Searles's Chinese Room
R Jones
It would seem to me that a book of rules having the capability
that Searle assumes in his Chinese room argument would, itself,
be intelligent. No book with such capabilities exists. It would
be a traditional production system.
R. Jones
Burt Voorhees
R Jones <jone...@esumail.emporia.edu> writes:
But the book of rules does not have any capability other than that
of being a book. Rather, it is Searle, inside the room (Hum, isn't
this like a Farside cartoon of Hell: Just stay in this room for the
next billion years and process chinese characters) who gives the
room the capacity to appear, from the outside, as if there was
an understanding of chinese going on inside.
bv
Oliver Sparrow
The room's a ROM. No one expects intelligent ROMs.
No one expects a program that looks like a ROM - packed
with declarative statements with no underpinning connectivity -
to be intelligent in appreanace or actuality. Searle argues
against a seventies model of AI.
_________________________________________________
Oliver Sparrow
oh...@chatham.demon.co.uk
David Longley
In article <367018...@chatham.demon.co.uk>
oh...@chatham.demon.co.uk "Oliver Sparrow" writes:
And as such, Searle just offered another variant of the classic
'free-will vs determinism' argument, couched in words which were
familiar to one generation and not to another. As I have said at
length here to no avail, such issues have nothing to do with the
real problems of AI, but such are the vicissitudes of intensional
contexts.....
--
David Longley
Jim Balter
Oliver Sparrow wrote:
>
> The room's a ROM. No one expects intelligent ROMs.
> No one expects a program that looks like a ROM - packed
> with declarative statements with no underpinning connectivity -
> to be intelligent in appreanace or actuality. Searle argues
> against a seventies model of AI.
Wrong. As far as Searle is concerned, you can give the room dynamic
memory, arms, legs, a mouth, a body, all of that, as he makes clear
in his response to the Robot reply, and it still won't "understand"
merely by virtue of "manipulating formal symbols".
The debate about the CR certainly won't be furthered by continually
missing the point of what it *purports* to show.
--
<J Q B>
Robin Faichney
Oliver Sparrow wrote:
>
> The room's a ROM. No one expects intelligent ROMs.
> No one expects a program that looks like a ROM - packed
> with declarative statements with no underpinning connectivity -
> to be intelligent in appreanace or actuality. Searle argues
> against a seventies model of AI.
As I understand it, the "underpinning connectivity" is
implicit in the rules. After all, that the CR behaves
as if understanding is one of the premises. Which
seems to imply that intelligence is granted.
--
Robin Faichney
r.j.fa...@stirling.ac.uk (delete ".nospam" to reply to me)
http://www.stir.ac.uk/envsci/staff/rjf1/
Oliver Sparrow
Jim Balter <j...@netcom.com> writes:
> Wrong. As far as Searle is concerned, you can give the room dynamic
> memory, arms, legs, a mouth, a body, all of that, as he makes clear
> in his response to the Robot reply, and it still won't "understand"
> merely by virtue of "manipulating formal symbols".
My point and Searle's: a look up table, with or without actuators,
will not 'understand'. True, O! Wise One: none shall disagree. But
who says that an AI is a look up table? Searle. Not me.
The CR can be reduced to the statement that a room which contains no
understanding cannot be expected to exhibit this property; and a room
with it, can. This is what we in the trade call a 'tautology',
however. Searles insensate room is the propositional equivalent of a
string of void parentheses: ((((())))). Om.
> The debate about the CR certainly won't be furthered by continually
> missing the point of what it *purports* to show.
Indeed it will not, Jim. Indeed so.
_________________________________________________
Oliver Sparrow
oh...@chatham.demon.co.uk
Jim Balter
Robin Faichney wrote:
>
> Oliver Sparrow wrote:
> >
> > The room's a ROM. No one expects intelligent ROMs.
> > No one expects a program that looks like a ROM - packed
> > with declarative statements with no underpinning connectivity -
> > to be intelligent in appreanace or actuality. Searle argues
> > against a seventies model of AI.
>
> As I understand it, the "underpinning connectivity" is
> implicit in the rules. After all, that the CR behaves
> as if understanding is one of the premises. Which
> seems to imply that intelligence is granted.
Quite; the Sparrow pecks at a straw Searle. Intelligence is granted,
understanding is not.
--
<J Q B>
Gary Forbis
It is quite common for arguments to take the form:
A premise
B premise
...
(A and B) => (~A or ~B) some rule of logic
therefore ~A or ~B
or
therefore A => ~B
The degreee to which one holds A or B determines ones prejudices
towards ~A or towards ~B. I think most hold a rock doesn't have
to understand the laws of physics in order to obey them.
Both humans and machines are obeying the laws of physics. Machines
have a further restriction in that in order to be labeled as properly
functioning they must follow the human model of physics.
Layered on top of physics is our model of understanding. We believe
humans can understand. Some of us believe machines following the
rules of our model of understanding can understand in the same way
we understand.
Goedel came up with his Incompleteness Theorem.
Many people claim to understand integer arithmatic even though
the Incompleteness Theorem tells us any formal model we might
concieve of integer arithmatic is either inconsistent or incomplete.
I am lead to wonder how anyone can believe both that they understand
integer arithmatic and that a machine understands the same way they
do merely (and I mean *mearly*) by way of following the rules of
a formal model of understanding..
--
--gary
for...@accessone.com
Jim Balter
Oliver Sparrow wrote:
>
> Jim Balter <j...@netcom.com> writes:
>
> > Wrong. As far as Searle is concerned, you can give the room dynamic
> > memory, arms, legs, a mouth, a body, all of that, as he makes clear
> > in his response to the Robot reply, and it still won't "understand"
> > merely by virtue of "manipulating formal symbols".
>
> My point and Searle's: a look up table, with or without actuators,
> will not 'understand'. True, O! Wise One: none shall disagree. But
> who says that an AI is a look up table? Searle. Not me.
Nowhere does Searle say that the CR contains a lookup table,
unless you consider a Turing machine to be a lookup table,
which would make you a computational Humpty Dumpty.
> The CR can be reduced to the statement that a room which contains no
> understanding cannot be expected to exhibit this property; and a room
> with it, can.
Well certainly it seems to me that Searle's argument can be reduced to
that, but Searle would disagree. In any case, that has nothing
to do with lookup tables.
> This is what we in the trade call a 'tautology',
> however. Searles insensate room is the propositional equivalent of a
> string of void parentheses: ((((())))). Om.
Searle's room is not insensate; it can sense questions written in
Chinese.
> > The debate about the CR certainly won't be furthered by continually
> > missing the point of what it *purports* to show.
>
> Indeed it will not, Jim. Indeed so.
So stop doing it.
--
<J Q B>
Neil Rickert
In <334F32...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>Jim Balter wrote:
>Layered on top of physics is our model of understanding. We believe
>humans can understand. Some of us believe machines following the
>rules of our model of understanding can understand in the same way
>we understand.
That appears to be based on a presumption that understanding is a
matter of following rules of some model. I think it is fair to say
that nobody knows what are the rules of understanding, if there are
such things. I think the idea that understanding is the following of
rules is mistaken.
Gary Forbis
Neil Rickert wrote:
>
> In <334F32...@accessone.com> Gary Forbis <for...@accessone.com> writes:
> >Jim Balter wrote:
Everything Jim Balter wrote was deleted.
I wrote:
> >Layered on top of physics is our model of understanding. We believe
> >humans can understand. Some of us believe machines following the
> >rules of our model of understanding can understand in the same way
> >we understand.
And Niel Rickert responded:
> That appears to be based on a presumption that understanding is a
> matter of following rules of some model. I think it is fair to say
> that nobody knows what are the rules of understanding, if there are
> such things. I think the idea that understanding is the following of
> rules is mistaken.
And I agree with this. It is for this reason I disbelieve the System's
Reply to the CR.
--
--gary
for...@accessone.com
Neil Rickert
In <334FF9...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>Neil Rickert wrote:
>I wrote:
>> >Layered on top of physics is our model of understanding. We believe
>> >humans can understand. Some of us believe machines following the
>> >rules of our model of understanding can understand in the same way
>> >we understand.
>And Niel Rickert responded:
>> That appears to be based on a presumption that understanding is a
>> matter of following rules of some model. I think it is fair to say
>> that nobody knows what are the rules of understanding, if there are
>> such things. I think the idea that understanding is the following of
>> rules is mistaken.
>And I agree with this. It is for this reason I disbelieve the System's
>Reply to the CR.
The main point of the Systems Reply is that Searle did not prove his
claim, because he overlooked the possibility that the understanding
might be in the system as a whole, rather than the component that
Searle considered.
In any case, there could be more going on in computer systems than
following rules. (And, to further complicate things, AW insists that
computers are not actually following rules.)
Gary Forbis
Neil Rickert wrote:
>
> In <334FF9...@accessone.com> Gary Forbis <for...@accessone.com> writes:
> >Neil Rickert wrote:
>
> >I wrote:
>
> >> >Layered on top of physics is our model of understanding. We believe
> >> >humans can understand. Some of us believe machines following the
> >> >rules of our model of understanding can understand in the same way
> >> >we understand.
>
> >And Niel Rickert responded:
>
> >> That appears to be based on a presumption that understanding is a
> >> matter of following rules of some model. I think it is fair to say
> >> that nobody knows what are the rules of understanding, if there are
> >> such things. I think the idea that understanding is the following of
> >> rules is mistaken.
>
> >And I agree with this. It is for this reason I disbelieve the System's
> >Reply to the CR.
>
> The main point of the Systems Reply is that Searle did not prove his
> claim, because he overlooked the possibility that the understanding
> might be in the system as a whole, rather than the component that
> Searle considered.
I am confused by the Systems Reply if it is only a bald assertion. I
take
it as something more. The System in question is just following rules.
Do those supporting it think it is doing something more? If so, what
and if so isn't this extending the system defined by Searle?
> In any case, there could be more going on in computer systems than
> following rules. (And, to further complicate things, AW insists that
> computers are not actually following rules.)
I fall into this camp. I think computers are obeying the laws of
physics and are devices designed in a way where their behavior may
be easily interpretted as following formal rules. When they fail
to adhere to our expectations we say they are malfunctioning even though
they have no choise but to follow the laws of physics (as opposed to
our models of the laws of physics.)
--
--gary
for...@accessone.com
Neil Rickert
In <33507B...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>Neil Rickert wrote:
>> In <334FF9...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>> >Neil Rickert wrote:
>> >I wrote:
>> >> >Layered on top of physics is our model of understanding. We believe
>> >> >humans can understand. Some of us believe machines following the
>> >> >rules of our model of understanding can understand in the same way
>> >> >we understand.
>> >And Niel Rickert responded:
>> >> That appears to be based on a presumption that understanding is a
>> >> matter of following rules of some model. I think it is fair to say
>> >> that nobody knows what are the rules of understanding, if there are
>> >> such things. I think the idea that understanding is the following of
>> >> rules is mistaken.
>> >And I agree with this. It is for this reason I disbelieve the System's
>> >Reply to the CR.
>> The main point of the Systems Reply is that Searle did not prove his
>> claim, because he overlooked the possibility that the understanding
>> might be in the system as a whole, rather than the component that
>> Searle considered.
>I am confused by the Systems Reply if it is only a bald assertion. I
>take
>it as something more. The System in question is just following rules.
Is it? That is not obvious.
My automobile is a mechanism. It can be descibed as following
rules. Those rule specify, for example, the relation of the opening
of the distributer contacts to the position of the pistons in the
cylinders. They specify the relation of the steering wheel position
to the orientation of the front wheels.
Does my automobile get me to work each day by just following rules?
It seems to me that the automobile is doing rule following on
internal data. But I don't see that it can be described as following
rules with respect to the external data (the potholes in the road,
for example). The rules which define the automobile's mechanical
operation say nothing about the external data.
Similarly, from the fact that a computer can be described as rule
following in terms of internal data, I don't see that it is
necessarily following rules with respect to external data. If I
looked at the rules which govern the operations of my current
computer, I don't think I would find any rule about fetching articles
from comp.ai.philosophy. Nonetheless, that is what my computer is
doing.
Gary Forbis
Yes. Being a physical system it's behavior will not always conform to
design specifications or may have undefined behavior outside the
design's
domain, none the less, within the design's domain a failure to conform
to
the rules is judged a system failure.
> It seems to me that the automobile is doing rule following on
> internal data. But I don't see that it can be described as following
> rules with respect to the external data (the potholes in the road,
> for example). The rules which define the automobile's mechanical
> operation say nothing about the external data.
The Outback is heavily advertised here in the north west US. The
advertisments talk about good handling on rough roads (and the name,
I suppose, is intended to invoke notions of ruggedness.) I would hope
the vehicle's behaviors upon hitting various sized potholes are well
defined. If the vehicle is wild when used in the expected way by the
target audience I hope government would have it taken off the market.
> Similarly, from the fact that a computer can be described as rule
> following in terms of internal data, I don't see that it is
> necessarily following rules with respect to external data. If I
> looked at the rules which govern the operations of my current
> computer, I don't think I would find any rule about fetching articles
> from comp.ai.philosophy. Nonetheless, that is what my computer is
> doing.
The system accessing c.a.p has rules about fetching articles from c.a.p.
I don't know how it could find the articles without having such rules.
Your computer is using software that gets the name of the various
newsgroups you read from a database of newsgroups names and parameters
and retrieves the articles based upon a Usenet access protocol from a
Usenet server of your choosing.
I don't know about you but I get upset when my software doesn't behave
according to specifications/expectations. Sometimes I get a GPF from
my browser. I don't accept this as part of its normal behavior but
mentally whine about "stupid programmers." When testing software I
assume the computer is behaving according to design rules and the
software
should be doing likewise.
--
--gary
for...@accessone.com
Jim Balter
Gary Forbis wrote:
> I am lead to wonder how anyone can believe both that they understand
> integer arithmatic and that a machine understands the same way they
> do merely (and I mean *mearly*) by way of following the rules of
> a formal model of understanding..
"I understand arithmetic" is just words. Machines can utter the
same words.
--
<J Q B>
Jim Balter
Neil Rickert wrote:
>
> In <334F32...@accessone.com> Gary Forbis <for...@accessone.com> writes:
> >Jim Balter wrote:
No, I didn't. Now you're doing it, Neil.
> >Layered on top of physics is our model of understanding. We believe
> >humans can understand. Some of us believe machines following the
> >rules of our model of understanding can understand in the same way
> >we understand.
>
> That appears to be based on a presumption that understanding is a
> matter of following rules of some model. I think it is fair to say
> that nobody knows what are the rules of understanding, if there are
> such things. I think the idea that understanding is the following of
> rules is mistaken.
Which is not to say that we cannot be seen to be following rules
as we go about understanding. But Forbis is probably going to
make the mistake of confusing the two. This is the confusion
that underlies the Lucas argument: as though there were an identity
between the model of a thing that solves problems and the model of
problem solving. But the former is a causal model, mechanistic,
potentially formalizable, and all that, not yet laid out, while the
latter is not a model in that sense at all, but rather a matter of
our currently held concepts as reflected in current language usage.
