As to "mind" in a way I would say there is no such thing.
Well, if so what is this "we" or "I"?
Anders N Weinstein
In article <333E0C...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>So here we have Aaron rather vigorously DENYING that he has shown that a
>CPU chip does NOT understand Chinese. Weinstein has some balls claiming
>that Aaron "freely concedes the validity of Searle's argument".
We are going to have a difficulty in that Sloman has made many comments
on Searle's argument over the years. In taking apart a complex issue
one can focus on many different unclarities, or different lines of
possible response.
I was referring explicitly to Sloman's 1985 paper on Searle -- I recall
the title as something like "Did Searle refute Strong Strong AI or Weak
Strong AI?" (a copy is on his web site). If you read it, I think you
will find it about as unambiguous as anything in philosophy ever is
that that paper conceded the validity of the Searle argument *as an
argument against "Weak Strong AI"*.
Of course there and elsewhere Sloman points out a great variety of
problems with Searle's argument. For example, he argues that a
refutation of Weak Strong AI is unexciting since the thesis on that
interpretation is very implausible, and he claims that is not what is
actually implicitly governing the thought of AI researchers (although
very few other respondants took this line against Searle).
I must reiterate: I never said that Sloman accepted the point of the
Searle argument as Searle evidently intended it. Merely that he
conceded its *logical validity* on a natural interpretation. So that
while he found much in Searle to take issue with, it was not what I
would call a trivial mistake in Searle's *logic*.
Now in this post, you are pointing to a very different context in which
Sloman argued against bad "linguistic dogmatists" that it is perfectly
appropriate to say that a CPU chip has some kind of understanding,
presumably an understanding of the operations or inputs a CPU deals with.
I don't think this idea of Sloman's really does furnish support for
your own view. First, Sloman *is* ascribing a kind understanding to the
CPU chip, and not to the whole system as the Systems Reply does. Second,
he does not, so far as I know, deploy this idea as the basis of his response
to Searle's Chinese Room argument.
That is perfectly understandable, since Sloman is not, so far as I can
tell, trying to ascribe anything like understanding of *Chinese* to the
CPU chip in an implementation of a Chinese understanding program. If I
understand the idea, what the CPU "Sloman-understands" would have to be
something like: memory addresses, the system bus, address decoding,
interrupts, registers and the like. (I might well be wrong about the
view here. The whole concept seems silly and pointless to me, so I
have to stretch to find anything worth saying using it.)
>With material from Aaron like this, Anders has some balls trying to use
>him as support for Searle's argument.
I am afraid you did not understand the *point* of my bringing Sloman's
careful analysis up at all. I do not myself think Searle's polemic is
free of problems. And of course I would never ascribe any such
view to Aaron Sloman. But I do believe the Searle argument is well
worth detailed and careful analysis and discussion, such as Sloman, in
my view, attempted to provide.
But you seemed to have a very different meta-view than I do. You seemed
to me to suggest, if I may paraphrase, that the argument just rested on
some trivial logical mistakes, some kind of howlers well-known to
everyone with the slightest bit of logical ability. These mistakes,
you suggest, have been adequately diagnosed time and again. So that
there really is no point in discussing the argument further at all -- it
would be like discussing any other textbook fallacy. One should just
refer people to the logic books.
I don't think any interesting disagreement in philosophy really does
have this character. Typically these disagreements stem not from
anything we could happily call "logic" but rather from the unclarity of
the crucial terms. But then it is well worth our effort to try to get a
more explicit understanding about the commitments involved in our terms
-- that is just what philosophy has always has been all about.
I brought up Sloman's response only to suggest that some very
knowledgeable AI people do not share your view of where the "well-known
logical errors" are. All I wanted was to defend the "meta-view" that
accurately diagnosing the flaws in Searle's argument is a very *subtle*
matter on which reasonable people can differ, not a trivial application
of logic.
Anders N Weinstein
In article <333DC0...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>Anders N Weinstein wrote:
>>
>> I was recalling the point in his 1985 paper where Sloman wrote of
>> Searle's reasoning:
>>
>> His logic is unassailable, if his premisses are true, in particular the
>> premiss that nothing understands Chinese in this situation.
>
>All Sloman has said here is that circular arguments are valid.
I just don't see how to read it that way. I certainly don't find the
charge that Searle's argument is circular anywhere in that paper.
I think one could fairly say that according to Sloman, Searle succeded
in refuting a straw man (Strong Strong AI). But then he bamboozled
people -- both himself and his respondants -- into thinking he had
refuted a more substantial target (Weak Strong AI). I suppose that might
be a kind of fallacy of equivocation, although both sides seem not to have
noticed it, on Sloman's view.
But to clarify the point: I never suggested Sloman was some kind of
ally of Searle. That would indeed be ridiculous. I was only pointing
out that Sloman evidently differs from you on where the errors are. And
that Sloman's response did not rest in some very simple way on taking
issue with the *logic* of Searle's polemic.
I am getting weary of what may just be a pointless terminological
quibble (since any disagreement with the argument, however subtle,
might in some sense be called a disagreement in logic). I mean only
this: Sloman's careful critique does not rest on charging Searle with
committing a trivial logical fallacy, as you frequently do.
Searle's argument may be unsuccessful. I just think that to suggest the
errors are *trivial* is grossly unfair.
Anders N Weinstein
In article <333DED...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>Where Weinstein shows what a MARTIAN he is (see Weinstein's previous
>suggestion that Martians who accepted affirming the consequent as an
>inference rule would not be recognized by us as thinking), since I never
>claimed that "Sloman diagnosed a simple error in deductive logic", and
>so to point out that he didn't is irrelevant; in fact, Sloman had bigger
>fish to fry, and so nothing follows from the fact that Sloman did not
>come out and say "Searle made a silly logic error".
You suggested that Searle's argument commits a well-known and rather
trivial logical error. I merely pointed out that Sloman's careful
critique does not see it that way, although it certainly finds errors.
Perhaps I *was* insensitive to Slomans "wry expository style". But
although the paper raises plenty of qualms along the way, it does not
rest on the Systems Reply; nor does it charge Searle with circular
reasoning in rejecting it.
In fact, as I see it, Sloman's position on the requirements for a
reliable implementation entail -- quite *contra* the Systems Reply --
that no understanding entity is constituted when a human being
hand-simulates the program. For that is just the wrong sort of
instantiation according to "Weak Strong AI".
>Weinstein also shows how DISHONEST he is, since his point was NOT
>"exactly" the negative that "Sloman's response did not take issue with
>the validity of Searle's argumentation", but rather the positive
>implication in "Didn't Aaron Sloman's response freely concede the
>validity of Searle's argument?" But in fact Sloman conceded no such
>thing, beyond conceding that Searle's argument is circular, which in
>fact is a subtle way to point out that Searle' *logic* is kid stuff, and
>not even worth addressing. But Sloman accepts neither Searle's premises
In my judgment you are projecting wildly when you say Sloman charged
Searle with circular reasoning. That may be a fair thing to say
on your interpretation, but I just don't find that charge
anywhere in Sloman's paper.
In the little bit I quoted, Sloman did include the qualifier "if his
premisses are true, including the premise that there is no
understanding..." I took this to express the author's reservation of
judgment at that point in the paper on whether this premise is in fact
true in the case at issue. So Sloman notes that a possible point of attack
is this premis, even though his main line of response does not take issue
with it.
Once again, recall that the deductive validity of an argument depends
only on its logical form, it does not depend on the truth of the
premises. I guess I must admit Sloman must have been talking about
soundness as well when he said, in informal language, "his logic is
unassailable" -- else why mention truth of premises at all?
But note: if Sloman had gone on to argue that the crucial premiss was
false, that would have shown Searle's argument to be unsound, but not
necessarily invalid. Again, I don't find the suggestion that it would
be circular anywhere in Sloman's paper.