We already "know" what we mean by "problem solving" or "understanding";
it isn't something to be discovered: "So what does it really *mean* to
understand?" is more of interest to lexicographists than to scientists.
But we don't know what the causal relations are in those things of which
we find ourselves saying "he/she/it understands"; that *is* something to
discover: "So what does it really take to produce something of which we
will say that it understands?" is of interest to scientists and not to
lexicographists.
We are machines, and we are understanding, but understanding not to be
found in our (not yet elaborated) machine model, it is to be found in
our judgements as social creatures. At this level, rather than ask
"What is the machine doing when it is understanding?", the question to
be asked is "What is the machine doing when it attributes
understanding?" Forbis confuses the folk conceptual model that we form
of "understanding" with the mechanical model that we seek of those
things, including other humans and computers, to which we, as
attributers, attribute understanding. And Weinstein recognizes that
these are two different levels, in fact makes a point of it, but stops
there, apparently denying that it is even legitimate to ask the latter
question; for him understanding *only* exists in its role in language
and the mechanization of the very process of using language is not seen
as having any bearing on what the words mean to us. But for people like
you, me, Sloman, Pindor, et. al., there are causal relationships between
what goes on in people's heads and which words get used in which
contexts, and examining those causal relationships is an important
aspect of AI research.
--
<J Q B>
Jim Balter
Gary Forbis wrote:
>
> Neil Rickert wrote:
> >
> > In <334F32...@accessone.com> Gary Forbis <for...@accessone.com> writes:
> > >Jim Balter wrote:
>
>
> I wrote:
>
> > >humans can understand. Some of us believe machines following the
> > >rules of our model of understanding can understand in the same way
> > >we understand.
>
>
> > That appears to be based on a presumption that understanding is a
> > matter of following rules of some model. I think it is fair to say
> > that nobody knows what are the rules of understanding, if there are
> > such things. I think the idea that understanding is the following of
> > rules is mistaken.
>
> Reply to the CR.
Belief has nothing to do with it; the SR simply points out that
Searle is arguing against a strawman. That is true whether or not
Strong AI is valid, whether or not Searle's conclusions or premises
are right, whether or not understanding is a matter of following
rules. The SR makes no claim whatsoever about such things.
All it says is "Whether the Chinese Room understands Chinese does not
hinge upon whether the Searle homunculus understands Chinese."
That's it; period.
--
<J Q B>
Jim Balter
Gary Forbis wrote:
>
> Neil Rickert wrote:
> >
> > >Neil Rickert wrote:
> >
> >
> > >> >Layered on top of physics is our model of understanding. We believe
> > >> >humans can understand. Some of us believe machines following the
> > >> >rules of our model of understanding can understand in the same way
> > >> >we understand.
> >
> > >And Niel Rickert responded:
> >
> > >> That appears to be based on a presumption that understanding is a
> > >> matter of following rules of some model. I think it is fair to say
> > >> that nobody knows what are the rules of understanding, if there are
> > >> such things. I think the idea that understanding is the following of
> > >> rules is mistaken.
> >
> > >And I agree with this. It is for this reason I disbelieve the System's
> > >Reply to the CR.
> >
> > claim, because he overlooked the possibility that the understanding
> > might be in the system as a whole, rather than the component that
> > Searle considered.
>
> I am confused by the Systems Reply if it is only a bald assertion.
How is that a bald assertion? If I say "I think my car weighs more
than a ton" and you say "Your steering wheel doesn't weigh a ton,
so you're wrong" is it "only a bald assertion" to point out that I
never *expected* my steering wheel to weigh a ton?
Also, you continue to confuse the negative Systems Reply with the
positive functionalist thesis that Searle labels Strong AI.
They aren't the same thing. Perhaps you want to argue against the
latter, not the former.
> I
> take
> it as something more. The System in question is just following rules.
> Do those supporting it
"it"? You mean Strong AI? Note again that that is not the same as
the Systems Reply.
> think it is doing something more? If so, what
> and if so isn't this extending the system defined by Searle?
You are just living. You are also just planning for you future.
And you are just trying to grapple with difficult concepts.
And you are just understanding English. And you are just
metabolizing. And you are just providing a vehicle for your genes.
Just as your brain can be just transmitting neural signals
consistent with the chemical and structural constraints of neurons
at the same time that it is just understanding English, the CR
can be just instantiating some program at the same time that it is
just understanding Chinese. There are no "extensions" involved,
but I can see how your insistence on misapplying words like "just"
and "merely", which are only appropriate with things that are
*mutually exclusive*, might lead you to think so.
> > In any case, there could be more going on in computer systems than
> > following rules. (And, to further complicate things, AW insists that
> > computers are not actually following rules.)
>
> I fall into this camp. I think computers are obeying the laws of
> physics
If that is "more going on", it is not anything that was ever denied or
at issue. The subject has always been instantiations of programs,
not programs in the abstract. Searle wants to make a *distinction*
between how the CR operates and how the human brain operates,
and that distinction certainly isn't that one obeys the laws of physics
but other doesn't. (Which doesn't even make sense; how could something
disobey the laws of physics? If something did, then they wouldn't
be the laws of physics after all. Russell already dealt with this
with "exist"; it isn't a predicate. Neither is "obeys the laws of
physics").
> and are devices designed in a way where their behavior may
> be easily interpretted as following formal rules.
The behavior of anything can be interpreted as following formal rules.
Not easily by some, apparently. The difference is that we already
know how to see computers as following rules because we started from
the direction of the rules, whereas with other things like humans,
we first have humans and then try to figure out how to view them as
following rules. Its a matter of engineering vs. reverse engineering,
except that in reverse engineering we still start from the assumption
that the thing can be seen as following rules. But with humans,
the damn things are so complex and there are so many emotional and
philosophical pitfalls that it is easy to simply deny that they can
be seen as following rules; after all, no one can prove otherwise.
Then this all gets more twisted due to failing to realize that
"understanding" is an *attribution* that we make based upon behavior,
rather than some intrinsic property. "If understanding
isn't a matter of following rules, then how can something that can be
seen as following rules be understanding?" Well, it's easy.
Since it isn't an intrinsic property, we can put various labels on some
behavior produced by something that is following rules, including
"understanding". We could, for instance, say that it is "making
progress", or that it is "missing the point", or "becoming confused", or
"solving a problem". None of these is itself "following rules", but
neither are any of them an "extension". There is no fact of the matter
as to whether something is "missing the point". These are all
attributions, judgements, that *we* form as we are stimulated by its
behavior.
> When they fail
> to adhere to our expectations we say they are malfunctioning even though
> they have no choise but to follow the laws of physics (as opposed to
> our models of the laws of physics.)
It's the same with us. It's all a matter of attribution. If I think
that total disregard for the interests of others is a malfunction,
I can label that sociopathy. If someone else thinks that putting
the interests of others before those of oneself is a malfunction,
they may write novels about the evils of altruism. Those that think
there is some fact of the matter, that perhaps the "law" of the survival
of the fittest rules that one is wrong and the other is right, are
very deeply confused.
But this is somewhat beside the point that Searle has mistaken
the steering wheel for the car.
--
<J Q B>
Neil Rickert
>> In <334F32...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>> >Jim Balter wrote:
>No, I didn't. Now you're doing it, Neil.
Sorry about that.
>> >Layered on top of physics is our model of understanding. We believe
>> >humans can understand. Some of us believe machines following the
>> >rules of our model of understanding can understand in the same way
>> >we understand.
>> That appears to be based on a presumption that understanding is a
>> matter of following rules of some model. I think it is fair to say
>> that nobody knows what are the rules of understanding, if there are
>> such things. I think the idea that understanding is the following of
>> rules is mistaken.
>Which is not to say that we cannot be seen to be following rules
>as we go about understanding. But Forbis is probably going to
>make the mistake of confusing the two.
You appear to have been right on both counts.
> This is the confusion
>that underlies the Lucas argument: as though there were an identity
>between the model of a thing that solves problems and the model of
>problem solving.
This confusion seems to be all too common.
Gary Forbis
Jim Balter wrote:
>
> Neil Rickert wrote:
> >
> > In <334F32...@accessone.com> Gary Forbis <for...@accessone.com> writes:
> > >Layered on top of physics is our model of understanding. We believe
> > >humans can understand. Some of us believe machines following the
> > >rules of our model of understanding can understand in the same way
> > >we understand.
> >
> > That appears to be based on a presumption that understanding is a
> > matter of following rules of some model. I think it is fair to say
> > that nobody knows what are the rules of understanding, if there are
> > such things. I think the idea that understanding is the following of
> > rules is mistaken.
>
> Which is not to say that we cannot be seen to be following rules
> as we go about understanding.
Darn it. Penrose looks at what he is doing when he does math and
concludes it cannot be seen as following rules. I believe Wittgenstein
goes as far as to say understanding is about the ability to generate
rules rather than to follow them.
While the substrate may be idealized as following rules it is obeying
the laws of physics rather than our idealization of the laws of physics.
Humans understand because on occasion they fail to adhere to our
idealization and the failure is fortuitous. On these occasions the
humans
will say something like "I had an insight."
One might counter by adding a randomization element to the rule
following
but I'd like to know how it can be done. Among what alternatives is
the rule follower to randomize? The designed substrate is given some
set of rules and may generate composites of these atomics but it cannot
generate new atomics as something other than simplifications of
composites
of the design atomics. If the designed machine understands it is not
because it is following rules but because it can generate rules about
that which it has understanding.
> But Forbis is probably going to
> make the mistake of confusing the two. This is the confusion
> that underlies the Lucas argument: as though there were an identity
> between the model of a thing that solves problems and the model of
> problem solving. But the former is a causal model, mechanistic,
> potentially formalizable, and all that, not yet laid out, while the
> latter is not a model in that sense at all, but rather a matter of
> our currently held concepts as reflected in current language usage.
The former is formalizable if and only if the solutions to the problems
are reachable by a finite set of rules in a finite amount of time.
Now it happens that most people don't need to understand math the way
Penrose understands math; we can get by by following rules. Machines
can be made to be as competent as most humans by following rules but
if one wants a machine that understands it the way Penrose understands
it will need to be able to generate new atomic rules (not just
composites
of designed rules.)
--
--gary
for...@accessone.com
Jim Balter
Gary Forbis wrote:
>
> Jim Balter wrote:
> >
> > Robin Faichney wrote:
> > >
> > > Oliver Sparrow wrote:
> > > >
> > > > The room's a ROM. No one expects intelligent ROMs.
> > > > No one expects a program that looks like a ROM - packed
> > > > with declarative statements with no underpinning connectivity -
> > > > to be intelligent in appreanace or actuality. Searle argues
> > > > against a seventies model of AI.
> > >
> > > As I understand it, the "underpinning connectivity" is
> > > implicit in the rules. After all, that the CR behaves
> > > as if understanding is one of the premises. Which
> > > seems to imply that intelligence is granted.
> >
> > Quite; the Sparrow pecks at a straw Searle. Intelligence is granted,
> > understanding is not.
>
> It is quite common for arguments to take the form:
>
> A premise
> B premise
> ...
> (A and B) => (~A or ~B) some rule of logic
Um, that's a contradiction, not some rule of logic.
Some rule of logic gives us (A and B) => ~(~A or ~B)
> therefore ~A or ~B
> or
> therefore A => ~B
I don't know what you are getting at; you have contradicted
your premises by employing an invalid argument.
> The degreee to which one holds A or B determines ones prejudices
> towards ~A or towards ~B.
They aren't prejudices, they are inferences. Holding to a premise
is a prejudice; that's why it is called a "pre"judice.
Holding that ~(~A or ~B), OTOH, is an inference from the premises;
it is not a prejudice, just a plain ol' "judice".
> I think most hold a rock doesn't have
> to understand the laws of physics in order to obey them.
I think your statements don't flow from one to another in any sort
of orderly fashion.
> Both humans and machines are obeying the laws of physics. Machines
> have a further restriction in that in order to be labeled as properly
> functioning they must follow the human model of physics.
Not at all. My car is functioning properly if it does what I
expect of it; I don't know that it is following the human model
of physics; that would depend upon whether the human model of
physics were accurate, and I really don't know that it is;
probably not in all particulars.
> Layered on top of physics
Is this physics, or our model of it?
> is our model of understanding. We believe
> humans can understand.
What do you think is our model of understanding? We have a *notion* of
understanding, and we observe humans being consistent with it.
If you have a model of understanding, do tell what it is.
For me, you don't fit my notion of understanding the subject at
hand, because you make mistakes that I would not expect to follow
from having what I think of as an accurate view. So for me,
understanding means something like "accepting the basic
propositions as true", but I can't know whether you do in fact
accept them to be true, I can only know whether your statements
are consistent with them. And I haven't laid out exactly what these
basic propositions are, I can only detect when your statements
appear consistent or inconsistent with what I hold to be true,
including the logical inferences I can make from what I hold to be true.
So, I can only judge whether you understand this stuff relative to my
own understanding of it, and I have no model for that either.
As for "humans can understand", understand what? If you just mean
in general, then by my "model" this means that humans can accept
some basic propositions in some area as true. This isn't a matter
of belief, but rather an interpretation of humans as accepters of
propositions. I can interpret programmed systems as accepters of
propositions, too, and I often do. So I guess this means that
I "believe" that machines can understand whenever I am interpreting
them as accepters of propositions. But you seem to think that
there is some true fact as to whether a programmed system can
understand, and so my "belief" might be accurate or inaccurate.
But I believe that this is because you don't understand the issues,
nor how to go about thinking about them.
> Some of us believe machines following the
> rules of our model of understanding can understand in the same way
> we understand.
Who are these people? What rules? What model? The Strong AI
thesis is not that machines that follow some rules that we have
already identified must be understanding, but rather that there is
*some* program such that, if the machine were controlled by it,
the machine would do all the same sorts of things that people do
when we say that they understand, and there wouldn't be anything
(soul? experience? ether?) missing that was relevant to understanding.
> Goedel came up with his Incompleteness Theorem.
>
> Many people claim to understand integer arithmatic even though
> the Incompleteness Theorem tells us any formal model we might
> concieve of integer arithmatic is either inconsistent or incomplete.
Do you have any reason to think you understand this stuff? Unless the
formal model of you is the formal model of arithmetic, it matters not a
whit what you can prove about arithmetic, let alone what you might
"claim" to "understand". What is relevant is whether you can prove that
*you* are consistent; good luck. Assuming, that is, that "you" are
representable as an axiomatic system of the sort that satisfies
the constraints of GIT. But you aren't, since you are an open-ended
system. And the same thing is true of any machine you might be
concerned with. The CR gets arbitrary inputs from humans; humans are
open-ended systems. The CR is therefore an open-ended system. Goedel
isn't relevant to a Turing machine with a tape whose contents are
undetermined.
> I am lead to wonder how anyone can believe both that they understand
> integer arithmatic
Understanding integer arithmetic is not the same as being able to
prove, in arithmetic, every true proposition of arithmetic. Why do you
think "understanding" has anything to do with Goedel? Why do you think
that you understand this stuff?
> and that a machine understands the same way they
> do
"The same way" being able to accept basic propositions as true,
by my "model".
> merely (and I mean *mearly*) by way of following the rules of
> a formal model of understanding..
The claim concerns the *right* program, not one that some confused
person might think represents a formal model of understanding. BUT ...
Even assuming that we had a formal model of understanding, and it were
accurate, the program for an understander is *not* "the rules of a
formal model of understanding", any more than the program for a chess
player are "the rules of a formal model of chess playing". The formal
model of chess playing is pretty much the rules of chess plus some
stuff about actually being engaged in playing a game, but that isn't
anything like a program that a chess machine follows. Rather, a chess
machine follows one of many different sets of rules for *being a chess
player*, just as an understanding machine would follow one of many
different sets of rules for *being an understander*.
Is it possible, Gary, that you are "led to wonder" because you
*do not understand*?
--
<J Q B>
Jim Balter
Gary Forbis wrote:
>
> Neil Rickert wrote:
> > My automobile is a mechanism. It can be descibed as following
> > rules. Those rule specify, for example, the relation of the opening
> > of the distributer contacts to the position of the pistons in the
> > cylinders. They specify the relation of the steering wheel position
> > to the orientation of the front wheels.
> >
> > Does my automobile get me to work each day by just following rules?
>
> Yes. Being a physical system it's behavior will not always conform to
> design specifications or may have undefined behavior outside the
> design's
> domain, none the less, within the design's domain a failure to conform
> to
> the rules is judged a system failure.
The rules that describe the automobile's internal behavior, such as
those that Neil mentioned, are not the rules of external behavior it is
supposed to follow, such as not rushing forward when you hit the brakes.
Neil gave *descriptive* rules. The car cannot fail to conform to the
rules that *describe* its internal behavior. OTOH, it certainly can
fail to conform to the rules that *prescribe* its external behavior.
But those aren't the sort of rules that the CR is said to following.
They aren't the sorts of rules in Searle's rule book.
> > It seems to me that the automobile is doing rule following on
> > internal data. But I don't see that it can be described as following
> > rules with respect to the external data (the potholes in the road,
> > for example). The rules which define the automobile's mechanical
> > operation say nothing about the external data.
>
> The Outback is heavily advertised here in the north west US. The
> advertisments talk about good handling on rough roads (and the name,
> I suppose, is intended to invoke notions of ruggedness.) I would hope
> the vehicle's behaviors upon hitting various sized potholes are well
> defined. If the vehicle is wild when used in the expected way by the
> target audience I hope government would have it taken off the market.
You don't seem to have read what Neil wrote. Car wheels can be designed
so as to maximize the probability that the behavior is within a certain
range when it hits potholes, but there is nowhere in the car to put the
rule "bounce no more than yay far when encountering a pothole". Rather,
the rules are more like "be yay compressable", "turn yay fast", and so
on. "bounce no more ..." is a desired consequence, a prescription. The
others are internal descriptions. Programs are internal descriptions.
The internal behavior of a computer that is instantiating a program is
described by that program; if it weren't, then that wouldn't be the
program being instantiated. Whether the result is that it exhibits the
prescribed external behavior is another matter, a matter that pays
programmers' salaries. But being unhappy with the result and thus
throwing the machine in the trash has no bearing on what rules the
machine is actually following.
> > Similarly, from the fact that a computer can be described as rule
> > following in terms of internal data, I don't see that it is
> > necessarily following rules with respect to external data. If I
> > looked at the rules which govern the operations of my current
> > computer, I don't think I would find any rule about fetching articles
> > from comp.ai.philosophy. Nonetheless, that is what my computer is
> > doing.
>
> The system accessing c.a.p has rules about fetching articles from c.a.p.
> I don't know how it could find the articles without having such rules.
[Upon rereading this after writing the following, I see that Neil said
"rule about ..." but Gary said "rules about ...". I took them both to
mean a rule such as "fetch articles from c.a.p", and still take Neil
that way. But Gary seems to have misunderstood Neil to mean "rules that
tell how to fetch articles from c.a.p". But those are internal
descriptive rules, not a prescriptive rule like "get me those c.a.p
articles, damn it". Which is the point. So I'll leave the following as
I wrote it.]
But in fact there is no such rule. There are instructions that convert
a mouse location into an index, move the data (say,
"comp.ai.philosophy") from the indexed memory location to another to
another, and at some point some instructions are executed that move that
data into a hardware register. There are totally different instructions
that some time later read data from another hardware register, and that
data goes through various transformations, and eventually certain memory
locations that happen to have the shape, as we recognize it, of the
characters represented by the data read from the hardware register are
set, and this eventually results in certain phosphors being lit. Note
that this is a *description*. The prescription, "fetch an article from
c.a.p", is nowhere to be found. When that prescription is met, it is a
consequence of the description of the system being the *right*
description to result in the desired behavior. Getting the description
of a system's internal behavior to produce the prescribed external
behavior is the job of the programmer, and its a hell of a lot harder
then inserting the rule "fetch articles from c.a.p" (although modern
software technology is largely about developing other descriptions that
happen to be the right ones to "automatically" translate prescriptions
into descriptions that will bring about the prescribed behavior).
> Your computer is using software that gets the name of the various
> newsgroups you read from a database of newsgroups names and parameters
> and retrieves the articles based upon a Usenet access protocol from a
> Usenet server of your choosing.
But that is not "fetch articles from c.a.p", which is a *consequence* of
all this internal behavior.
> I don't know about you but I get upset when my software doesn't behave
> according to specifications/expectations. Sometimes I get a GPF from
> my browser. I don't accept this as part of its normal behavior but
> mentally whine about "stupid programmers." When testing software I
> assume the computer is behaving according to design rules and the
> software
> should be doing likewise.
But your expectations themselves do not determine what rules the machine
is following. The very fact that you sometimes get upset indicates that
the machine does not follow those rules, but rather is such that its
behavior only approximates your prescriptions. Those expectations led
to (smart) programmers attempting to cause the description of the
machine's internal behavior to be such that it's external behavior would
meet the prescription. They did that via an involved process of
scratching their heads, scribbling on pieces of paper, talking to each
other, and slamming their fingers against little pieces of plastic until
their tendons are sore. Then they watched the machine behave, noted
that it didn't meet the expectations, scratched some more, banged some
more, watched some more, and continued this process until the machine
behavior reached some acceptable approximation to the expectations. The
machine still doesn't follow the rules, but its internal behavior has
been tweaked to the point where its external behavior usually matches
the prescription. And as a result of all this tweaking we have a
description of the internal behavior of the machine, in a form that we
refer to as a "program".
--
<J Q B>
Neil Rickert
In <33516A...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>Jim Balter wrote:
>> Neil Rickert wrote:
>> > That appears to be based on a presumption that understanding is a
>> > matter of following rules of some model. I think it is fair to say
>> > that nobody knows what are the rules of understanding, if there are
>> > such things. I think the idea that understanding is the following of
>> > rules is mistaken.
>> Which is not to say that we cannot be seen to be following rules
>> as we go about understanding.
>Darn it. Penrose looks at what he is doing when he does math and
>concludes it cannot be seen as following rules. I believe Wittgenstein
>goes as far as to say understanding is about the ability to generate
>rules rather than to follow them.
Well, I happen to think that Penrose and Wittgenstein are both right
in those assessments. But then Penrose makes the mistake of thinking
this would somehow contradict the possibility of AI.
Keep in mind that Penrose bases his argument on Goedel's theorem.
Now Goedel's proof is as fine a piece of rule following as you can
find. If doing mathematics is not merely rule following, and if
Goedel's proof is the demonstration of this, as Penrose argues, then
there is something that doesn't fit with the way you and Penrose are
thinking about this.
The point is that Goedel, in proving his theory, can be carrying out
mere rule following in his metasystem. But in doing so he is not
following rules at the level of arithmetic, but is instead
investigating our ability to generate those rule, just as your
comment on Wittgenstein would require.
In the same way, an intelligent computer system could be following
rules of its own internal system, but doing so in such a way as to
allow it to investigate and generate rules that deal with the real
world.
>While the substrate may be idealized as following rules it is obeying
>the laws of physics rather than our idealization of the laws of physics.
>Humans understand because on occasion they fail to adhere to our
>idealization and the failure is fortuitous. On these occasions the
>humans
>will say something like "I had an insight."
I think it would be a mistake to suggest that "Aha, insight" events
come about as the fortuitous result of failing to follow rules. A
better explanation might be that the neurons are doing their own rule
following, where the rules are biochemical. One effect of that
neural rule following, is that our neurons structure themselves so as
to generate rules that we follow, and to improve those rules where
possible. I suggest that the "Aha, insight!" experiences arise when
we become consciously aware of the improved system of rules that the
neurons have been silently generating for us.
>Now it happens that most people don't need to understand math the way
>Penrose understands math; we can get by by following rules.
You won't get very far in mathematics by following rules.
I'm reminded of students who can solve the homework problems in
chapter 6, and can solve the homework problems in chapter 7. But they
do poorly in exams, because the exam questions don't tell them whether
the problem is from chapter 6 or from chapter 7.
The point is, that it is not enough to be able to follow rules.
Doing useful mathematics requires (a) reducing the real world problem
to a formal problem, and (b) applying rules to solve the formal
problem. Most of the hard work is in (a) rather than in (b). If one
does (a) by following rules, then they are not rules that deal with
real world entities, but rules that deal with the internal steps
required to organize the data in a way that will reduce the problem
to a formal one.
[Incidently, Penrose's books caused me to revise downwards my
estimate of his understanding of mathematics.]
Neil Rickert
In <335186...@netcom.com> Jim Balter <j...@netcom.com> writes:
>Gary Forbis wrote:
>> Neil Rickert wrote:
>> > If I
>> > looked at the rules which govern the operations of my current
>> > computer, I don't think I would find any rule about fetching articles
>> > from comp.ai.philosophy. Nonetheless, that is what my computer is
>> > doing.
>> The system accessing c.a.p has rules about fetching articles from c.a.p.
>> I don't know how it could find the articles without having such rules.
>[Upon rereading this after writing the following, I see that Neil said
>"rule about ..." but Gary said "rules about ...". I took them both to
>mean a rule such as "fetch articles from c.a.p", and still take Neil
>that way. But Gary seems to have misunderstood Neil to mean "rules that
>tell how to fetch articles from c.a.p". But those are internal
>descriptive rules, not a prescriptive rule like "get me those c.a.p
>articles, damn it". Which is the point. So I'll leave the following as
>I wrote it.]
Yes, you understood me as I had intended it to be understood.
Apparently Gary took a different interpretation.
Let me just summarize my position. In both of the cases I mentioned,
the automobile and the computer, the systems are following rules of
internal operation. My automobile has no rule which would have the
effect of getting me to my work. The computer has no rule which
would cause it to fetch c.a.p. articles. But in both cases the
effect of the internal rule following is to make the systems highly
responsive to external input. Thus my automobile gets me to work
because it responds to the inputs I provide at the steering wheel and
the other controls. My computer fetches c.a.p. articles because it
is highly responsive to the inputs I provide at the keyboard.
Now let's get back to "understanding." I agree with Gary, that it is
implausible that understanding is a matter of rule following. But I
think it plausible that understanding is a matter of being
sufficiently responsive to a wide range of external inputs. That
sort of responsiveness seems to be what we test when we examine
students to assess their understanding. If a suitable system based
on internal rule following can produce this responsiveness, then it
can produce understanding. Our current computers are not yet
sufficiently responsive. They are highly responsive, but only to a
too-limited range of types of input.
Jim Balter
Gary Forbis wrote:
>
> Jim Balter wrote:
> >
> > Neil Rickert wrote:
> > >
> > > >humans can understand. Some of us believe machines following the
> > > >rules of our model of understanding can understand in the same way
> > > >we understand.
> > >
> > > matter of following rules of some model. I think it is fair to say
> > > that nobody knows what are the rules of understanding, if there are
> > > such things. I think the idea that understanding is the following of
> > > rules is mistaken.
> >
> > Which is not to say that we cannot be seen to be following rules
> > as we go about understanding.
>
> Darn it. Penrose looks at what he is doing when he does math and
> concludes it cannot be seen as following rules.
When I write "X does not say Y", that is not so much to assert Y, as to
draw attention to the *distinction* between X and Y, a distinction that
you blithely ignore with your counter that some important person claims
Y. But in fact, Penrose does not claim Y; you, and perhaps he, fail to
recognize that there are different relevant points of view. The claim
that Penrose can be seen as following rules is *not* the claim that
Penrose is *consciously* following rules, it is the claim that the box,
the chunk of meat, the mechanism, that we call Penrose, can be seen to
be following rules. Perhaps the fact that you repeatedly miss this as
even a possible subject of the discussion is due to a predisposition
against the very notion, but by shutting your eyes to it and always
translating the third person into the first person, you will never know
whether there is any sense to be made of it or not. In *this* sense of
"what we are doing", Penrose cannot look at what he is doing when he
does math. He does not own a Penrose-brain analyzer or a
what-Penrose's-brain-is-doing-scope with which he can see this.
Therefore, your statement is false, in the sense that I meant, and that
is always meant when we are talking about a *mechanism* following rules.
You cannot simply discard this meaning and switch to another one. All
the "darn it"s in the world will not make up for this sort of failure to
comprehend.
As far as what Penrose is consciously doing, mostly he is sitting there
trying to get stuff to bubble up into his conscious thoughts, like we
all do when we are thinking. Of course this is not "following rules",
and no one ever said it was.
Consider a familiar example of something that *is* following rules in
the sense that Penrose is denying (and I concur): a chess player. A
chess player follows the rules of chess in playing chess; Penrose does
not follow any rules of how to do mathematics in doing mathematics. But
suppose the chess player is a computer. Then there are *two* sets of
rules; the rules of chess, and the rules that make up the program. The
latter are the sorts of rules that Searle is talking about when he talks
about a rule book in the CR; he is not talking about "rules of
understanding"; there are no more rules of understanding than there are
rules of doing mathematics. But just as a chess playing program may
follow LISP rules in order to be a chess player, a mathematics program
may follow LISP rules to do mathematics, even though there are no rules
for doing mathematics *in the other sense*. And an understanding
program may follow LISP rules in order to be an understander, even
though there are no rules for doing understanding. A computer playing
at chess is following programming rules that cause it to obey chess
rules; clearly, I would hope, these are not the same rules, nor even the
same sort of rules. We should use different terms; I think
"description" vs. "prescription" or "norm" would make things clearer. If
you really want to understand this stuff, then you should make it an
exercise to consider every instance of "rule" and translate it into
description or prescription. A chess playing machine satisfies a
description, given by a program, which results in it following the
prescription or norm, namely the rules of chess. There are no norms for
doing mathematics or for understanding, but there may well be (in fact,
how could there not be, asks the functionalist) descriptions of systems
that do mathmatics or understand Chinese. If there is a description of
Penrose as he goes about doing mathematics, then he, the mechanism, not
the conscious self, can be seen as "following rules".
> I believe Wittgenstein
> goes as far as to say understanding is about the ability to generate
> rules rather than to follow them.