> But Sloman accepts neither Searle's premises
>nor his conclusions as those conclusions apply to anything other
>than "Strong strong AI".
As I read it, Sloman's position in that paper certainly does have the
consequence that no understanding entity is constituted in the Chinese
room. Human beings hand-simulating do not constitute reliable
implementations, and so don't meet the "Weak Strong AI" criteria for
"instantiating a program". So I would say that Sloman does share that
premise with Searle, although he certainly arrives at it by a very
different route than Searle's "imaginative introspection".
Of course I agree Sloman doesn't share Searle's overall conclusions.
The whole paper is certainly a criticism of the Searle argument and I
apologize if I suggested or implied otherwise -- that was not my intent
at all.
I am afraid you may have read way too much in my narrow claim about
not taking issue with the logic.
Anders N Weinstein
In article <E7x3r...@gpu.utcc.utoronto.ca>,
Andrzej Pindor <pin...@gpu.utcc.utoronto.ca> wrote:
>People who expect an answer to the question if CR "experiences understanding"
>(or is supposed to) are hopelessly confused. As everyone here should know,
>we cannot with certainity say that anyone, except us, "experiences
>understanding" (or anything else). Being convinced that we, ourselves, do,
>we extrapolate this to others, which are "like us". If they are different, we
So nice to find someone who philosophizes as if Wittgenstein never
existed or wrote anything at all. Let me just go on the record: I know
no such thing.
"I saw he was in tremendous pain" is as acceptable in context as "I saw
there was a house in front of me". In both cases, claims of these form
can turn out on further inquiry to be false. But in the cases when they
are not false, they evidently include contents that can be known by
seeing.
Further, I think Wittgenstein quite demolished the myth that one could
acquire concepts of mental states by introspection from your own case.
And even if, per impossibile, one could do this magic, there would still
never be any justice in extrapolating from your own unique case to those
of others, if behavior is only externally related to mental states.
On the other hand Wittgenstein was not a behaviorist. In ascribing pain
third personally on the basis of someone else's expression you are not
aptly viewed as making an inference to the pain from some behavior.
Rather, you are conceptualizing it immediately *as* behavior expressive
of pain. But if the person really is in pain, then you are seeing the
expressive behavior as just what it is, not forming a hypothesis.
BTW Wittgenstein also suggested that "I know I am in pain" was, joking
or emphasis aside, nonsense. In part because it could not be used to
inform someone of some contingency that might or might not obtain.
Jim Balter
Anders N Weinstein wrote:
> >With material from Aaron like this, Anders has some balls trying to use
> >him as support for Searle's argument.
>
> I am afraid you did not understand the *point* of my bringing Sloman's
> careful analysis up at all.
You dishonest git. You offered Sloman's statement that Searle's logic
was "unassailable" out of context, as support for your claim
that Sloman "freely conced[ed] the validity of Searle's argument".
But the context shows Sloman's "careful analysis" is in fact not
a careful analysis of Searle's argument, but rather a careful analysis
of the notion of "Strong AI". Sloman's actual presentation of Searle's
argument that immediately preceded his "unassailable" comment, an
argument that provides the context and which you neglected to include,
is *not* the argument that Searle thinks he is offering or that is taken
as convincing by followers of Searle. What Sloman freely conceded
is that, if you deny from the outset that the room itself or the
man in it understands Chinese, then it follows that the Chinese room has
no mental states. There is no careful analysis of that, because it
requires none; there is simply
His logic is unassailable, if his premisses are true, in particular
the premiss that nothing understands Chinese in this situation.
Which is all you offered, out of context, without a whit of
understanding, but merely your usual ideologically driven "philosophy"
that replaces thought and analysis with "Sloman said", Strawson said",
"Gibson said".
--
<J Q B>
Jim Balter
Anders N Weinstein wrote:
> I mean only
> this: Sloman's careful critique does not rest on charging Searle with
> committing a trivial logical fallacy, as you frequently do.
I have already pointed out how stupid, irrelevant, and Martian that is.
Sloman's failure to present a certain argument has no bearing on
the argument; Sloman has other interests. He may not even be aware of
the argument, because it is already clear to him that Searle's argument
fails miserably on other grounds, grounds more interesting to Sloman.
> Searle's argument may be unsuccessful. I just think that to suggest the
> errors are *trivial* is grossly unfair.
That doesn't change the fact that they are trivial. The ad reductio
premise of his scenario is that his mind works on computational
principles. His argument fails to test that premise; it fails
trivially, and all your guff doesn't change that.
--
<J Q B>
Jim Balter
Anders N Weinstein wrote:
>
> In article <333DC0...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
> >Anders N Weinstein wrote:
> >>
> >> Searle's reasoning:
> >>
> >> premiss that nothing understands Chinese in this situation.
> >
>
> I just don't see how to read it that way. I certainly don't find the
> charge that Searle's argument is circular anywhere in that paper.
That's because you are too dishonest to actually look at
how Sloman presented Searle's argument:
Searle claims that provided Pc is expressed in a programming language
he understands, he can take the place of the computer and produce an
instantiation of Pc which will fool others outside the room into
thinking that something inside the room understands Chinese (if Pc is
a good program). But, since he doesn't understand Chinese, and (he
^^^
claims) the total system consisting of him executing the program
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
doesn't understand Chinese, it follows that the alleged mental state
^^^^^^^^^^^^^^^^^^^^^^^^^^
does not exist. So C(x) (that is B(x) instantiating Pc) is not
sufficient for M(x) in this case.
The underlined material is the negation of the Systems reply *as a
premise*. *If* there is no understanding present, *then* "the alleged
mental state", namely "e.g. can understand symbols, or use them to refer
to other things" (from earlier in Sloman's paper) "does not exist", and
so C(x) isn't sufficient to produce M(x). Like, duh. The question then
becomes, under what conditions is there understanding present. That's
beyond Searle; he thinks it is obvious there isn't, and Sloman points
out why he thinks that Searle thinks that: because static structures
don't have the required causal powers. Sloman then goes on, way beyond
Searle, to discuss other forms of Strong AI, non-static forms, that
escape this obviousness. Which not everyone finds obvious BTW, and in
fact Sloman agreed here recently that it was possible to extend
"understanding" in such a way as to apply to static structures, in which
case even Strong Strong AI might be true; I think that such extension
must be granted, because a space-time description of causal
relationships is static.
In any case, Searle thinks he has shown that there is no understanding
in the room simply by pointing at the man in the room. The Systems
Reply is that his argument simply hasn't looked for understanding where
the functionalists claim it is to be found. Rather than accept the
burden of constructing an argument that the room as a whole cannot
understand, Searle crams the room into his head and says "see, I still
don't understand Chinese". But this is mere assertion and sloppy
thinking. Suppose he had crammed the brain of a Chinese person into his
head; what then? Without the proper connections, he would fail to be
able to satisfy the requirements of being able to converse in Chinese.
Make those connections, and now we have Searle walking around, talking
Chinese, and perhaps denying (in English) that he understands it. but
we don't know that in fact he would then deny it, and if he did it would
be far from obvious that such a denial was valid. But Searle and those
who appreciate his having presented an argument that seems to support
their prior biases are too intellectually dishonest to actually
explore these issues. Rather, Searle just says that it is ridiculous
to imagine that he together with "bits of paper" could understand
Chinese
when he alone could not, which is just another *assertion* that programs
aren't relevant to understanding, and such argument by assertion is
*trivially* fallacious.
--
<J Q B>
Chris Malcolm
>(b) If the CR is *not* claimed to experience understanding,
>is understanding attributed to it on a purely behavioural
>basis, or is there some other consideration?
Consider 3 kinds of understanding.
1. UnderstandingT, defined behaviourally (passes Turing Test).
2. UnderstandingS, defined both behaviourally, and because the
right kind of internal machinery is doing the work (e.g. Smith's KR
hypothesis).
3, UnderstandingB, as S, but includes the experience of
understanding (as Brentano, and later, Searle).
Schank originally suggested that his burger-bar script-based program
should be considered as showing a little understanding, in the sense
that it could be developed further to have more of the kind of
understanding we have. In effect this was a claim that he had a built an
understander of type S, and considered that understandingS was a step on
the way to understandingB.
Searle's CR replied to this by mocking understandingT, and then
demonstrating that the most wonderful implementation of understandingT
(or S) would not be understandingB. Which is hardly surprising, since
strong AI (understandingT or S) explicitly ignores the issue of
subjective experience, and will only succeed in implementing
understandingB if type B happens by a fortunate accident to supervene on
a comprehensive implementation of type T or S.
The interesting issue which is buried beneath all the cross-purposes
argumentation of the CR is whether, as Schank suggested, understanding
of type S is a component of understanding of type B, of if it is an
interesting alternative route to another kind of understanding, or else,
alas, doomed for lack of the consciousness-causing widget which is
essential to the best understanding.
--
Chris Malcolm c...@dai.ed.ac.uk +44 (0)131 650 3085
Department of Artificial Intelligence, Edinburgh University
5 Forrest Hill, Edinburgh, EH1 2QL, UK DoD #205
Anders N Weinstein
In article <E7xAA...@gpu.utcc.utoronto.ca>,
Andrzej Pindor <pin...@gpu.utcc.utoronto.ca> wrote:
>In article <5ha3ep$n...@usenet.srv.cis.pitt.edu>,
>Anders N Weinstein <ande...@pitt.edu> wrote:
>
>In other words, something "being a table" is due to 'interactions' of this
>object with human society. This is a crucial thing I have been pointing out
Sure. Note it might alos be that being a person or having mental states
are similarly constituted by this interaction. Think, for example of the
predicates "knows what a table is" or "can recognize a table" or, better,
"understands tables [and the practices in which they have role]". These
also relate one to the institutional facts.
>to you several times. The long-winded answers of yours come down to this fact.
>I do not see why it is necessary to write down tens and tens of lines if it
>all can be succinctly expressed by the word "interaction" - what makes
>an object a table is its interaction with humans. Perhaps philosophers are
>paid by number of words (:-)), but most of these words do not advance
>the issue. They even obscure it.
Well most of the words have been devoted to taking issue with your view
that tablehood is "subjective". Certainly you cannot infer from "due
to interactions" to "subjective".
Secondly, I am not entirely happy with the formulation since it tends
to blur the distinction between constitutive and causal relations.
That tables are social objects is a constitutive fact about tables.
It is not just a fact about causal history.
Finally, I expect I differ from you on what the interactions that
constitute tables as such are. In particular, I am willing to hold
that conscious persons and [such things as] tables can be simultaneously
constituted as part of a social system. So that you don't explain the
constitution of the social in terms of a prior level of individual
mental states.
>Briefly speaking, you are differentiating the immediate physical aspects
>of an event (or an object) from effects of its interaction with a wider
>system (human society). I have pointed out to you that interactions within
>a system are also a subject of physical science. Are you listening?
Of course physical science can deal with causal interactions. It does not
have a notion of a "norm" or correct use of an object, however. In the
physical world there are no norms, but cultural worlds are constituted by
norms and creatures that can act in accordance with them.
So the interactions that I am concerned with are mainly interactions
between an understanding person and a system of institutionalized
artifacts and tools. These descriptions are incommensurable with those
of physical science, I think, although each *particular* act of, say,
pulling your chair up to a table and eating with knife and fork, might
also be explicable in physical terms. But not, I think, the whole
pattern of which that is a part.
But perhaps. There are reductionist programs in social science too. I
find them implausible and unmotivated, but I can't claim a knockdown
argument that none could succeed. Your own sketchy remarks about
"interactions" are hopelessly programmatic, it seems to me.
>>> The mind is indeed trapped, but behind (veil
>>>or not) inputs. Its only contact with the world is through these inputs. If
>>
>>I disagree. It's contact with the world can be through information carried
>>in patterns of (what you call) "inputs". Information is in a sense an
>
>I do not understand - first you say that you disagree and the next sentence
>seem to be in agreement with what I said.
There is a difference between input qua physical energy, and input qua
information-bearing *pattern*.
Here is a very simple example of JJ Gibson's conceptualization:
according to Gibson, there is no need to "infer" that objects remain of
constant size as they approach, even though the angle subtended by
their retinal projection gets larger as they move. Why not? He
suggested that something like the following was possible: information
about the invariant size of the object was naturally specified or coded
in a constant *ratio* in the proximal "input". For example, the ratio
of image size to number of lines of background texture subtended might
be constant.
So, Gibson says: one can think of the mature neural circuitry as a
complex functional component -- a black box -- whose response exhibits
sensitivity to this ratio. Constant size perception is explained as the
result of the operation of this functional component. Because
information about constant size is coded in the pattern of stimulation,
the device does not have to *infer* distance, it just has to be
sensitive to the ratio. If you like, it's input is not just the light,
but the ratio *in the light* that specifies constant size.
Of course this ratio is itself a perfectly good "physical property",
albeit in a certain sense a "higher-order property" of the stimulus
(since it depends on "first-order" properties like the intensity at
various points in the array). So there is no reason on physicalist
grounds that we can't speak of a physical device which is sensitive to
(exhibits a response which reliable varies with) such a property.
In this way, the complex device detects constant size in roughly
the way a simple transducer like a nueron detects light intensity.
On the other hand: that such a physical device constitutes or makes up
a reliable size detector (as opposed to a detector of some other
information) clearly depends on certain facts about the normal
environment of the organism. For example, the facts summed up in the
idea that this pattern of proximal input reliable specifies the
property of distal size, and that the organism comes to rely on it in
the guidance of behavior that ought to be sensitive to size.
Anyway, in this sense we might say: perception requires neural input,
but *not* inference. Rather it is an *epistemically* or *cognitively*
unstructured process of information *detection* -- it does not resolve
into a set of inferential steps -- even if it is physically structured
(it requires sensitivity to lots of neurons and requires lots of parts
to implement).
So the idea is that at the right functional level of description,
the perceptual system is simply detecting information carried *in* the
input, not basing an inference on a representation *of* the input.
In this way, there is really no limit on what can be directly detected.
For example, the information that a mailbox in front of me affords
letter mailing can, in Gibson's sense, be carried in the light and
directly detected, as can the meaning in a glance or facial expression,
or the meanings in a word. Elements of socially constituted reality
like tables and chairs might also be non-inferentially recognized as such.
>>abstraction. For example, when I detect the information that this
>>wastebasket (here I'm pointing in my office to W1) is empty, what I detect is
>>in a way an abstraction -- the proposition that this wastebasket [W1]
>>is empty. But it is something that I think might also be taken to be
>>immanent in the patterned radiation reaching my retina in some way.
>>
>Are you saying that the patterned radiation reaching your retina has
>information in it about wastebasket?? Obviously the discussions with you are
Yes! That is pretty much exactly what I am saying. I don't see why
this cannot be -- an awful lot depends on what you mean by "information".
I would have thought btw that almost all cognitive science concurs that in
some sense the pattern in the light is the source of my information about
the wastebasket.
>I'll take a simpler example, so as to minimize a danger of you sidetracking
>the issue. Consider looking at a white surface and seeing a letter "A"
>written on it. From what you have said it seems that in your view light
>reflected from the surface and falling on your retina will contain in its
>physical patterns an information about letter "A", right?
Yes, now you seem to be catching on to the right way of talking.
>You will agree that even for other human cultures the same pattern may have
>another meaning, right? Even more so for other possible minds, Martian, etc,
>right? There is no reason that there could not be a virtual infinity of
Right -- it takes learning and culturation to develop the ability to
detect the information that is there to be detected.