But again this is in the sense of norms, not descriptions. If you are
going to think *correctly* about something, you must do it like *yay*.
The one who understands can *write* the book, rather than have to refer
to it. But what about writing that book, how is that done? The great
writer doesn't need a book on how to write. How does she do it? Well,
she's a mechanism (I say) such that the results of her functioning is
writing great books. And there's a (incredibly complex) description of
that mechanism. And it could be encoded and instantiated elsewhere,
says the functionalist.
> While the substrate may be idealized as following rules it is obeying
> the laws of physics rather than our idealization of the laws of physics.
I don't quite know what "obeying the laws of physics" means other than
"is physical". The world is as it is; we try to describe it by
crafting laws of physics.
> Humans understand because on occasion they fail to adhere to our
> idealization and the failure is fortuitous. On these occasions the
> humans
> will say something like "I had an insight."
So humans only understand on occasion, and its always a failure to
adhere to some idealization? Doesn't sound right to me. I think of an
insight as recognizing a particularly juicy pattern, especially one that
was hidden in the space of possibilities, or that has tendrils that
connect seemingly unconnected things. The recognition of a pattern
equates to the generation of a rule.
> One might counter by adding a randomization element to the rule
> following
> but I'd like to know how it can be done. Among what alternatives is
> the rule follower to randomize? The designed substrate is given some
> set of rules and may generate composites of these atomics but it cannot
> generate new atomics as something other than simplifications of
> composites
> of the design atomics. If the designed machine understands it is not
> because it is following rules but because it can generate rules about
> that which it has understanding.
There's a pattern space through which we search. Randomization balances
the search, keeps from spending too much time going down dead ends,
avoids search patterns that "resonate" with the search space in such a
way as to repeatedly miss some branches. Perhaps you should familiarize
yourself with this model; yours seems confused and degenerative.
>
> > But Forbis is probably going to
> > make the mistake of confusing the two. This is the confusion
> > that underlies the Lucas argument: as though there were an identity
> > between the model of a thing that solves problems and the model of
> > problem solving. But the former is a causal model, mechanistic,
> > potentially formalizable, and all that, not yet laid out, while the
> > latter is not a model in that sense at all, but rather a matter of
> > our currently held concepts as reflected in current language usage.
>
> The former is formalizable if and only if the solutions to the problems
> are reachable by a finite set of rules in a finite amount of time.
Yet another example of the same confusion; you seem to have no idea what
I am talking about. We don't have to solve the problems that a problems
solver can solve in order to formalize the problem solver. Otherwise
there would be no point to programming, where we formalize our problem
solver precisely so that it will go about solving unsolved problems.
Hell, when I was 15 I wrote a program to calculate pi by successive
approximation. There's a problem the solution of which is not solvable
in finite time, yet I had no trouble in formalizing it. Your claim is
blatantly false, but it seems to rest on the same confusion between what
something *does* and what it *is*.
> Now it happens that most people don't need to understand math the way
> Penrose understands math;
I do hope you aren't holding Penrose up as a great mathematician!
> we can get by by following rules. Machines
> can be made to be as competent as most humans by following rules but
> if one wants a machine that understands it the way Penrose understands
> it will need to be able to generate new atomic rules (not just
> composites
> of designed rules.)
It has to be able to search a space. Not exactly something known to be
beyond computation.
It helps to have the right models, and to recognize them as right.
i.e., understanding.
--
<J Q B>
Gary Forbis
Jim Balter wrote:
>
> Gary Forbis wrote:
> > It is quite common for arguments to take the form:
> >
> > A premise
> > B premise
> > ...
> > (A and B) => (~A or ~B) some rule of logic
>
> Um, that's a contradiction, not some rule of logic.
> Some rule of logic gives us (A and B) => ~(~A or ~B)
>
> > therefore ~A or ~B
> > or
> > therefore A => ~B
>
> I don't know what you are getting at; you have contradicted
> your premises by employing an invalid argument.
That would be true if A and B were atomic so clearly this couldn't have
been
the case. While it may be common to employ invalid arguments I should
hope
you won't see me promoting such an approach as appropriate. I had hope
you could have infered as much. For instance take Euclids proof
1. X is the greatest prime.
....
n. There is no greatest prime.
Are you trying to tell me Euclid followed an invalid argument simply
because
the conclusion contradicts the premise?
> > The degreee to which one holds A or B determines ones prejudices
> > towards ~A or towards ~B.
>
> They aren't prejudices, they are inferences. Holding to a premise
> is a prejudice; that's why it is called a "pre"judice.
> Holding that ~(~A or ~B), OTOH, is an inference from the premises;
> it is not a prejudice, just a plain ol' "judice".
> What do you think is our model of understanding? We have a *notion* of
> understanding, and we observe humans being consistent with it.
> If you have a model of understanding, do tell what it is.
And then later:
> an understanding machine would follow one of many
> different sets of rules for *being an understander*.
Which happens to be very close to one of the premises used by Searle.
The premise is: A machine can understand by virture of following the
rules for being an understander. He didn't choose this premise because
he believed it but because he didn't believe it.
> Is it possible, Gary, that you are "led to wonder" because you
> *do not understand*?
Sure.
--
--gary
for...@accessone.com
Jim Balter
> >> Neil Rickert wrote:
>
> >> > If I
> >> > looked at the rules which govern the operations of my current
> >> > computer, I don't think I would find any rule about fetching articles
> >> > from comp.ai.philosophy. Nonetheless, that is what my computer is
> >> > doing.
>
> >> The system accessing c.a.p has rules about fetching articles from c.a.p.
> >> I don't know how it could find the articles without having such rules.
>
> >[Upon rereading this after writing the following, I see that Neil said
> >"rule about ..." but Gary said "rules about ...". I took them both to
> >mean a rule such as "fetch articles from c.a.p", and still take Neil
> >that way. But Gary seems to have misunderstood Neil to mean "rules that
> >tell how to fetch articles from c.a.p". But those are internal
> >descriptive rules, not a prescriptive rule like "get me those c.a.p
> >articles, damn it". Which is the point. So I'll leave the following as
> >I wrote it.]
>
> Apparently Gary took a different interpretation.
>
> Let me just summarize my position. In both of the cases I mentioned,
> the automobile and the computer, the systems are following rules of
> internal operation.
I take this as meaning that they have some behavior which we
can describe in fairly parsimonious terms as "a rule". When we are
looking at how these systems operate internally, we are taking Dennett's
design stance. From that stance "follow" might be a bit misleading.
They do X because they
are constructed in such as way as to do X, and thus
doing X becomes our description of their behavior. If something
changes (which we might interpret, relative to our expectations,
as a malfunction), then the system might start doing Y instead of X.
It would be a bit awkward to say that it stopped following one rule
and started following another; it seems more a matter of leading than
following. Since it isn't being viewed from the intentional stance,
it isn't an agent, and so I prefer to be neutral: it is simply doing
something else, and we craft a different description to track it.
> My automobile has no rule which would have the
> effect of getting me to my work. The computer has no rule which
> would cause it to fetch c.a.p. articles. But in both cases the
> effect of the internal rule following is to make the systems highly
> responsive to external input.
Yes, I didn't address your point about interaction directly, but
I touched on it in talking about wheels being designed in order
to likely react certain ways to potholes, and in my description
of fetching news articles I simply stuck some bytes (c o m p ...)
into a hardware register and some time later other bytes magically
appeared elsewhere (F r o m ...). Obviously news articles showing
up is beyond the scope of the control system of the computer.
Perhaps less obviously, rolling down the road is beyond the scope
of the control system of the car: there's got to be a road!
Both systems will fail miserably if their environments are wrong,
but that is no failure of the system. And Penrose would not be
able to do mathematics in a (literal) vacuum, but that is no fault
of Penrose. Much as the doctor may tell him to eat his fruits and
vegetables, he cannot follow that rule if he is stuck in a dungeon
with only food and water.
> Thus my automobile gets me to work
> because it responds to the inputs I provide at the steering wheel and
> the other controls. My computer fetches c.a.p. articles because it
> is highly responsive to the inputs I provide at the keyboard.
And to the data coming in on the wire, which might not even be
c.a.p articles, despite all that behavior that was designed to fetch
them.
> Now let's get back to "understanding." I agree with Gary, that it is
> implausible that understanding is a matter of rule following. But I
> think it plausible that understanding is a matter of being
> sufficiently responsive to a wide range of external inputs.
But don't we have to know what we are talking about before we can
talk about what it can be a matter of? And does that phrase mean
"is constituted of" or "is caused by" or "requires"? It doesn't seem to
me to be plausible that any amount of response to external inputs
is in itself "understanding"; that just isn't what I mean by it.
It seems to me to mean that someone holds the right things to be true.
Someone who understands LISP will say the right things about LISP
syntax and semantics, and they won't have to look it up. If they
say the wrong things or don't know what to say, then they don't
understand it. Holding a set of propositions to be true does not
seem to be following rules, although there may be rules that one
"understands" need to be followed in order to carry out some activity.
And being responsive could certainly contribute to reaching
understanding, and would probably be required to do so in pactice,
but a system that understands something could be cloned, and the
clone would still understand, contrary to Anders' views on history.
> That
> sort of responsiveness seems to be what we test when we examine
> students to assess their understanding.
I don't see how we are testing responsiveness per se.
If someone holds the basic propositions to be true, then they
will produce the right responses whereever those propositions
come into play, and that translates into responsiveness. But if
they have the wrong propositions, they may be able to apply them
to many circumstances but get the wrong results. We have to test for
context in order to be sure that it is really the general form of
the proposition, and not some specific case, that is held true,
and we test for a variety of complex contexts for coverage, to
hit many different propositions at once. It may be very difficult
to actually isolate what the invariants of the entire body of
knowledge we are testing for are. If we test for a number
of complex cases and we get the right answers, then we can have
some faith that the basis (mathematical sense) of the body of
knowledge is adhered to, even if we don't know what that basis is.
(I think this concept can be found in multivariate analysis, but
I have only a tangential, um, understanding of the area).
> If a suitable system based
> on internal rule following can produce this responsiveness, then it
> can produce understanding.
This seems to me to mostly be ontogenic; a sufficiently
interactive system can be exposed to a wide search space from which it
can extract patterns that translate into a basis of
understanding.
> Our current computers are not yet
> sufficiently responsive. They are highly responsive, but only to a
> too-limited range of types of input.
And thus they are only able to cover a very small part of the search
space. We humans are exposed to a vast array of data, certainly
no Poverty of Stimulus, from which we are constantly extracting
patterns which become an interlocking knowledge base. There are a
vast number of things that we hold to be true and from which we draw
further inferences that interact with other inferences and observations,
leading us to hold yet more things to be true. And they are to a
large degree the right things, because we are constantly faced with
consequences if we take the wrong things to be true (except in
philosophy, sigh). This powerful learning process and the stream of
inputs that drive it are missing from our computer systems. Programming
by hand introduces knowledge at such a low rate in comparison that
it is no wonder that people take the vast practical gulf between us
and our machines to be a theoretical gulf,
BTW, I've been up for about 36 hours, so if I seem a bit rambling,
repetitive, and just generally whacked out, that's part of the
explanation.
--
<J Q B>
Jim Balter
Gary Forbis wrote:
>
> Jim Balter wrote:
> >
>
> > > It is quite common for arguments to take the form:
> > >
> > > A premise
> > > B premise
> > > ...
> > > (A and B) => (~A or ~B) some rule of logic
> >
> > Um, that's a contradiction, not some rule of logic.
Whoops, actually it's not. It simply asserts ~A or ~B.
But there are no grounds for such an assertion; it is certainly
no rule of logic.
> > Some rule of logic gives us (A and B) => ~(~A or ~B)
> >
> > > therefore ~A or ~B
> > > or
> > > therefore A => ~B
> >
> > I don't know what you are getting at; you have contradicted
> > your premises by employing an invalid argument.
>
> That would be true if A and B were atomic so clearly this couldn't have
> been
> the case.
*What* couldn't have been the case? You are making no sense.
> While it may be common to employ invalid arguments I should
> hope
> you won't see me promoting such an approach as appropriate. I had hope
> you could have infered as much. For instance take Euclids proof
>
> 1. X is the greatest prime.
> ....
> n. There is no greatest prime.
>
> Are you trying to tell me Euclid followed an invalid argument simply
> because
> the conclusion contradicts the premise?
I think I can safely say you haven't the faintest idea what you
are talking about. Euclid's argument does not employ a bogus
"rule of logic" like
(A and B) => (~A or ~B)
>
> > > The degreee to which one holds A or B determines ones prejudices
> > > towards ~A or towards ~B.
> >
> > They aren't prejudices, they are inferences. Holding to a premise
> > is a prejudice; that's why it is called a "pre"judice.
> > Holding that ~(~A or ~B), OTOH, is an inference from the premises;
> > it is not a prejudice, just a plain ol' "judice".
>
> > What do you think is our model of understanding? We have a *notion* of
> > understanding, and we observe humans being consistent with it.
> > If you have a model of understanding, do tell what it is.
>
> And then later:
>
> > an understanding machine would follow one of many
> > different sets of rules for *being an understander*.
>
> Which happens to be very close to one of the premises used by Searle.
> The premise is: A machine can understand by virture of following the
> rules for being an understander.
That asserts that there are such machines. I didn't assert it
("would" makes it conditional), but as it is the Strong AI thesis,
I'm certainly sympathetic to it. But in any case you have totally
missed the point (actually several points).
> He didn't choose this premise because
> he believed it but because he didn't believe it.
Um, so bloody what? Searle and Euclid each presented an argument
from contradiction. Euclid's was valid, Searle's wasn't.
Whether or not you intended your argument to be an argument from
contradiction, it is invalid, and much more obviously than Searle's,
because it employs this bogus "rule of logic":
(A and B) => (~A or ~B)
> > Is it possible, Gary, that you are "led to wonder" because you
> > *do not understand*?
>
> Sure.
It has gone way beyond possibility.
1+1=2 premise
2+2=4 premise
(1+1=2 and 2+2=4) => (1+1!=2 or 2+2!=4) some rule of Forbis-logic
Therefore 1+1!=2 or 2+2!=4
Therefore 1+1=2 => 2+2!=4 Forbis-QED
Yeah, right. That's about standard for Forbis-thinking.
--
<J Q B>
Phil Roberts, Jr.
> The premise is: A machine can understand by virture of following the
> he believed it but because he didn't believe it.
>
I would state this a bit differently, i.e., that there is at present no
particular reason to suppose that our capacity to cognize algorhythms is
itself an algorhythm, although there are probably a number of emotional
reasons why strong AI advocates want to believe it must be so.
--
Phil Roberts, Jr.
Feelings of Worthlessness from the Perspective of
So-Called Cognitive Science
http://www.geocities.com/Athens/5476
Jim Balter
Anders N Weinstein wrote:
>
> In article <5ipnoa$4...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
> > (And, to further complicate things, AW insists that
> >computers are not actually following rules.)
>
> his investigations into following a rule.
>
> On the other hand GOFAI and related rule-and-representation designs for
> intelligent systems are based on the model of a person following very
> exact rules,
They are based upon the model of the mechanism that generates the
"person" (self, personality, self-awareness, whathaveyou)
following very exact rules/descriptions/instruction. Whether the
implemented person follows exact rules/prescriptions/norms is a
different matter, since those rules are imposed externally.
> indeed following them quite "mindlessly". So an AI
> researcher like yourself who is pushing alternative designs can still
> make the point you want to make against rule-based designs, I would say.
The rules that we do or don't follow, as when Penrose
says he doesn't follow rules in doing mathematics, are not *designed*
rules, they are norms. A GOFAI could theoretically implement
a system with a "conscious" module that produced the same sort of
reports as Penrose; from its point of view, it would not be following
rules, and in fact it would not be, since no norms had ever been
laid down for its behavior *at that level*. That does not mean that it
is not implemented rigidly, any more than Penrose not following rules
in doing mathematics means that his implementation is not constrained
by the facts of chemistry and physics. These are two totally different
things, just as the rules of a GOFAI chessplaying program are not the
rules of the game it plays. Mixing them up will forever lead to the
pointless sort of debate and confusion seen here. I think one of
the reasons for the mixup is that the systems people are familiar with
*do* follow design norms, and the connection between the
implementation rules and those norms of behavior are relatively
transparent. But that isn't *necessary*; indeed, a goal of AI
is to build systems that break out of the mold of following norms;
systems with behavior that is not predictable, that invent their own
norms, consistent only with some basics like "go solve some problems
in such and such domain"; that's the sort of rule Penrose is following
when he does mathematics, and I see no a priori reason why we can't
create systems of that sort, even though their internal behavior is
rigidly describable as a Turing machine, just as Penrose's internal
behavior is rigidly describable in terms of chemical bonds and so on.
--
<J Q B>
Anders N Weinstein
In article <5ipnoa$4...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
> (And, to further complicate things, AW insists that
>computers are not actually following rules.)
I said they were not doing what *Wittgenstein* was interested in in
his investigations into following a rule.
On the other hand GOFAI and related rule-and-representation designs for
intelligent systems are based on the model of a person following very
exact rules, indeed following them quite "mindlessly". So an AI
Anders N Weinstein
In article <335279...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>The rules that we do or don't follow, as when Penrose
>says he doesn't follow rules in doing mathematics, are not *designed*
>rules, they are norms. A GOFAI could theoretically implement
As I understand it all rules are norms, as are all designs.
We need to distinguish rules in the sense of norms from mere
tendencies or regularities.
E.g. water tends by law to flow downhill, and indeed its trajectory
minimizes some quantity. But I would not say it is following a rule,
as opposed to conforming to some lawful description we can give.
I think I differ from Rickert in wanting to deny that the mechanical
parts of a car 'follow rules' in the sense in which the microprocessors
in the car 'follow rules' of their program. There is a design which
sets a norm against which we assess their tendencies, but not
anything like consultation of an inner instruction book, which I take
to be involved in the metaphorical sense in which computer programs
follow rules.
You are of course correct that the (non-deterministic)
norms of legal chess play are different than the computatonal rules governing a
symbolic chess-playing computer program. That is akin to the competence
performance distinction in linguistics.
>a system with a "conscious" module that produced the same sort of
>reports as Penrose; from its point of view, it would not be following
>rules, and in fact it would not be, since no norms had ever been
>laid down for its behavior *at that level*. That does not mean that it
OK. I accept the distinction of levels and am mainly interested in the
metaphysics of the higher level, in particular the behavioral norms
laid down by the culture at that level, which I take to include the
norms governing the use of linguistic expressions in which our concepts
are embodied.
>is not implemented rigidly, any more than Penrose not following rules
>in doing mathematics means that his implementation is not constrained
>by the facts of chemistry and physics. These are two totally different
Correct. But I think there is no particular reason to suppose there is
a *rule-governed* sub-cognitive level above the level of physics and
chemistry. As opposed to say a level where descriptions are like that
of water falling downhill, physical tendencies, not cases of
subcognitive *rule-following*. One should not infer from "physical"
to "computational", for computational explanation is rather special,
involving such things as representations, syntax and semantics in a way
physical explanation of tendencies doesn't.
It might also be that no system can be discerned at this level, even
though each particular event is in accord with the basic laws of
physics and chemistry.
Neil Rickert
>> Now let's get back to "understanding." I agree with Gary, that it is
>> implausible that understanding is a matter of rule following. But I
>> think it plausible that understanding is a matter of being
>> sufficiently responsive to a wide range of external inputs.
>But don't we have to know what we are talking about before we can
>talk about what it can be a matter of? And does that phrase mean
>"is constituted of" or "is caused by" or "requires"? It doesn't seem to
>me to be plausible that any amount of response to external inputs
>is in itself "understanding"; that just isn't what I mean by it.
Ok, I was not precise enough. Responding any which way would not be
good enough. Some sort of intelligent response (or adaptive
response) would be needed.
>It seems to me to mean that someone holds the right things to be true.
I'm not so sure about that. It seems to me that someone could
understand natural selection, yet not hold it to be true because of
religious convictions. (I must admit that most of the people who
have religious convictions against natural selection also seem to
misunderstand it).
>> That
>> sort of responsiveness seems to be what we test when we examine
>> students to assess their understanding.
>I don't see how we are testing responsiveness per se.
>If someone holds the basic propositions to be true, then they
>will produce the right responses whereever those propositions
>come into play, and that translates into responsiveness.
This is not obvious. Indeed, I'm not sure that we ever know what
exactly we should consider the basic propositions, or whether there
are such things. I doubt that the brain can be considered a
repository for propositions. Perhaps you are just saying that if
they respond appropriately, we will ascribe to them what we believe
to be the basic propositions. But that would agree with my early
point about judging students on their responses.
Neil Rickert
>I think I differ from Rickert in wanting to deny that the mechanical
>parts of a car 'follow rules' in the sense in which the microprocessors
>in the car 'follow rules' of their program. There is a design which
>sets a norm against which we assess their tendencies, but not
>anything like consultation of an inner instruction book, which I take
>to be involved in the metaphorical sense in which computer programs
>follow rules.
The distinction you are making seems purely arbitrary. In the early
days of computing, the programming was hard wired into the computer.
The choice to place the program in memory, where it can be considered
an inner instruction book, is purely one of convenience. I would
think that the shape of the cam shaft and the shape of the crankshaft
are inner rule books which control the sequencing of valve opening,
spark plug firing, and piston moving in an automobile engine.
Jim Balter
> > The premise is: A machine can understand by virture of following the
> > rules for being an understander. He didn't choose this premise because
> > he believed it but because he didn't believe it.
> >
>
> particular reason to suppose that our capacity to cognize algorhythms is
> itself an algorhythm, although there are probably a number of emotional
> reasons why strong AI advocates want to believe it must be so.
The capacity to cognize algorithms is not an algorithm, it is a
capacity. However, there may well be algorithms that produce
systems that can do such cognizing. There are at present very good
reasons for thinking so, mainly the Church-Turing thesis.
The proposition that Strong AI advocates do so only for reasons
of emotional satisfaction is itself a matter of emotional satisfaction
and gross ignorance.
--
<J Q B>
Phil Roberts, Jr.
Jim Balter wrote:
>
> Phil Roberts, Jr. wrote:
> >
> > I would state this a bit differently, i.e., that there is at present no
> > particular reason to suppose that our capacity to cognize algorhythms is
> > itself an algorhythm, although there are probably a number of emotional
> > reasons why strong AI advocates want to believe it must be so.
>
> The capacity to cognize algorithms is not an algorithm, it is a
> capacity. However, there may well be algorithms that produce
> systems that can do such cognizing. There are at present very good
> reasons for thinking so, mainly the Church-Turing thesis.
> The proposition that Strong AI advocates do so only for reasons
> of emotional satisfaction is itself a matter of emotional satisfaction
> and gross ignorance.
>
Thanks Jim. I knew I could count on you for my much needed humility
lesson. ;)
Gary Forbis
Jim Balter wrote:
>
> Phil Roberts, Jr. wrote:
> >
> > > The premise is: A machine can understand by virture of following the
> > > rules for being an understander. He didn't choose this premise because
> > > he believed it but because he didn't believe it.
> > >
> >
> > particular reason to suppose that our capacity to cognize algorhythms is
> > itself an algorhythm, although there are probably a number of emotional
> > reasons why strong AI advocates want to believe it must be so.
>
> The capacity to cognize algorithms is not an algorithm, it is a
> capacity. However, there may well be algorithms that produce
> systems that can do such cognizing. There are at present very good
> reasons for thinking so, mainly the Church-Turing thesis.
Hmmm...
I don't see how the Church-Turing thesis helps. Doesn't one have to
first
first assume such an algorithm exists before one can conclude it can be
implemented on a Turing machine?
--
--gary
for...@accessone.com
Anders N Weinstein
In article <5iud7i$a...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5iual2$8...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>I think I differ from Rickert in wanting to deny that the mechanical
>>parts of a car 'follow rules' in the sense in which the microprocessors
>>in the car 'follow rules' of their program. There is a design which
>>sets a norm against which we assess their tendencies, but not
>>anything like consultation of an inner instruction book, which I take
>>to be involved in the metaphorical sense in which computer programs
>>follow rules.
>
>The distinction you are making seems purely arbitrary. In the early
>days of computing, the programming was hard wired into the computer.
>The choice to place the program in memory, where it can be considered
>an inner instruction book, is purely one of convenience. I would
I would say, first, it is not "purely arbitrary", since it can be of
importance for designers. The stored program concept is widely held to
be a great achievement. It supports the idea of the universality of
certain architectures, including of course Turing machines. In their
brief for the Physical Symbol System hypothesis, Newell and Simon also
seem to me to be very impressed with the flexibility attainable when
symbols can be bound to *procedures*, symbol structures which can be
constructed, stored, accessed, and then *executed* ("interpreted"),
perhaps ultimately "compiled" into fast routines. (And removing a layer
of interpretation can be important in attaining performance.)
But second, even in the case where the program is "hard-wired" it is
still understood as *computing* in accordance with a program. I take it
that means it is transforming symbolic representations into others in a
way that accords with rules. E.g. an adder instantiates a procedure for
symbolic addition of binary digits, even if it is not interpreting
microcode instructions but "hard-wired".
But I guess you are correct to suggest I should not be relying on the
distinction between the "consulting of explicit rules" done by a
stored-program computer and the analog of primitive "rule following"
without "consultation" we employ in describing the operation of a
hardwired computer.
The more important distinction is between a digital computer as
transformer of symbolic representations and other sorts of mechanisms
that are not computers in this specific sense. Of course some have
been called "analog" computers.
> I would
>think that the shape of the cam shaft and the shape of the crankshaft
>are inner rule books which control the sequencing of valve opening,
>spark plug firing, and piston moving in an automobile engine.
I find this usage trivializing and idle -- you might as well say every
physical system is a rule-following computer, water is a computer
because it finds the shortest path to the bottom of a valley, a falling
rock is a computer, a diving bell *computes* its equilibrium depth,
and so on.
I have disagreed with you on this in the past. I take it (along with
Van Gelder) that in explaining various competences using the concepts
say of dynamical systems theory, rather than the *special* vocabulary
of computation as transformation of representations, as is done in
explaining the Watt Steam Engine governer, we are using an
*alternative* to computation. In this alternative physical causation
takes the place of representation and inference-like operations. Of
course every physical object including a computer is a dynamical
system, but I think computational explanation must be held to involve
some distinctive form of explanation, else any claims made on behalf of
computation get trivialized to the claim that human beings are physical
systems.
Neil Rickert
In <5iuqro$a...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5iud7i$a...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>In <5iual2$8...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>>I think I differ from Rickert in wanting to deny that the mechanical
>>>parts of a car 'follow rules' in the sense in which the microprocessors
>>>in the car 'follow rules' of their program. There is a design which
>>>sets a norm against which we assess their tendencies, but not
>>>anything like consultation of an inner instruction book, which I take
>>>to be involved in the metaphorical sense in which computer programs
>>>follow rules.
>>The distinction you are making seems purely arbitrary. In the early
>>days of computing, the programming was hard wired into the computer.
>>The choice to place the program in memory, where it can be considered
>>an inner instruction book, is purely one of convenience. I would
>I would say, first, it is not "purely arbitrary", since it can be of
>importance for designers. The stored program concept is widely held to
>be a great achievement.
Well, I agree it was a great achievement. But the discussion was
about rule following, rather than about computation. What I am
calling arbitrary, is the restrictions you want to put on what
mechanical systems can be said to be rule following.
>But second, even in the case where the program is "hard-wired" it is
>still understood as *computing* in accordance with a program. I take it
>that means it is transforming symbolic representations into others in a
>way that accords with rules. E.g. an adder instantiates a procedure for
>symbolic addition of binary digits, even if it is not interpreting
>microcode instructions but "hard-wired".
Again, you have changed the subject to "computation" rather than
"rule following." I don't find a whole lot of evidence that we only
use the term "rule following" to refer to transformations of symbolic
representations.
>> I would
>>think that the shape of the cam shaft and the shape of the crankshaft
>>are inner rule books which control the sequencing of valve opening,
>>spark plug firing, and piston moving in an automobile engine.
>I find this usage trivializing and idle -- you might as well say every
>physical system is a rule-following computer, water is a computer
>because it finds the shortest path to the bottom of a valley, a falling
>rock is a computer, a diving bell *computes* its equilibrium depth,
>and so on.
At this stage, I don't much care whether you want to say that the
water or the automobile is computing. The subject, after all, was
rule following, rather than computing. If I recall my history
correctly, in the early Newcomen steam engine humans followed rules
as to when to open the steam valves. They they discovered that they
could automate this rule following. The push rods and cam shafts are
really the modern instantiation of this way of mechanizing what
historically began as a rule following activity. So I still say that
you are making arbitrary distinctions.
I am reminded of a debate I attended many years ago. A philosopher
at Yale wanted to discuss the question of whether skating is an art
or a sport. As the discussion proceeded, it became clear what was
going on. It was common to classify figure skating as an art. What
this philosopher wanted to do was to create a bogus
pseudo-justification of his subjective opinion supporting common
usage. He rejected out of hand the simple explanation that it was
merely a social custom to so classify, and insisted on his so-called
justification.
It seems to me that you are in the business of creating a bogus
pseudo-justification of your subjective opinion that computing can be
considered rule following while the actions of the cam shaft cannot.
But then maybe the creation of bogus pseudo-justifications is about
all that philosophy is good for.
Neil Rickert
In <5j0geb$g...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5iv27j$a...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>OK, rephrase: on your usage everything in nature is a rule following?
I thought we were discussing only the case of the automobile. I took
that to be a relatively uncontroversial example of rule following.
Yet you challenged me on it.
Is everything rule following? I would be more inclined to say that
nothing is rule following. That is, whether something is
rule-following is a question of our interpretation, rather than a
question of intrinsic properties of reality. In the same way, I
would say that nature is not following the laws of physics, but that
we find it convenient and predictively efficacious to interpret
nature as if it were following the laws of physics. (Incidently, I
don't see anything intrinsic in intentionality either.)