But I notice there is a possible ambiguity in "same pattern" here --
it could mean something *of the same shape* but produced on Mars as part of
Martian practices. But it could also mean a single concrete particular token,
produced, say, as an English word but looked at by a Martian.
It is clear that different languages can make use of the same shapes for
different purposes -- different languages use the same Roman alphabet
and even similar-shaped words with different meanings, for example. As
Wilfrid Sellars taught us, this is like using the same playing cards or
game pieces to play different games. When we talk about meaning, we are
not talking about physical structure but role in a cultural system of
rule-like norms for the use of linguistic expressions.
So yes, Martians might use similar shapes with different meanings. No
problem here, if meaning is a matter of socio-functional role, not
material. That's just a social analogue of functionalism in philosophy
of mind.
On the other hand, maybe you are talking about a case where a Martian
comes to Earth and encounters a mark I have produced as an English
sentence. In that case, of course it is -- objectively -- an English
sentence whether or not the Martian can recognize it as such. The
Martian would probably take it wrongly at first, but would ultimately
detect the error and try to learn about the system of rules *we* use in
order to perceive it's social function aright.
That also is no problem, since quite generally, one must learn
to see, even if what one sees is there already. If I take an aborigine
to a baseball game, I do not expect him to be able to see what I can
see in it. The same patterns reach his retina, but he has not yet have
the right understanding of the game, including the right detectional
capacities.
>right? There is no reason that there could not be a virtual infinity of
>other minds, each of which would see something different in the pattern of
>light. Are you then saying that all this possible interpretations are
>contained in the pattern of light? As far aa I am concerned this much more
I am not sure what you mean in terms of the above ambiguity. It sounds
as if you are talking about one and the same concrete token, say, a
particular English sentence uttered by me. Now the Martians might take
such a thing to be a Martian sentence -- but then they would just be
wrong, for it is not a Martian sentence. And it would not really stand
up under this interpretation -- since I don't go on to say further
things that make sense in Martian. They would have to learn English to
see it as what it is.
So although a potential infinity of minds might *at first* interpret a
concrete token of English differently, they could not all be right. And
there is no reason they couldn't come to recognize this. They would
almost certainly bump up against gross failures of fit further on down
the line if they really tried to make sense of me as speaking Martian.
This is no more threatening than the fact that the Polish word "ja" is
not the same as the German word "ja" (yes). Most of the time we have no
difficulty discovering the difference, although a single case all by
itself can of course be ambiguous.
There can be some funny cases. A bilingual businessman might be juggling
a phone call on one telephone to England and one on on another to
Germany. His English friend asks him how many men on a baseball team.
The German asks him -- in German -- if he will be placing an order. Seeing
an opportunity to kill two birds with one stone, he holds both
receivers in front of him at once and utters the sound "Nine" (or
"nein").
In this case one and the same noise-making would make up or constitute
two quite distinct linguistic moves, one, an answering of the number
"nine" to England, the other, an issuing of a negative answer to the
query from Germany. That's a little bit odd, but fully intelligible,
given the way linguistic or semantic attributes supervene on physical
attributes of events like noise-makings.
That is far from the usual case, but in any case it is not much of
a threat. I just told you everything you need to know to make it
perfectly intelligible.
If on the other hand you are talking about different cultures using the
same token with different linguistic roles, that also is possible but
poses no threat either. FOr the systematic rules and relationships
which constitute the various meanings are still perfectly discoverable,
even if not necessarily discovered on first encounter.
You keep raising such possibilities as if they presented some kind of
intolerable difficulty, but you never explain to me why there is any
problem to be found here.
> As far aa I am concerned this much more
>inconceivable than assuming that letter "A" is an interpretation of the
>physical pattern of light provided by the particular human mind, and
But you are not dealing with the question of what makes it a *correct*
interpretation in a particular case. Surely, I am not automatically
right every time I take something as a letter "A", just as not every
time I take something to be a house am I always right -- sometimes
further experience shows me false and I have to revise my
interpretation. (Look, I am responsive to empirical evidence, and you
don't seem to be here)
I believe the Greek lower-case lambda is very similar to the Chinese
glyph for Man. I might not be able to distinguish them without further
information about history. But given a particular historical artifact
it is certainly not up to my interpretation which is which. If somebody
hangs some decorative piece of Chinese calligraphy on the wall and I
walk into their house and take this figure for a Greek lambda, it makes
perfect sense for them to explain to me, no, you're wrong, it's not a
Greek lambda (though it looks like one), it's actually Chinese writing
and it means Man.
And when Aristotle, say, looks at some Greek writing and detects
the information that there is a lambda on the page (or more
realistically, detects a word or meaning contained therein), that is
correct only because it is the exercise of an ability that results from an
upbringing in Greece amidst Greek practices of using letters.
So it is a historical fact about the object that it is a lambda. But it
just as much a historical fact about the *subject* that Aristotle (or,
if you really want, a system in his brain) is a reliable
lambda-detector and not a detector of Chinese figures. (Of course part
of the difference will also show up in differences in other parts of
his brain, but I don't think this needs to be the case).
>resulting from the structure of the brain, including all changes to it
>arising from the of social interactions (learning the alphabet, etc), much
>in the same way as a given input to a computer (a sequence of 0's and 1's)
Of course this capacity requires changes in the brain. But the level of
brain science will not explain what it is that is being *detected* -- that
depends on history. In one case, it is English letters, in another
Martian say.
>can rtesult in a different output, depending on what program is runing in
>the computer. You are not hopefully saying that a given sequence of 0's and
>1's contains in it all posible outputs which it can lead to with all possible
>programs which can take this input, are you?
Sure, but we are not really interested in the description in terms of
1's and 0s'. That is a non-semantic description -- it does not tell you
what the things stand for.
We are interested in a different level of description,
the level of "detecting information about words [or letters, or
tables]". That is a semantic level of description. It concerns what these
things mean.
>Now, unless you can provide a counterargument making sense, please do not
>repeat the same claim. As far as above goes, it does not hold water.
I don't know what I can say, since you have not raised any real problem that
I can see. I take it there are historically determined facts. Facts about
the semantics of your mental states or recognitional capacities would seem
to be among them.
You seem not to have heard of "externalism" even though it has been much
discussed in philosophy of mind since Putnam's Twin Earth hypothetical.
I am an "externalist" -- I think the semantic facts about mental states
are not determined solely by what's in your brain, narrowly described, but
also by the material and social environments. This is not to say there
aren't non-semantic levels of description you can also use. It has
to do with the way the semantic attributes supervene on the nonsemantic
attributes.
We had a similar case illustrating the idea of one layer of fact
supervening upon on another in the case of the businessman realizing
two different linguistic moves with a single noise-making.
>>In this way I do not in any way *infer* this content from
>>representations of patterns on my retina -- I certainly have no
>>conscious representations of patterns on my retina, so I could not
>>perform any such inference. Rather I am now able to *extract* the
>>information the information about the world that is already (I know not
>>how) coded there.
>>
>As far as I can see, the problem here is that you never seem to consider
>what is this "I" above you are talking of, objectively speaking. This
I take the subject of psychological predicates to be a living human being
or what resembles (behaves like) one. Mostly I am thinking of persons,
animals that have developed certain innate potentialities for rational
thought, action, and self-consciousness through the acquisition of
language.
>is why I )and others) see as dualist, because you refuse to make a connections
>of this "I" with the physical system which a human is. Hiding behind
Nonsense. The "I" just *is* the system which a living human being is, so
the connection is there staring you in the face.
Suppose someone were to say: you never make the connection between the
piece of wood and the pawn in a chess game, therefore you are a dualist.
We can explain what how the piece of wood comes to constitute a pawn, but
not if we look only to physical science attributes of the wood.