>Or is it only a system such that there are design norms dictating what
>it is supposed to do?
I would paraphrase that as "Or is it only a system that we choose to
interpret as rule following?"
>I am reminded of Kant's claim that although everything in nature act in
>*accordance* with rules (laws of nature),
I'm not sure I agree with that. I don't know that there really are
any laws of nature. To put it differently, nature is under no
obligation to follow our scientific laws. Rather, the obligation is
on us to design laws which fit nature as closely as is reasonably
possible.
>I think we want some distinction on pain of trivialization of the claim.
I don't see any problem of trivialization arising. If "rule
following" is a matter of useful interpretation, then such a claim is
non-trivial to the extent that it is useful. As long as interpreting
a computer as a rule following device allows us to program it to
solve problems, then our interpretation has demonstrated its
usefulness. As long as treating the automobile motor as a rule
following device allows us to adjust (and thus program) the cam shaft
to maximize the operating efficiency of the motor, then our
interpretation has demonstrated its usefulness.
>Recall I recommend keeping in mind two distinctions: first is the
>personal/sub-personal level distinction, the distinction between a
>whole person following a public, institutionalized rule at the social
>behavioral level; and a sub [or non]-personal mechanism like a computer
>or bit of functional neural circuitry "following" a rule at its level.
>That is a philosopher's conceptual distinction, in my view.
The main distinction, from my point of view, is that interpreting a
computer as following a rule is far more predictively efficaceous
than is interpreting a whole person as following a rule at the social
level. Generally, I think people are poor rule followers.
>The second is a subsidiary distinction between two types of *design* of
>sub-personal mechanisms: first, those that are designed like digital
>computers to emulate human symbolic rule-followers; second, those whose
>function is explained on other principles, e.g. those of dynamical
>systems or "analog" computation. These are intended to explicate engineer's
>distinctions.
I would characterize that as largely a distinction without a
difference. In the first place, everything is a dynamical system,
including computers. Moreover, the digital computer is an analogue
device which we are able to interpret as digital (or symbolic),
largely because we have designed it to work in a non-linear region.
But beyond that, we don't always design digital systems so as to
emulate human symbolic rule followers. In the CD music recording
system, we use digital computers in a way which emulates and improves
on analog recording methods, but which does not come close to
emulating more typical uses of symbols by humans. The choice of
digital over analog technology is largely a pragmatic one. We choose
analog methods because they offer advantages in reliability and
precision. Presumably our languages are digital systems (with the
phonemes considered digits) because nature too has discovered those
pragmatic advantages of digital systems.
>It might be right that the distinction at the sub-personal level
>is merely heuristic and threatens to break down if pressed at all.
I would say that the distinction between the personal and subpersonal
levels is mostly a matter of putting ourselves on a pedestal as part
of our species-centric way of looking at things. The distinction
between analog and digital at the personal level is mostly a matter
of irrational religious dogma (or philosophical dogma) intended to
support that species-centric viewpoint.
> I mean
>it to be of a piece by other distinctions intuitively used by system designers,
>e.g. that between analog and digital, between pictorial and propositional
>representations, between stored program and hard-wired program.
>All of these are in fact very difficult to make out precisely, I believe,
>although they do seem to appear in the thought of designers.
I suggest that these distinctions emanate from the overemphasis on
the symbolic that results from the religious dogma which is western
philosophy.
> (There is
>a literature attempting to make out an analog/digital distinction, but
>no real consensus, I think on how to do it).
If there is no real consensus, I would take that as supportive of my
view that there is less of a difference than meets the eye.
>I am taking it that digital computation is one very special sort of
>design, one that invites descriptions in terms of representations and
>rules for manipulating them. I take it achieving the behavior by brute
>physical structures is an alternative form of design.
But our computers work only by virtue of the behavior of brute
physical structures. Again, I think you put digital computation and
symbolic processes on too high a pedestal.
>I would be a little surprised if you wanted no distinction at all
>between two alternative styles of control system design, say one that
>uses a digital microprocessor and symbolic operations, and another that
>exploits only physical structure of the parts and their dynamics.
In my studies of digital electronics, I have not come upon a digital
microprocessor which did not work by exploiting the physical
structure of the parts and their dynamics. I think the only
distinctions are pragmatic. And, yes, there are solid pragmatic
grounds for making distinctions.
Anders N Weinstein
>At this stage, I don't much care whether you want to say that the
>water or the automobile is computing. The subject, after all, was
>rule following, rather than computing. If I recall my history
OK, rephrase: on your usage everything in nature is a rule following?
Or is it only a system such that there are design norms dictating what
it is supposed to do?
I am reminded of Kant's claim that although everything in nature act in
*accordance* with rules (laws of nature), only some creatures can act in
accordance with a *conception* (i.e. a representation) of a rule.
(In fact he suggested that we were "prescribing" the laws to nature
in the act of constructing our representations of it)
I think we want some distinction on pain of trivialization of the claim.
I am resting on the analogy with a person operating with symbols.
Recall I recommend keeping in mind two distinctions: first is the
personal/sub-personal level distinction, the distinction between a
whole person following a public, institutionalized rule at the social
behavioral level; and a sub [or non]-personal mechanism like a computer
or bit of functional neural circuitry "following" a rule at its level.
That is a philosopher's conceptual distinction, in my view.
The second is a subsidiary distinction between two types of *design* of
sub-personal mechanisms: first, those that are designed like digital
computers to emulate human symbolic rule-followers; second, those whose
function is explained on other principles, e.g. those of dynamical
systems or "analog" computation. These are intended to explicate engineer's
distinctions.
It might be right that the distinction at the sub-personal level
is merely heuristic and threatens to break down if pressed at all. I mean
it to be of a piece by other distinctions intuitively used by system designers,
e.g. that between analog and digital, between pictorial and propositional
representations, between stored program and hard-wired program.
All of these are in fact very difficult to make out precisely, I believe,
although they do seem to appear in the thought of designers. (There is
a literature attempting to make out an analog/digital distinction, but
no real consensus, I think on how to do it).
>correctly, in the early Newcomen steam engine humans followed rules
>as to when to open the steam valves. They they discovered that they
>could automate this rule following. The push rods and cam shafts are
>really the modern instantiation of this way of mechanizing what
>historically began as a rule following activity. So I still say that
>you are making arbitrary distinctions.
I am taking it that digital computation is one very special sort of
design, one that invites descriptions in terms of representations and
rules for manipulating them. I take it achieving the behavior by brute
physical structures is an alternative form of design.
>It seems to me that you are in the business of creating a bogus
>pseudo-justification of your subjective opinion that computing can be
>considered rule following while the actions of the cam shaft cannot.
>But then maybe the creation of bogus pseudo-justifications is about
>all that philosophy is good for.
My major philosophical distinction is between the personal/sub-personal
level -- that is what I am making when I say nothing at the
sub-personal level is really rule-following, in Wittgenstein's sense.
But I take it that leaves open an engineer's distinction at the
sub-personal level, between designs that "follow rules" (digital computers)
and those that don't. But perhaps this does break down, that's ok with me.
My minor distinction is intended as a start on explicating the
intuitions actually employed by system designers. If it is not helpful
then I can easily drop it. It is not essential for my own purposes,
since what, if anything, happens literally inside our heads does not
seem to me to be specially relevant for philosophical inquiries into
the nature of the personal level.
I would be a little surprised if you wanted no distinction at all
between two alternative styles of control system design, say one that
uses a digital microprocessor and symbolic operations, and another that
exploits only physical structure of the parts and their dynamics.
Perhaps you want it to be made out in other terms than "rule-following"
-- since both are equally liable to description in terms of the
designed norms (rules) of proper function that we impose.
Jim Balter
Neil Rickert wrote:
>
> In <3351B5...@netcom.com> Jim Balter <j...@netcom.com> writes:
> >Neil Rickert wrote:
>
> >> implausible that understanding is a matter of rule following. But I
> >> think it plausible that understanding is a matter of being
> >> sufficiently responsive to a wide range of external inputs.
>
> >But don't we have to know what we are talking about before we can
> >talk about what it can be a matter of? And does that phrase mean
> >"is constituted of" or "is caused by" or "requires"? It doesn't seem to
> >me to be plausible that any amount of response to external inputs
> >is in itself "understanding"; that just isn't what I mean by it.
>
> good enough. Some sort of intelligent response (or adaptive
> response) would be needed.
Which leads us from one unknown to another. :-)
But couldn't you be like Henry Ford, and intelligently
respond to a wide range of inputs by delegating authority?
This gives only one sort understanding that covers many
areas. Or one can understand one thing very well. Depth vs. breadth.
I'm having trouble seeing how this fits our folk meaning
of understanding. Intelligence seems required for understanding,
and breadth seems required for intelligence, but I'm having
trouble with the transitivity; it's natural language, but not
predicate logic. :-)
> >It seems to me to mean that someone holds the right things to be true.
>
> I'm not so sure about that. It seems to me that someone could
> understand natural selection, yet not hold it to be true because of
> religious convictions. (I must admit that most of the people who
> have religious convictions against natural selection also seem to
> misunderstand it).
Excellent example (we can charitably dismiss the parenthetical,
since clearly people *can* reject the truth of what they understand).
But I don't mean "holds true" at a *conscious* level; there aren't
nearly enough propositions that we hold true at that level.
I mean something more functional. That is, given premises
(possibly evidentiary claims), the right inferences are made.
The understanding skeptic would be able to make the right
arguments. The fundamentalist who understood natural selection could
pass the test with flying colors and still mutter "but I don't believe
a whit of it". But no one could pass the test merely by being
committed to the validity of natural selection. So "the right things"
are what it takes to make the right inferences. Someone who understands
Chinese can infer the right responses to Chinese utterances, without
consciously doing much of anything. I even imagine "mutespeak",
akin to "blindsight", where someone could speak Chinese but deny
that they know what any of the words mean. In fact there is the
case of Pat Hayes, a cognitive scientist who posts to JCS-online,
who has some episodic aphasia; he has reported that there are stages
of his aphasia where he speaks but doesn't know what he is saying,
although others tell him that he spoke appropriately. His conscious
self (begging that question) doesn't understand English at those times.
but some other part of him certainly appears to; it is able to make
the right inferences from its inputs to the appropriate outputs;
some part of his brain knows what to say; understands.
> >> That
> >> sort of responsiveness seems to be what we test when we examine
> >> students to assess their understanding.
>
> >I don't see how we are testing responsiveness per se.
> >If someone holds the basic propositions to be true, then they
> >will produce the right responses whereever those propositions
> >come into play, and that translates into responsiveness.
>
> This is not obvious. Indeed, I'm not sure that we ever know what
> exactly we should consider the basic propositions, or whether there
> are such things.
That don't know what should be the basic propositions is what I was
getting at with some of my other material. What it is we are testing
for is not set, but it seems to me to be related to what inferences
*we* make; we capture some of these inferences as tests, apply them,
and see if the responses match our own. By this process we correlate
the other person's inferences to our own; the better they match,
the more we hold the other to be understanding. If I think that
someone understands Searle's argument, it would certainly seem
to be my own understanding that I think they share. If I later
decide that I don't understand it after all (I was making the
wrong inferences), but they haven't changed, I would then hold that
they *don*t understand it, *and didn't*.
> I doubt that the brain can be considered a
> repository for propositions.
Perhaps not, but I think it can be seen as an inference engine,
which my "proposition" idea was a first cut at (I'm really doing
this on the fly). Your point about denying the truth of what one
understands moved me from the absolutist "true propositions" to the
relativist/functional "correct inferences".
> Perhaps you are just saying that if
> they respond appropriately, we will ascribe to them what we believe
> to be the basic propositions.
I don't think I am *just* saying that; I'm saying something about
checking the correlation or registration between our two inference
engines. I hadn't brought in the symmetric aspect before (because
I didn't realize/recognize/understand? it), but I think it is important,
and I think you also have a relativist notion of epistemology,
don't you? If you have a debate with Anders or a Jehovah's Witness,
you each judge each other's understanding; none holds a position of
intrinsic authority.
> But that would agree with my early
> point about judging students on their responses.
I don't know that I ever disagreed, or that it matters.
I started out saying that we don't have a model of understanding,
but got a flash and started thinking that maybe there was some
hope of crafting one. I think that the idea that understanding
is a judgement or attribution (which I always thought), but that it
is a certain sort of judgement, namely of registration between
one's own inferences and those of the one being judged (this isn't
the conscious content of the judgement, but rather a third person
description of the process that produces the judgement), has some
promise. It seems to me that your notion of breadth of responsiveness
has more to do with intelligence than with understanding as I
see it, but as such is also promising. Both seem much more promising
than the sort of dichotomous "has it or doesn't" coming from the
other camp that channels energy away from constructive attempts at
analysis.
--
<J Q B>
Jim Balter
Phil Roberts, Jr. wrote:
>
> Jim Balter wrote:
> >
> > Phil Roberts, Jr. wrote:
> > >
> > > I would state this a bit differently, i.e., that there is at present no
> > > particular reason to suppose that our capacity to cognize algorhythms is
> > > itself an algorhythm, although there are probably a number of emotional
> > > reasons why strong AI advocates want to believe it must be so.
> >
> > The capacity to cognize algorithms is not an algorithm, it is a
> > capacity. However, there may well be algorithms that produce
> > systems that can do such cognizing. There are at present very good
> > reasons for thinking so, mainly the Church-Turing thesis.
> > of emotional satisfaction is itself a matter of emotional satisfaction
> > and gross ignorance.
> >
>
> Thanks Jim. I knew I could count on you for my much needed humility
> lesson. ;)
Apparently you consider anything you happen to not to agree with
or not to understand a "humility lesson" that you can smirk at. Such a
person could indeed use a big lesson in humility, but I don't think that
quality is within your reach.
Smirking is no substitute for a rebuttal, yet you will go on blathering
about AI advocates "wanting to believe" this or that, content in
your gross ignorance of why they do in fact believe this or that.
--
<J Q B>
Jim Balter
Gary Forbis wrote:
>
> Jim Balter wrote:
> >
> > > > The premise is: A machine can understand by virture of following the
> > > > rules for being an understander. He didn't choose this premise because
> > > > he believed it but because he didn't believe it.
> > > >
> > >
> > > particular reason to suppose that our capacity to cognize algorhythms is
> > > itself an algorhythm, although there are probably a number of emotional
> > > reasons why strong AI advocates want to believe it must be so.
> >
> > The capacity to cognize algorithms is not an algorithm, it is a
> > capacity. However, there may well be algorithms that produce
> > systems that can do such cognizing. There are at present very good
> > reasons for thinking so, mainly the Church-Turing thesis.
>
>
> I don't see how the Church-Turing thesis helps. Doesn't one have to
> first
> first assume such an algorithm exists before one can conclude it can be
> implemented on a Turing machine?
Perhaps you are under the false impression that the Church-Turing thesis
claims that algorithms can be implemented on a Turing machine,
rather than claiming that anything computable by a human is computable
by a machine.
--
<J Q B>
Anders N Weinstein
In article <335437...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>>
>hope of crafting one. I think that the idea that understanding
>is a judgement or attribution (which I always thought), but that it
>is a certain sort of judgement, namely of registration between
>one's own inferences and those of the one being judged (this isn't
>the conscious content of the judgement, but rather a third person
>description of the process that produces the judgement), has some
>promise.
You might be interested in the (rather involved) theory of Pitt's Bob
Brandom in his book _Making it Explicit_. He has a different theory of
what an inference is -- a move in a norm governed social practice which
determines commitments and entitlements, a bit like a game. But he
basically has an "inferential role" theory of content, in that he takes
the conceptual content of extra-logical vocabulary to be largely a
matter of the material inference rules governing them (also of the
worldly circumstances that prompt assertion).
Anyway, it is absolutely central to his account that each player
"keeps score" on the other players in *practice* (without consciously
thinking about it). This includes assessing what assertional
commitments they have undertaken and therefore what they are committed
or entitles to go on to assert. But in evaluating the others, each player
necessarily uses his own inference rules to assess these consequences.
In that way what you commit yourself to can outrun what you acknowledge
yourself to be committed to. This is crucial to Brandom's account of
why assertional practice is objective -- because from every point of
view within it there is a difference between the commitments you
acknowledge and the commitments others will hold you to.
Anyway, I think it's a fascinating account (although I have some problems
with it). I'd be interested in discussing it if anyone is interested.
Anders N Weinstein
In article <5j0unp$d...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5j0geb$g...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>In article <5iv27j$a...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>OK, rephrase: on your usage everything in nature is a rule following?
>
>I thought we were discussing only the case of the automobile. I took
>that to be a relatively uncontroversial example of rule following.
>Yet you challenged me on it.
Doesn't look uncontroversial to me. I don't know of many people who
want to say that the sparkplugs, pistons, camshafts etc are following
rules when they move. We explain them in terms of norms of design, and
in that there is a way they are *supposed* to behave. Perhaps that is
all you mean by following rules.
>Is everything rule following? I would be more inclined to say that
>nothing is rule following. That is, whether something is
>rule-following is a question of our interpretation, rather than a
>question of intrinsic properties of reality. In the same way, I
OK, I can rephrase again: is everything usefully interpreted as a
system that follows rules on your view? Clearly not, I hope. So we
still want a difference, don't we, even if it is merely a matter
of "useful interpretation".
I think I am skeptical of the distinction you are purporting to draw.
Is anything an "intrinsic" property of reality in your sense?
My view is that if you like, you are free to say that everything we
ever think of as a fact is just a matter of what is a useful
interpretation of "reality", whatever that is supposed to be.
But all such putative facts come out in the same boat.
I think most invocations of a distinction between "intrinsic"
properties and "matters of useful interpretation" are vacuous. Then
everything really comes out as "matter of useful interpretation" --
including physical science -- so there is nothing really being
distinguished. I find it simpler to call them all objective facts.
>would say that nature is not following the laws of physics, but that
>we find it convenient and predictively efficacious to interpret
>nature as if it were following the laws of physics. (Incidently, I
>don't see anything intrinsic in intentionality either.)
Perhaps there is nothing intrinsic in anything. It doesn't seem to
matter whether something is intrinsic or not as long as there are statements
we can make that are as objectively constrained, as factual, as
anything we say ever gets.
>>Or is it only a system such that there are design norms dictating what
>>it is supposed to do?
>
>I would paraphrase that as "Or is it only a system that we choose to
>interpret as rule following?"
Too simple, our interpretations can be wrong.
If a device structurally similar to a calculator fell from the
sky, it might be useful to interpret it as a calculator, but we could
find out it is not really a calculator (although usable as such) but
something else, since it was produced by the Martians with different
intentions. In that sense this fact is not *just* a matter of what is useful
to us, it is a very complex socially determined fact in which Martian
practices and intentions are involved.
>>I am reminded of Kant's claim that although everything in nature act in
>>*accordance* with rules (laws of nature),
>
>I'm not sure I agree with that. I don't know that there really are
>any laws of nature. To put it differently, nature is under no
>obligation to follow our scientific laws. Rather, the obligation is
>on us to design laws which fit nature as closely as is reasonably
>possible.
In some fashion Kant indeed thought we were "under the obligation" to
cognize nature by imposing laws in its representation. That is
basically his definition of "nature" -- a system of appearances in
accordance with laws, or something like that. He thought there must be
lawfulness in nature because we must put it there.
>>I think we want some distinction on pain of trivialization of the claim.
>
>I don't see any problem of trivialization arising. If "rule
>following" is a matter of useful interpretation, then such a claim is
>non-trivial to the extent that it is useful. As long as interpreting
Right. And that is why we need a distinction -- for it is not particularly
useful in the case of water flowing downhill.
There is still a question of why it should be useful. After all, it
licenses no new predictions about the system to be told it is following
rules. What does it really add, in your view?
> As long as treating the automobile motor as a rule
>following device allows us to adjust (and thus program) the cam shaft
>to maximize the operating efficiency of the motor, then our
>interpretation has demonstrated its usefulness.
OK. But I don't see tuning the properties of cam shaft as helpfully
compared to "programming" it to follow a different rule. If I reshape
the body for less wind drag, am I really "programming it"? The case
of tuning an engine seems similar to me.
>>Recall I recommend keeping in mind two distinctions: first is the
>>personal/sub-personal level distinction, the distinction between a
>>whole person following a public, institutionalized rule at the social
>>behavioral level; and a sub [or non]-personal mechanism like a computer
>>or bit of functional neural circuitry "following" a rule at its level.
>>That is a philosopher's conceptual distinction, in my view.
>
>The main distinction, from my point of view, is that interpreting a
>computer as following a rule is far more predictively efficaceous
>than is interpreting a whole person as following a rule at the social
>level. Generally, I think people are poor rule followers.
That's as may be, it is irrelevant to the idea whether people are
specially good or bad at it.
The claim is that only people are rule-followers at all, and this claim
has nothing special to do with predicting what noises and motions they
will make, it has rather to do with constituting their acts as
meaningful at all.
>>The second is a subsidiary distinction between two types of *design* of
>>sub-personal mechanisms: first, those that are designed like digital
>>computers to emulate human symbolic rule-followers; second, those whose
>>function is explained on other principles, e.g. those of dynamical
>>systems or "analog" computation. These are intended to explicate engineer's
>>distinctions.
>
>I would characterize that as largely a distinction without a
>difference. In the first place, everything is a dynamical system,
>including computers. Moreover, the digital computer is an analogue
Granted. But not everything is "usefully explained (interpreted)" in
terms of computational as opposed to dynamical concepts.
>But beyond that, we don't always design digital systems so as to
>emulate human symbolic rule followers. In the CD music recording
>system, we use digital computers in a way which emulates and improves
>on analog recording methods, but which does not come close to
>emulating more typical uses of symbols by humans. The choice of
Well it is not a typical "application" for human rule followers, but of
course it *does* involve the kind of emulation of a symbolic
rule-follower in my sense -- insofar as it is *digital* computation at
all. For that means it is explained as doing rule-governed computations
over discrete symbolic representations of acoustic signals.
Our *purpose* in designing this system is ultimately to get good sound out
of the process. But the digital part is still like a little man doing
sums or long division in this sense, it is pushing around discrete
symbols in accordance with formal rules.
>digital over analog technology is largely a pragmatic one. We choose
Of course, of course. I am just trying to rely on the
idea that there *is* a difference.
> The choice of
>analog methods because they offer advantages in reliability and
>precision. Presumably our languages are digital systems (with the
>phonemes considered digits) because nature too has discovered those
>pragmatic advantages of digital systems.
I presume you meant "digital".
>>It might be right that the distinction at the sub-personal level
>>is merely heuristic and threatens to break down if pressed at all.
>
>I would say that the distinction between the personal and subpersonal
>levels is mostly a matter of putting ourselves on a pedestal as part
>of our species-centric way of looking at things. The distinction
>between analog and digital at the personal level is mostly a matter
I never made any such a distinction at the personal level. The symbolic
activity of human beings in the world would seem to be *neither*
"digital" nor "analog".
For example, the "slab" game uses discrete symbols and uses them to
stand for ditinct clearly demarcated objects. So is it "digital"?
But playing it also
involves the ability to move reach and pick up and carry the building
stones. Which would seem to involve continuously varying quantities (or
processes best modelled that way), hence an "analog" component.
I tend to think of our operations with discrete meaningful public
symbols as floating on and modulating a background flow of non-symbolic
activity. Because you can't have one without the other it does not
seem to make sense to ask "analog" or "digital". These concepts just
don't apply to the whole process. They might apply to the
representations that obtain their significance by virtue of being
caught up in and playing some role within it, these then are digital
but that is not the whole story about their significance.
I think we differ from the animals in that such things a spublic
symbols have come to play a role in our lives, in ways continuous with
but infinitely more complex than any animal signalling systems.
>> (There is
>>a literature attempting to make out an analog/digital distinction, but
>>no real consensus, I think on how to do it).
>
>If there is no real consensus, I would take that as supportive of my
>view that there is less of a difference than meets the eye.
You used the analog/digital distinction yourself and remarked on pragmatic
reasons for choosing one over the other. Evidently you do think
there is a difference.
>>I am taking it that digital computation is one very special sort of
>>design, one that invites descriptions in terms of representations and
>>rules for manipulating them. I take it achieving the behavior by brute
>>physical structures is an alternative form of design.
>
>But our computers work only by virtue of the behavior of brute
>physical structures.
I don't think anyone is denying this, so I don't know why you feel the
need to keep reminding me of this.
>physical structures. Again, I think you put digital computation and
>symbolic processes on too high a pedestal.
I am not putting it on a pedestal, I am just trying to demarcate it
as a very distinctive thing. Fodor, or Haugeland expounding GOFAI, or
Newell and Simon are right to say it is a very special new idea.
E.g. the Physical Symbol System hypothesis around which a lot of
cognitive science research was and still is organized would have be
*false* if it turned out if the function of the neural circuitry was
best explained as an "analog" computer or in terms of the concepts
of dynamical systems. Those are *alternatives* or *rivals* to N+S's
hypothesis. That is all I mean.
>>I would be a little surprised if you wanted no distinction at all
>>between two alternative styles of control system design, say one that
>>uses a digital microprocessor and symbolic operations, and another that
>>exploits only physical structure of the parts and their dynamics.
>
>In my studies of digital electronics, I have not come upon a digital
>microprocessor which did not work by exploiting the physical
>structure of the parts and their dynamics. I think the only
Right, who would deny that? But not every physical system is explained
as a digital computer.
>distinctions are pragmatic. And, yes, there are solid pragmatic
>grounds for making distinctions.
Then we are in agreement, so I don't see your beef. I said I was only
interested, insofar as I was speaking about sub-personal
design-stance stuff, in what was pragmatically motivated.
Neil Rickert
In <335437...@netcom.com> Jim Balter <j...@netcom.com> writes:
>Neil Rickert wrote:
>> Ok, I was not precise enough. Responding any which way would not be
>> good enough. Some sort of intelligent response (or adaptive
>> response) would be needed.
>Which leads us from one unknown to another. :-)
>But couldn't you be like Henry Ford, and intelligently
>respond to a wide range of inputs by delegating authority?
Sure. I would say that would be adaptive -- adapting to the
availability of people to whom you can delegate responsibility.
>> I'm not so sure about that. It seems to me that someone could
>> understand natural selection, yet not hold it to be true because of
>> religious convictions. (I must admit that most of the people who
>> have religious convictions against natural selection also seem to
>> misunderstand it).
>Excellent example (we can charitably dismiss the parenthetical,
>since clearly people *can* reject the truth of what they understand).
>But I don't mean "holds true" at a *conscious* level; there aren't
>nearly enough propositions that we hold true at that level.
>I mean something more functional. That is, given premises
>(possibly evidentiary claims), the right inferences are made.
Ok. I can agree with that. But I am not sure that "holding true" is
the right standard at the functional level. You are really talking
about abilities rather than beliefs. That may include the ability to
make certain types of inference, of course.
> In fact there is the
>case of Pat Hayes, a cognitive scientist who posts to JCS-online,
>who has some episodic aphasia; he has reported that there are stages
>of his aphasia where he speaks but doesn't know what he is saying,
>although others tell him that he spoke appropriately. His conscious
>self (begging that question) doesn't understand English at those times.
>but some other part of him certainly appears to; it is able to make
>the right inferences from its inputs to the appropriate outputs;
>some part of his brain knows what to say; understands.
Yes, I followed much of that discussion. I found it unsurprising. I
have never thought that my reasoning was linguistic. Rather, it has
always seemed that the linguistic expression is merely coming along
for a free ride. I suppose that is part of my skepticism about the
way philosophy uses logic. Pinker uses "The Language Instinct" to
refer to a presumed grammatical engine inside. For me, "the language
instinct" should refer to our habit of attaching language to our
activities, including our non-linguistic thinking.
>> >I don't see how we are testing responsiveness per se.
>> >If someone holds the basic propositions to be true, then they
>> >will produce the right responses whereever those propositions
>> >come into play, and that translates into responsiveness.
>> This is not obvious. Indeed, I'm not sure that we ever know what
>> exactly we should consider the basic propositions, or whether there
>> are such things.
>That don't know what should be the basic propositions is what I was
>getting at with some of my other material. What it is we are testing
>for is not set, but it seems to me to be related to what inferences
>*we* make; we capture some of these inferences as tests, apply them,
>and see if the responses match our own.
That sounds ok. Let me give an illustration from computer science.
We might have two different compilers for the same language (say
Pascal, for example). One is based on an LL(1) parser, while the
other is based on an LR(1) or SLR(1) parser (say generated by
'yacc'). The LL(1) compiler was built using an LL(1) grammar for the
language, while the LR compiler used an LR(1) grammar. Thus the two
compilers may be based on different grammars for the same language.
In terms of the way philosophers talk, we might say that the grammars
are the conceptual structures of the compilers, and the productions
in those grammars are the general statements (beliefs) used by the
compilers to conceptualize their input (pascal programs). Thus the
two compilers are based on different, and perhaps incompatible belief
systems. But they still correctly compile the same pascal programs.
So I want to measure the knowledge of the compilers as their
abilities to compile programs, rather than as their systems of belief
(the productions in their grammars). And, incidently, I think it
reasonable to say that the compiler understands the programs that it
correctly compiles.
>> I doubt that the brain can be considered a
>> repository for propositions.
>Perhaps not, but I think it can be seen as an inference engine,
>which my "proposition" idea was a first cut at (I'm really doing
>this on the fly). Your point about denying the truth of what one
>understands moved me from the absolutist "true propositions" to the
>relativist/functional "correct inferences".
That sounds reasonable.
>> Perhaps you are just saying that if
>> they respond appropriately, we will ascribe to them what we believe
>> to be the basic propositions.
>I don't think I am *just* saying that; I'm saying something about
>checking the correlation or registration between our two inference
>engines. I hadn't brought in the symmetric aspect before (because
>I didn't realize/recognize/understand? it), but I think it is important,
>and I think you also have a relativist notion of epistemology,
>don't you?