Similarly we can explain how a person comes to be an understanding entity,
but not in terms of neural -- or even neuro-computational -- properties
of the brain. Rather in terms that make reference to of the social
normative context.
>philosophical jargon like "epistemic subject" or like is ducking the issue.
I do not see an issue. Descartes held that there were two kinds of *substances*
(objects, entities), mental and physical. I emphatically do not believe that.
I have only one substance, the living human being. Perhaps there is a dualism
of predicates or attributes. But that should be no more problematic in the
case of persons then in the case of chess pieces or money.
>Now you are hiding behind technical difficulties. If you are not saying
>that such thing is in principle impossible, why raise the issue?
Well it might be important to note that the only way to get the
complexity is with a real world. Just so we don't exaggerate the
threat, if nothing else.
>Consider only an audio input. You can have a full illusion of being in
>front of a waterfall (with your eyes closed) even if the audio inputs which
>make you to.think so would be produced by a couple of well placed
>loudspeakers in an empty room.
OK: now think about the conditions under which such a state could
*mean* "that's a waterfall". Could a brain that was always in a vat
have a state that means "that's a waterfall"? What would determine that
meaning as opposed to something else? The whole idea seems ludicrously
far-fetched to me -- after all, to "decode" states of the brain, you
always need to look to its history and the normal environment.
>>But in any case I am talking about perception of the real reality, for
>>example my office. If you are in a virtual reality then you are not
>>really perceiving anything, you only seem to be. I don't recall
>>saying that one could always tell the difference.
>>
>But if you can't tell the difference, what sense does it make to claim that
>we see (or hear) directly the objects of the world?? Our brain/mind responds
Why not? Suppose John is in a house of mirrors and is unsure whether
he is seeing someone in front of him or merely seeming to. Still we can say:
he is in fact seeing someone in front of him, even if he cannot tell this
for sure.
(And we don't particularly give a damn about his "mind/brain" when we say this.)
>then only to the physical inputs, and not the "real" objects of the world.
>So what sense does your claims make? It looks like they are quasi-religious
>beliefs, and are not subject to logical or empirical verification. If you
>disagree, please explain what logical sense do your claims above make?
I am just talking about what we say when we talk about seeing in ordinary
language. We do not really talk about the "mind-brain" responding to
physical inputs at all, we talk about persons who can form and express
judgements about the world around them.
>And how does the above answer the basic question: if we can be conscious of
>a waterfall even if we only receive sounds from a couple of loudspeakers in
You cannot be conscious of a waterfall in that case -- you can only seem to
be. But there is no answer to the question "which waterfall are you conscious
of" in such a situtation.
As I use the term, to be "conscious of x" requires that you are
actually aware of a real x.
>an otherwise empty room, what sense does it make to claim that we are
>"directly conscious of the world"?
When someone is really hearing an actual waterfall in the world, the
logic is different, even if you yourself can't tell the difference
at the time. I can ask which particular waterfall you are hearing and
get an answer -- e.g. "Niagara".
>..............
>>Well sure, if you are really seeing a wastebasket then there really is
>>a thing you see, for example, and your experience is in some way
>>responsive to *it*, the real wastebasket which anyone else can see and
>>
>You seem to overlook the fact that you are relying on the presence of
>an "external" arbiter who knows whether there is a 'real" wastebasket or
>not. Where (and/or who) is this arbiter to tell us that there really is
So you think I am going out on a limb when I say there is a wastebasket
in an office? Maybe that's a little too risky? But most of us can tell a
wastebasket when we see one, and if there are errors, they are routinely
corrected by ordinary tests. Doubt and testing come to an end somewhere.
I am very surprised you think I ought to be a skeptic about the
existence of wastebaskets. That seems to be a very odd stance for a
scientist to take. After all, scientists do measurements and
experiments in laboratories all the time and expect to be able to communicate
and reproduce each other's results.
What if you were to object to an experimental team wishing to publish their
results in Science: and who was the arbiter who guaranteed there
actually was a laboratory and voltmeter or test tube and lab notebook
and process of measurement and recording and checking of results?
Before you can report your results for other scientists to try to reproduce,
you *really* ought to first establish that there is an external world in
which there *are* other scientists and laboratory equipment at all who
occupy a common space and that you can manage to communicate such results
to other people using language in a way that they can test them. Please
resubmit when you have furnished adequate support for these crucial
hypotheses.
I hardly think many practicing scientists would take you seriously.
There is simply no role at all for supposedly private subjective
experience in science, it is essentially an intersubjective process
that rests on the constitutive presupposition of the possibility of
communicating shareable contents about a common world. What is
completely private and incommunicable is of no concern to
intersubjective science. And anyway it would be downright silly for
*Science* to demand such steps, since any such skepticism would also
demolish any rationale for even having such a journal as Science. If
one could never be sure of a common world including other scientists,
there would no point trying to communicate with them. One ought really
just to retreat to solipsistic nirvana and soak one's consciousness
mindlessly in the sensory stream.
>a wastebasket there and not something very different which affects all our
>senses in a way which we interpret as a wastebasket in a analogous (but very
>limited) way to the way the sounds form the loudspeakers in the example above
>persuaded us that we are in front of a waterfall ? You are clearly assuming
You would need to provide some grounds for this hypothesis, I think.
And I am wondering why most scientists don't trouble to work to rule out such
idle hypotheses before they publish their results.
>that there exists, what is termed by some people "a God's eye view". Apart
No, I am talking only about common-sense wastebaskets as they figure as
paraphernalia in ordinary human life. That is, as they appear to us
from our own point of view as people who have a need to throw out
the trash.
If I ask the maintenance staff when the wastebaskets are emptied, I do
not often meet the reply "How should I know, I don't have a god's eye view
and so can not be sure there are wastebaskets". Unless you think that
would be a perfectly sensible reply to my inquiry to the staff, I would
say it is you that are making meaningless claims. For me, a word-form
is shown meaningful if we can imagine some ordinary context in which it
can be significantly used.
You seem to be forgetting that what show up from the ordinary human
point of view are *not* mainly subjective sense data, but rather
such things as meaningful paraphernalia -- elements that have some function
in our everyday goal-directed pursuits -- in the common world.
>Fine, but if you do this you will be unable to understand/explain
>a relationship of this "conscious subject" to the physical brain. By refusing
I guess I think we don't really have to explain this relationship beyond
noting that the brain is necessary for the person to have psychological
states, and that malfunctions in the brain can causally interfere with
the thought of the person.
But I wonder why you don't worry that the *brain* you seem to see when you
look at certain scans or other devices is just an illusion. After all,
you seem to be mighty skeptical of the wastebaskets. How can you then go
on to help yourself to the real existence of brains? Brains are surely the
same sort of public spatial object as wastebaskets; in fact, most of us
knew about wastebaskets long before we ever learned about the brains in
our heads.
Do you have a "god's eye view" from which you have verified that there
really are brains in physical space (while wastebaskets are risky
hypotheses)? Don't you really only know that there are your own private
sense-data? Or course if you do some experimentation or learning it may
*seem* to you that there are brains, and it may appear (in the virtual
reality you have always been in) that these appearances called brains
are causally linked to your own private experiences. But after all, if
you are worried that the whole objective world for you is a virtual
reality, surely you ought to worry that you might also be in gross
error about the causal relevance of *brains* to your experiences -- for
that is something you only verify them to have as phenomena discovered
*inside* your hallucination (virtual reality), not in reality as it is
in itself.
Since I am not worried about skepticism, I don't have that problem.
More seriously, I take it there are many possible ways of explaining
the relationship between the brain and psychological capacities. Some
of them posit only very "weak" relationships, such as I alluded to as
"supervenience" (which has been discussed in philosophy). You seem to
be presupposing a very particular strong relationship according to
which the brain *must* somehow furnish *outputs* which then serve as
*inputs* to the thought of a conscious phenomenological subject. I see
no reason why an acknowledgement that the brain is causally necessary
for the person to perceive means that the brain furnishes *inputs* to
the person.
>to see such relationship you are falling into the trap of believing in
>"direct seeing" instead of realising that this "seeing" may come from
>transformations of physical inputs by subconscious processes in the brain,
You know, in some sense of these words I accept this. The question is
whether the fact that Aristotle is seeing a Greek lambda could be explained
by properties of the brain alone. I think, clearly no, this is a historical
fact that depends on the existence of institutionalized functional roles
of the marks.
>>As to "mind" in a way I would say there is no such thing.
>
>Well, if so what is this "we" or "I"?
We are human animals that have developed a set of psychological
capactities, e.g. the capacity to see, to interact with objects in the
world, also to think, plan, infer, imagine, remember, feel sensations
and emotions and the like. But we are not ourselves "minds", if that
means we are things that float like ghosts apart from our bodies. Nor
do we *have* minds as parts or subsystems. The thing that thinks, or
plans, or feels an itch is the very same entity as the one that walks
or waves and says "hello" or scratches its arm or cries out in agony.
(One can of course say that someone has a quick or slow mind, but that
is completely intelligible in Rylean terms, as referring to various
broadly dispositional traits.)
Anders N Weinstein
In article <33408C...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>Anders N Weinstein wrote:
>
>> I mean only
>> this: Sloman's careful critique does not rest on charging Searle with
>> committing a trivial logical fallacy, as you frequently do.
>
>the argument; Sloman has other interests. He may not even be aware of
>the argument, because it is already clear to him that Searle's argument
>fails miserably on other grounds, grounds more interesting to Sloman.
I remind you the Sloman diagnosis does also suggest that virtually all
the interlocutors in the original BBS review missed the crucial
ambiguity he finds.
And don't Sloman's constraints on implementation on the "reasonable"
interpretation thesis of "Weak Strong AI" have -- or at least permit -- as
a consequence the premise Searle uses, that no understanding entity is
constituted in the case where a man hand-simulates a program in the
room?
>> Searle's argument may be unsuccessful. I just think that to suggest the
>> errors are *trivial* is grossly unfair.
>
>That doesn't change the fact that they are trivial. The ad reductio
>premise of his scenario is that his mind works on computational
>principles. His argument fails to test that premise; it fails
More precisely, it is that an entity could have intentional states,
including those (Searle takes to be) required for understanding a
language, *in virtue of* implementing or instantiating the right
program. Searle's own view is compatible with the idea that the course
of semantic or intentional mental states and operations could be
*modelled* by a simulation in a computer program. Also with the idea
that the operations of the mind might also have a true description as a
computer program, just as the digestive operations in the stomach do.
Anders N Weinstein
>premise*. *If* there is no understanding present, *then* "the alleged
>mental state", namely "e.g. can understand symbols, or use them to refer
>to other things" (from earlier in Sloman's paper) "does not exist", and
>so C(x) isn't sufficient to produce M(x). Like, duh. The question then
Note my own view is in partial agreement with you. I agree that in some
versions of the Robot Reply there could be an entity other than Searle that
understands Chinese, namely the Robot (call him Robbie). I agree with
you that Searle's appeals do fail to tell against this possibility.
However as I understand these matters, this is in fact little
consoloation to defenders of Strong AI -- it is rather an abandonment
of Strong AI. For as I see it the Robot that understands in that case
is not a computer, nor a program, nor a running instance of a program
(a computational process). Better, perhaps, since you can always insist
on describing it in any of these terms, is not taken to understand in
virtue of implementing or instantiating a computer program (although it
might have parts inside it that do this).
I think if anything Robbie understands in the right sort of case in
virtue of a capacity for being-in-the-world, where that is, if you
like, a kind of emergent level of description that has little to do
with computation and everything to do with cultivated molar capacities
for interacting with things in practical contexts, and expressing
opinions, giving arguments, reporting silent thoughts or imaginings and
the like.
BTW I think the SR is also little consolation to defenders of
materialist or physicalist ontology, since it seems to rest on the idea
of emergent entities called Systems that are not wholly individuated by
the physical science descriptions of their material parts. I think this
is no real problem, but then I think physicalism is nearly worthless.
In any case I think it is quite worthwhile to explore the metaphysics
implicit in the Systems Reply. To me it just seems to lend ssupport to
a belief in emergent properties of wholes -- Aristotlean forms, if you
like -- that are irreducible to those of the basic physical science
explanations of the parts.
However, I don't think Searle's argument is quite so *blatantly* stupid
as you suggest. For Searle's rejection of the idea that understanding
is constituted in the case is not a totally groundless assumption after
all. It trades, first, on the idea that from his perspective, he Searle
does not come to understand Chinese by virtue of being given explicit
rules to follow -- from his perspective the symbols are meaningless
bits of type. That is not unreasonable, especially if you think his
perspective is simply different than that of the understanding entity.
But it also stems, second, from his conviction that he Searle is the
thing that implements or instantiates a computer program. For remember
the thesis was that something understands in virtue of implementing a
program. Searle is taking it that he, Searle, must be the subject of
this predicate in the case.
I agree this is not right, but it does call for more inquiry into the
concept of "instantiation". After all, Searle is using one somewhat
fair interpretation of what it means to be a computer, an
interpretation that is linked to a definite ordinary sense of
computation in which computation is always something done by an *agent*
who is actively manipulating or operating with symbols, e.g. doing long
division. This is behind the common enough idea of computers as
automatic devices for doing the same thing, replacing the person that
manipulates symbols.
There is of course a somewhat different conception according to which an
automatic computing device should really not be spoken of as an agent
of any kind. Rather it is a medium in which symbols get
manipulated without a manipulator. I recall Hofstadter pushing this line.
On this conception it is a mistake to identify the subject that understands
with the agency that manipulates symbols -- for there just is none.
One wonders if this means that on the computational philosophy there
really is no self or subject of mental states -- as some philosophers
have independently held. And yet here is the Systems Reply insisting
that the subject is the System, whatever, exactly, that entity is.
Moreover, on the Robot-Systems Reply, the thing that comes to
understand Chinese -- Robbie, say -- is it would seem, an entity that
is simply *never* conscious of the symbols inside the room, just as you
are not conscious of the symbols that might be discovered to be inside
your head. So in what sense should we say that *that* entity really
does manipulate or compute with *those* symbols? Again, it seems to
me that the Systems Reply (Robot version) is, properly viewed, no real
consolation to the (to me, bizarre) idea that *computers* might think.
Brains don't think, people do.
>becomes, under what conditions is there understanding present. That's
>beyond Searle; he thinks it is obvious there isn't, and Sloman points
>out why he thinks that Searle thinks that: because static structures
>don't have the required causal powers. Sloman then goes on, way beyond
>Searle, to discuss other forms of Strong AI, non-static forms, that
>escape this obviousness. Which not everyone finds obvious BTW, and in
>fact Sloman agreed here recently that it was possible to extend
>"understanding" in such a way as to apply to static structures, in which
>case even Strong Strong AI might be true; I think that such extension
>must be granted, because a space-time description of causal
>relationships is static.
Could you explain what you mean by "static structures"? The symbols in
the CR seem to be dynamic insofar as Searle is manipulating them, every bit
as dynamic as the symbols inside a STRIPS-governed robot, say.
>In any case, Searle thinks he has shown that there is no understanding
>in the room simply by pointing at the man in the room. The Systems
>Reply is that his argument simply hasn't looked for understanding where
>the functionalists claim it is to be found. Rather than accept the
>burden of constructing an argument that the room as a whole cannot
>understand, Searle crams the room into his head and says "see, I still
>don't understand Chinese". But this is mere assertion and sloppy
I think he can (somewhat charitably) be viewed as resting on a sort of
materialist premise: if two entities have all their material parts in
common they are the identical. And trying to construct a case in which
he and the System have all the material parts in common ("there is
nothing to the system other than me", or something like that). Of
course this is trickier than Searle allows -- when he memorizes the
program and steps through it in his imagination, for example, the
System might with justice be taken to be something *smaller* than
Searle, rather than something larger, so not materially coincident.
In any case, I think the premise is just false. But an awful lot of
self-styled physicalists seem to find its falsity deeply problematic.
As usual, I think the thought experiment raises very interesting issues
about the metaphysics of material substances and the subjects of
psychological predicates. True these are not fully explored. I think
that if properly followed out, they lead to a rather non-physicalist
metaphysics. As for example in the neo-Aristotlian view in Wiggins,
_Sameness and Substance_. Yet proponents of the SR often take
themselves at the same time to be making the world safe for something
like the supposed modern physicalist world-view.
>thinking. Suppose he had crammed the brain of a Chinese person into his
>head; what then? Without the proper connections, he would fail to be
>able to satisfy the requirements of being able to converse in Chinese.
>Make those connections, and now we have Searle walking around, talking
>Chinese, and perhaps denying (in English) that he understands it. but
Because I distinguish between the sub-personal representations and the
person-level states, I actually think the case is woefully
underdescribed. For example, the sub-personal states in your brain
might be said to deal with, say, raw acoustic input. The computations
would have to parse or segment it into words and phrases. But you the
conscious subject cannot obviously do this. In general you might find
it extremely difficult or even impossible to become aware of the
low-level acoustic properties of the stimulus pattern. At the person
level, you often just hear meanings directly without inference or
computation.
Similarly in the case of vision -- the visual system might do
computations that deal with two images, one from each retina. But the
person might not be able to become consicous of the two images as such
and so simply could not perform these computations.
So I don't know what it means exactly for Searle himself to internalize
the whole program for a sub-personal Chinese control system in his own
person. It would mean he would have to recover the low-level acoustic
representations at his eardrum and step through the computations done
by a sub-personal sentence parser. Similarly for low-level visual input
and the computations done by the visual system; e.g. he would have to
find out about the *TWO* images on each retina and imaginatively step
through a visual system algorithm, maybe a pseudo-parallel
instantiation of Marr and Poggio's relaxation algorithm, to take a
concrete example.
There is no real reason to suppose Searle himself could recreate all
this sub-personal detail simply because he can himself see or hear
Chinese utterances or marks or see recognizable objects in the world.
I guess it is not impossible in principle. But I do think it is worth
noting that he is being rather hasty in the description of the case.
Again, this worry flows naturally once we distinguish sharply between
the sub-personal input to a computational sub-system in the brain from
the input information to a conscious person in the world.
>be far from obvious that such a denial was valid. But Searle and those
>who appreciate his having presented an argument that seems to support
>their prior biases are too intellectually dishonest to actually
>explore these issues. Rather, Searle just says that it is ridiculous
I am very happy to explore them, in particular the question about what
the subject of psychological predicates is supposed to be, according to
the computational philosophy (which seems to me to be the real issue).
I have given a few general considerations about substance and the
subject of psychological attributions that I think allow us to make
good sense of the case -- but according to which the thing that thinks
is not the thing that computes.
So I think they have as a natural but not really worrisome consequence
that Strong AI is just false under any interpretation, mainly a kind of
category mistake. Since intentional psychological attributes are,
properly viewed, "emergent" attributes of whole living organisms in an
environment; in particular, of human beings in the world.
>to imagine that he together with "bits of paper" could understand
>Chinese
>when he alone could not, which is just another *assertion* that programs
>aren't relevant to understanding, and such argument by assertion is
>*trivially* fallacious.
I think Searle relies on some somewhat slightly plausible grounds for
this, e.g. the materialist composition principle (in the case in which
he internalizes the system), the explicable but ultimately inadequate
conception of a computer as a symbol manipulating *agency*, and the
claim that understanding requires phenomenology which would be absent
in the case in which the conscious perspective of the agent of the
computation consists of contents presenting the symbols only as formal
or meaningless entities.
I take it to point out important features about the need to distinguish
between different perspectives, on the one hand that of the conscious
person in the world and on the other the supposed perspective of the
alien thing, whatever it is, that might be held to compute or manipulate
symbols or other representations inside your brain.
andrew babian
>> careful analysis up at all.
>
>You dishonest git.
Woah. now there's an angle. Do we actually see dishonesty on this
subject? Then what is the purpose? For there are some good reasons to
confuse an issue.
If you're goal is to be able to write endlessly about something without
ever having to come to any conclusion or produce any results, dishonesty
is your ticket. Hey, isn't that the whole point of philosophy? Using
loose terminology and switching definitions around is good too. You can
write hundreds of lines without resorting to cut-and-paste.
Its quite possible that someone might have many reasons for not believing
in the possibility of ai but that it is easier to simply cloud the issue
than defend them.
I mean, i assumed that "comp.ai.philosophy" would be about general
theories about putting artificial intelligence on computers, as opposed
to vacuous arguments that it can't be done. If all we want is smoke,
then by all means lets bring on the alien religions. At least it gets
deflected from comp.ai.
Anders N Weinstein
In article <E7z8v...@cruzio.com>, andrew babian <aba...@cruzio.com> wrote:
>I mean, i assumed that "comp.ai.philosophy" would be about general
>theories about putting artificial intelligence on computers, as opposed
But why must it be on computers, as opposed to, say, in robots? There
are deep philosophical motives for this conviction, having to do with
the idea of an interface between the separated reasoning agent and the
world.
I wonder why you think a group with "philosophy" in its name would be
free of, well, philosophy. With all the amorphousness of subject
matter, method, and what counts as a "result" that goes with the
discipline.
Neil Rickert
> I agree that in some
>versions of the Robot Reply there could be an entity other than Searle that
>understands Chinese, namely the Robot (call him Robbie). I agree with
>you that Searle's appeals do fail to tell against this possibility.
>However as I understand these matters, this is in fact little
>consoloation to defenders of Strong AI -- it is rather an abandonment
>of Strong AI.
Perhaps it is an abandonment of a strawman version of Strong AI, and
I will admit that there are times when I think Searle was merely
arguing against such a strawman.
> For as I see it the Robot that understands in that case
>is not a computer, nor a program, nor a running instance of a program
>(a computational process).
Then I guess that the 'computer' on my desk is not a computer, nor a
program, nor a running instance of a program (a computational
process). But whatever the defenders of strong AI mean by
"computer", most of them do intend the term to cover the machine on
my desk, and to exclude that wall in Searle's office which Searle
points to as an instantiation of the WordStar program.
>BTW I think the SR is also little consolation to defenders of
>materialist or physicalist ontology, since it seems to rest on the idea
>of emergent entities called Systems that are not wholly individuated by
>the physical science descriptions of their material parts.
The SR is not intended to offer consolation to anyone. The purpose
of the SR is to point out a gaping hole in Searle's argument. You
seem to be suggesting that any materialist reduction of Robbie would
necessarily include a step labelled "... and then a miracle
happened." But there is no need for you to impose your own confusion,
and your own dogmatic anti-reductionism on those who don't much care
for your way of looking at cognition.
> I think this
>is no real problem, but then I think physicalism is nearly worthless.
Presumably physicalism does not matter to solipsists.
>In any case I think it is quite worthwhile to explore the metaphysics
>implicit in the Systems Reply.
By contrast, it is metaphysics that I think of as nearly worthless.
> To me it just seems to lend ssupport to
>a belief in emergent properties of wholes -- Aristotlean forms, if you
>like -- that are irreducible to those of the basic physical science
>explanations of the parts.
That is where your "... and then a miracle happened" step shows
up. As far as I can see, that step arises out of your confusion and
your very deliberate but dogmatic rejection of any possible
reduction.
>However, I don't think Searle's argument is quite so *blatantly* stupid
>as you suggest. For Searle's rejection of the idea that understanding
>is constituted in the case is not a totally groundless assumption after
>all. It trades, first, on the idea that from his perspective, he Searle
>does not come to understand Chinese by virtue of being given explicit
>rules to follow -- from his perspective the symbols are meaningless
>bits of type. That is not unreasonable, especially if you think his
>perspective is simply different than that of the understanding entity.
My automobile is not following explicit rules about pistons and spark
plugs. Nevertheless there are components of my automobile which do
follow such rules, and my automobile would not function without that
rule following. It is a category mistake to ascribe such rule
following to the automobile as a whole. It seems to me that you are
suggesting that Searle was making a similar category mistake.
Personally I prefer to take a more charitable view and assume that
Searle was not making such a blatant category mistake.
>One wonders if this means that on the computational philosophy there
>really is no self or subject of mental states -- as some philosophers
>have independently held. And yet here is the Systems Reply insisting
>that the subject is the System, whatever, exactly, that entity is.
Again, that seems to be based on a misunderstanding of the Systems
Reply. If Searle were to announce that there are no such things as
mental states, most proponents of the Systems Reply would take this
admission a supporting their position against the validity of the CR
argument.
>Moreover, on the Robot-Systems Reply, the thing that comes to
>understand Chinese -- Robbie, say -- is it would seem, an entity that
>is simply *never* conscious of the symbols inside the room, just as you
>are not conscious of the symbols that might be discovered to be inside
>your head. So in what sense should we say that *that* entity really
>does manipulate or compute with *those* symbols?
This seems to be at the heart of your confusion, and Searle's
confusion. There is no need for any symbols that might be found in
my brain to be objects of my consciousness. Nor is there any need
for symbols processed by the CR to be the symbols processed by the
computational processes going on inside the CR. It might be that the
rules of the CR program call for making many measurements of the
source Chinese characters. Thereafter, the rule following processes
might deal only with those measurements. It is not necessary that
the rule following processes treat the Chinese characters as
symbols. These computational processes might use computations with
measurements to produce markings on paper, which only the CR as a
whole would interpret as symbols.
Analogously, the rule following in the automobile engine has to do
with pistons and spark plugs. The effect for the automobile as a
whole has to do with roads and transportation. There is no need for
the engine to deal with roads and transportation, and there is no
need for the automobile to follow rules about spark plugs and
pistons.
Neil Rickert
In <5hs4hs$m...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <E7z8v...@cruzio.com>, andrew babian <aba...@cruzio.com> wrote:
>>I mean, i assumed that "comp.ai.philosophy" would be about general
>>theories about putting artificial intelligence on computers, as opposed
>But why must it be on computers, as opposed to, say, in robots?
There is no need to treat computers as opposed to robots. The
distinction between them, and the question of whether there is an
important distinction, should be allowable topics of discussion.
> There
>are deep philosophical motives for this conviction, having to do with
>the idea of an interface between the separated reasoning agent and the
>world.
I suspect that Andrew would prefer the discussion to be driven by
deep scientific motives rather than by deep philosophical motives.
>I wonder why you think a group with "philosophy" in its name would be
>free of, well, philosophy.
I didn't take Andrew as suggesting that the group be free of
philosophy. What he objects to, I take it, is free philosophy rather
than philosophy directed toward the problems of AI.
Jim Balter
andrew babian wrote:
> I mean, i assumed that "comp.ai.philosophy" would be about general
> theories about putting artificial intelligence on computers, as opposed
> to vacuous arguments that it can't be done.
Yes, exactly, which is why I come down so hard on people like
Anders Weinstein, Phil Roberts, R. Alan Squires, and Searle, whose
destructive ideologies coupled with shoddy logic disrupt any valid
discussion, just as the creationists disrupt discussion in talk.origins.
Their sophistry drives out people like McCarthy, Minsky, Moravec,
McCullough, McDermott, etc. who actually have intelligent things to say
that could move us forward. Fortunately, those people are all busy
doing useful work elsewhere.
--
<J Q B>
Jim Balter
Neil Rickert wrote:
>
> In <5hs4hs$m...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>
> >>I mean, i assumed that "comp.ai.philosophy" would be about general
> >>theories about putting artificial intelligence on computers, as opposed
>
>
> There is no need to treat computers as opposed to robots. The
> distinction between them, and the question of whether there is an
> important distinction, should be allowable topics of discussion.
Anders is being his usual dishonest Martian self. No one says
it must be on computers; rather, Anders says it must *not* be on
computers.
> > There
> >are deep philosophical motives for this conviction, having to do with
> >the idea of an interface between the separated reasoning agent and the
> >world.
>
> I suspect that Andrew would prefer the discussion to be driven by
> deep scientific motives rather than by deep philosophical motives.
Yes, there's little to distinguish "deep philosophical motives"
as Anders uses the phrase from "deeply held prejudices".
Convictions are the tools of people afraid of being wrong.
> >I wonder why you think a group with "philosophy" in its name would be
> >free of, well, philosophy.
>
> I didn't take Andrew as suggesting that the group be free of
> philosophy. What he objects to, I take it, is free philosophy rather
> than philosophy directed toward the problems of AI.
Anders is being his usual dishonest or forgetful self, since Aaron
Sloman recently quite clearly pointed out the distinction between
constructive and useful philosophy and the sort of destructive
dogmatism that Anders practices.
--
<J Q B>
Jim Balter
Anders N Weinstein wrote:
>
> In article <333E0C...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
> >So here we have Aaron rather vigorously DENYING that he has shown that a
> >CPU chip does NOT understand Chinese. Weinstein has some balls claiming
> >that Aaron "freely concedes the validity of Searle's argument".
>
> on Searle's argument over the years. In taking apart a complex issue
> one can focus on many different unclarities, or different lines of
> possible response.
>
> I was referring explicitly to Sloman's 1985 paper on Searle -- I recall
> the title as something like "Did Searle refute Strong Strong AI or Weak
> Strong AI?"
This shows how pathetically dishonest and ideologically driven you are,
since the title is "... attack ...", not "refute". You just replaced
"attack" with "refute" because you are an intellectually dishonest git
who uses shoddily inferior intellectual methods. While your fantasized
title supports your claim, the actual title undermines it.
> (a copy is on his web site). If you read it, I think you
> will find it about as unambiguous as anything in philosophy ever is
> that that paper conceded the validity of the Searle argument *as an
> argument against "Weak Strong AI"*.
Aside from your (subconscious?) mistake in switching "Weak Strong AI"
for "Strong Strong AI", this is false. He never concedes any such
thing, certainly not unambiguously. An *honest* reading of the paper
reveals that Sloman concludes that Strong Strong AI is "absurd"; this is
based upon Sloman's own argument, not Searle's.
" What's Right in Searle's Argument
I conclude that Searle is right to suggest that, as far as our
ordinary concepts of calculating and understanding are concerned, not
just any production of patterns fitting an (appropriate) program
specification is enough to produce understanding. He is right to
suggest that the processes must be produced by a system with the right
causal powers. This is a requirement not only for mental processes,
but for any computational process which is used as a reliable basis
for taking decisions or controlling anything. The mere fact that a
pattern of leaves or water molecules happens to solve a mathematical
problem, does not make either the leaves or the water into a
calculator."
Since Sloman only mentions things that Searle has *suggested* among what
is right with his argument, it should certainly follow by your own
Martian thinking that he doesn't consider anything else to be right
about it, including its logic. But in fact Sloman considers the logic
of an argument from *the premise* that the room as a whole doesn't
understand Chinese to the conclusion that the room doesn't possess the
mental states associated with understanding to be "unassailable". The
only problem is that Searle would not agree that he took that as a
premise, but rather that he showed it.
--
<J Q B>