Right. But, as you probably realize, my mild type of relativism
should not be equated with the radical relativism of someone like
Feyerabend.
> If you have a debate with Anders or a Jehovah's Witness,
>you each judge each other's understanding; none holds a position of
>intrinsic authority.
Right.
>> But that would agree with my early
>> point about judging students on their responses.
>I don't know that I ever disagreed, or that it matters.
>I started out saying that we don't have a model of understanding,
>but got a flash and started thinking that maybe there was some
>hope of crafting one.
Let me suggest one. To understand X, is to be able to pick up and
use the information arising from X. This makes 'understanding' a
matter of degree, since the amount of information picked up is a
matter of degree. It is also why I suggested that the compiler
understand the program it is compiling. But perhaps an optimizing
compiler understands the program better than a simple one-pass
compiler, and it bases its optimization on the additional information
it has been able to recover from the source program during its
optimizing phase.
> I think that the idea that understanding
>is a judgement or attribution (which I always thought), but that it
>is a certain sort of judgement, namely of registration between
>one's own inferences and those of the one being judged (this isn't
>the conscious content of the judgement, but rather a third person
>description of the process that produces the judgement), has some
>promise.
That sounds ok. Your inferences are a measure of how much
information you are picking up and are able to use, and then you
compare this with a similar measure of the person you are judging.
> It seems to me that your notion of breadth of responsiveness
>has more to do with intelligence than with understanding as I
>see it, but as such is also promising.
I would suggest that intelligence and understanding are related. I
suggest that intelligence particularly has to do with the ability to
learn -- that is, the ability to increase one's understanding -- by
discovering previously unrecognized information.
> Both seem much more promising
>than the sort of dichotomous "has it or doesn't" coming from the
>other camp that channels energy away from constructive attempts at
>analysis.
Agreed.
Neil Rickert
In <5j17ko$k...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5j0unp$d...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>Is everything rule following? I would be more inclined to say that
>>nothing is rule following. That is, whether something is
>>rule-following is a question of our interpretation, rather than a
>>question of intrinsic properties of reality. In the same way, I
>OK, I can rephrase again: is everything usefully interpreted as a
>system that follows rules on your view? Clearly not, I hope.
No, I would not suggest everything is usefully interpreted as rule
following. But what is considered useful will depend on what are
your aims. So we should not expect universal agreement on what
should count as rule following.
>I think I am skeptical of the distinction you are purporting to draw.
>Is anything an "intrinsic" property of reality in your sense?
I would suppose that it is unknowable as to what is intrinsic.
>My view is that if you like, you are free to say that everything we
>ever think of as a fact is just a matter of what is a useful
>interpretation of "reality", whatever that is supposed to be.
But then scientific changes, such as those brought about by
Copernicus and Einstein, are changes in interpretation. This seems
right to me.
>Too simple, our interpretations can be wrong.
>If a device structurally similar to a calculator fell from the
>sky, it might be useful to interpret it as a calculator, but we could
>find out it is not really a calculator (although usable as such) but
>something else, since it was produced by the Martians with different
>intentions.
As long as we were using it as a calculator, it was a calculator.
What happened to "meaning is use" that you so recently supported. So
I see nothing wrong with such an interpretation. We may later
discover an even better interpretation. I think the standard for
interpretation is one of usefulness, rather than right or wrong.
>> As long as treating the automobile motor as a rule
>>following device allows us to adjust (and thus program) the cam shaft
>>to maximize the operating efficiency of the motor, then our
>>interpretation has demonstrated its usefulness.
>OK. But I don't see tuning the properties of cam shaft as helpfully
>compared to "programming" it to follow a different rule.
We could be changing from "open the valve at 5 degrees before top
dead center" to "open the valve at 5 degrees after top dead center".
That sounds like changing the rule as far as I am concerned.
>The claim is that only people are rule-followers at all, and this claim
>has nothing special to do with predicting what noises and motions they
>will make, it has rather to do with constituting their acts as
>meaningful at all.
This seems to me to be a strange sense of "rule following", and a
strange sense of "meaningful."
>Granted. But not everything is "usefully explained (interpreted)" in
>terms of computational as opposed to dynamical concepts.
But not everything is usefully explained as a dynamical system
either. The attribution "dynamical system" by itself is too vague to
be of much use. The motion of the planets is better explained by
Newton's laws, than by designating the solar system as a dynamical
system.
>>But beyond that, we don't always design digital systems so as to
>>emulate human symbolic rule followers. In the CD music recording
>>system, we use digital computers in a way which emulates and improves
>>on analog recording methods, but which does not come close to
>>emulating more typical uses of symbols by humans. The choice of
>Well it is not a typical "application" for human rule followers, but of
>course it *does* involve the kind of emulation of a symbolic
>rule-follower in my sense -- insofar as it is *digital* computation at
>all. For that means it is explained as doing rule-governed computations
>over discrete symbolic representations of acoustic signals.
>Our *purpose* in designing this system is ultimately to get good sound out
>of the process. But the digital part is still like a little man doing
>sums or long division in this sense, it is pushing around discrete
>symbols in accordance with formal rules.
However, there is a big difference. The sort of problem that has to
be solved by the CD sound system is really a continuous problem,
rather than a discrete problem. Logic is primarily the tool for the
analysis of discrete problems. Entirely different methods are
available for analyzing continuous phenomena. The CD sound system in
some sense is built upon the type of analysis appropriate for
continuous systems. The digital processing is merely used as a
convenience at the implementation level because of the reliability
and precision that it provides.
>>I would say that the distinction between the personal and subpersonal
>>levels is mostly a matter of putting ourselves on a pedestal as part
>>of our species-centric way of looking at things. The distinction
>>between analog and digital at the personal level is mostly a matter
>I never made any such a distinction at the personal level. The symbolic
>activity of human beings in the world would seem to be *neither*
>"digital" nor "analog".
Symbols are discrete entities. Symbol processing is digital, in any
important sense of the term.
>For example, the "slab" game uses discrete symbols and uses them to
>stand for ditinct clearly demarcated objects. So is it "digital"?
>But playing it also
>involves the ability to move reach and pick up and carry the building
>stones. Which would seem to involve continuously varying quantities (or
>processes best modelled that way), hence an "analog" component.
Most of what faces us in our daily activity involves continuous
components. Logic is not the best tool to handle those
circumstances. This is perhaps why I am so cynical about the way
philosophy uses logic. Induction, for example, is posed as a
discrete symbolic operation. Thus we hear of the logic of induction,
when it should be the illogic of induction. The methods for
advancing science depend much more on finding continuity, and using
interpolation and a limited degree of extrapolation from measurements
of continuous data. The importance of such interpolation and
extrapolation seems to be missed in the usual philosophical
discussions of scientific discovery.
>I tend to think of our operations with discrete meaningful public
>symbols as floating on and modulating a background flow of non-symbolic
>activity.
But then you should be concerned about that non-symbolic activity,
and the methods we use to introduce symbols as a way of solving
non-symbolic problems. But whenever I try to discuss those things,
you try to turn everything back into symbolic activity.
>I think we differ from the animals in that such things a spublic
>symbols have come to play a role in our lives, in ways continuous with
>but infinitely more complex than any animal signalling systems.
>You used the analog/digital distinction yourself and remarked on pragmatic
>reasons for choosing one over the other. Evidently you do think
>there is a difference.
I think one should examine the problem. Discrete problems call for
discrete mathematics, including logic, in the methods of solution.
Continuous problems call for the methodologies of continuous
mathematics (such as geometry) for their solution. Ultimately, at
the final implementation stage, there is an independent choice as to
whether the solution will be implemented with a digital technology or
an analog technology.
>> Again, I think you put digital computation and
>>symbolic processes on too high a pedestal.
>I am not putting it on a pedestal, I am just trying to demarcate it
>as a very distinctive thing. Fodor, or Haugeland expounding GOFAI, or
>Newell and Simon are right to say it is a very special new idea.
>E.g. the Physical Symbol System hypothesis around which a lot of
>cognitive science research was and still is organized would have be
>*false* if it turned out if the function of the neural circuitry was
>best explained as an "analog" computer or in terms of the concepts
>of dynamical systems. Those are *alternatives* or *rivals* to N+S's
>hypothesis. That is all I mean.
I think you are drawing the wrong conclusions. We have to
distinguish the nature of the problem from the nature of the
solution. The trouble with much of GOFAI and much of cognitive
science, is that it has supposed that there is a symbolic or discrete
problem to be solved. And it has then attempted to use the methods
of discrete mathematics to solve this presumed discrete problem. But
if the problem to be solved is primarily a continuous one, then the
analysis is wrong. We should be looking at the methods of solving
continuous problems. Once we have done that, we can separately
choose whether we should use digital or analog methods at the
implementation level.
I take it that the brain is using continuous methods to solve
continuous problems, but is using a digital implementation of those
methods of solution. So this would not contradict the physical
symbol hypothesis at the implementation level, but would suggest that
it should not be relied on at the analysis level.
Anders N Weinstein
In article <5j37ho$g...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5j17ko$k...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>I think I am skeptical of the distinction you are purporting to draw.
>>Is anything an "intrinsic" property of reality in your sense?
>
>I would suppose that it is unknowable as to what is intrinsic.
So what is the point of a distinction if it is unknowable whether
anything falls on one side? It fails to (knowably) distinguish anything.
For example, it does not distinguish the understanding attained by
the sciences of nature from that attained by the disciplines
that study culture, at least not in point of getting at something
"intrinsic".
>>My view is that if you like, you are free to say that everything we
>>ever think of as a fact is just a matter of what is a useful
>>interpretation of "reality", whatever that is supposed to be.
>
>But then scientific changes, such as those brought about by
>Copernicus and Einstein, are changes in interpretation. This seems
>right to me.
Sure, but they can also be as much of a factual discovery as anything we
say ever gets to be.
>>If a device structurally similar to a calculator fell from the
>>sky, it might be useful to interpret it as a calculator, but we could
>>find out it is not really a calculator (although usable as such) but
>>something else, since it was produced by the Martians with different
>>intentions.
>
>As long as we were using it as a calculator, it was a calculator.
Perhaps if we began to use it as a calculator it would *become* a
calculator by virtue of being caught up in our practices. But as I
have been trying to argue against Pindor, there is still the
anthropologist's question "what is it" said of an *artifact*, which
is not just a matter of its being usefully interpreted as one by us.
(It might be said to involve an interpretation of a different kind,
namely of the practices and processes that surrounded its production).
>What happened to "meaning is use" that you so recently supported. So
I am trying to respect that -- what it is as a Martian artifact is
largely a matter of its role in Martian practices.
>I see nothing wrong with such an interpretation. We may later
>discover an even better interpretation. I think the standard for
>interpretation is one of usefulness, rather than right or wrong.
That is a standard for one kind of interpretation, but it is the wrong
one if you are trying to speak about an artifact. I take it facts about
artifactual computers are social facts. The idea that parts of the
brain are *natural* computers is comparable to the claim that the heart
is a natural pump, that it has this function by virtue of its role in
maintaining the life of the organism of which it is a part.
>>Granted. But not everything is "usefully explained (interpreted)" in
>>terms of computational as opposed to dynamical concepts.
>
>But not everything is usefully explained as a dynamical system
>either. The attribution "dynamical system" by itself is too vague to
>be of much use.
Of course, I was alluding to the sort of example Van Gelder pointed to
in his paper and intro to his co-edited anthology, _Mind as Motion_. I
take it the particular explanations within this particular school *are*
fleshed out in more substantial ways.
Again, the point is only to delineate an alternative to the symbolic
computational paradigm.
>>>But beyond that, we don't always design digital systems so as to
>>>emulate human symbolic rule followers. In the CD music recording
>>>system, we use digital computers in a way which emulates and improves
>>>on analog recording methods, but which does not come close to
>>>emulating more typical uses of symbols by humans. The choice of
>
>>Well it is not a typical "application" for human rule followers, but of
>>course it *does* involve the kind of emulation of a symbolic
>>rule-follower in my sense -- insofar as it is *digital* computation at
>>all.
>
>However, there is a big difference. The sort of problem that has to
>be solved by the CD sound system is really a continuous problem,
>rather than a discrete problem. Logic is primarily the tool for the
>analysis of discrete problems. Entirely different methods are
>available for analyzing continuous phenomena. The CD sound system in
I think this use of "logic" is threatening to get a little too broad.
After all, it is not as if the algorithms employed by a
digital signal processor are not "logical" or lack semantics.
>some sense is built upon the type of analysis appropriate for
>continuous systems. The digital processing is merely used as a
>convenience at the implementation level because of the reliability
>and precision that it provides.
OK, I see what you mean. That particular choice is at the implementation
level with respect to the continuous problem to be solved, yes.
Perhaps I ought to rest on the distinction between "symbolic"
computation and non-symbolic (but possibly digital) computation,
depending on whether the discrete symbols have semantics at the level
of the cognitive domain.
>>>I would say that the distinction between the personal and subpersonal
>>>levels is mostly a matter of putting ourselves on a pedestal as part
>>>of our species-centric way of looking at things. The distinction
>>>between analog and digital at the personal level is mostly a matter
>
>>I never made any such a distinction at the personal level. The symbolic
>>activity of human beings in the world would seem to be *neither*
>>"digital" nor "analog".
>
>Symbols are discrete entities. Symbol processing is digital, in any
>important sense of the term.
If I write discrete symbols on some discrete cards and juggle them, am
I doing something 'digital'?
>>For example, the "slab" game uses discrete symbols and uses them to
>>stand for ditinct clearly demarcated objects. So is it "digital"?
>>But playing it also
>>involves the ability to move reach and pick up and carry the building
>>stones. Which would seem to involve continuously varying quantities (or
>>processes best modelled that way), hence an "analog" component.
>
>Most of what faces us in our daily activity involves continuous
>components. Logic is not the best tool to handle those
>circumstances. This is perhaps why I am so cynical about the way
This is a little misleading -- it suggests that logic is a kind of tool
for solving problems to be discarded when it is not useful. I don't
think of it that way at all.
It might be better to say that representation and reasoning are not the
best approaches to practical problems, e.g. walking to the next room,
or catching a fly ball. But that is not because one is somehow being
"illogical" when one does that, it is rather because walking or
catching a fly ball is a bodily skill, not a cognitive problem.
> Induction, for example, is posed as a
>discrete symbolic operation. Thus we hear of the logic of induction,
>when it should be the illogic of induction. The methods for
>advancing science depend much more on finding continuity, and using
>interpolation and a limited degree of extrapolation from measurements
>of continuous data. The importance of such interpolation and
>extrapolation seems to be missed in the usual philosophical
>discussions of scientific discovery.
Well that sure can't solve the "problem of induction", if there is one,
since it can't get you a reasonable inference from what has
been effective in the past to what will continue to be effective
in the future without relying on induction. So I am not sure what
you are getting at.
It is true, I think, that we don't walk confidently without thinking
because we have ourselves inductively confirmed that the ground
is generally solid. We generally just walk and are in a world in
which that habit works well enough.
>>I tend to think of our operations with discrete meaningful public
>>symbols as floating on and modulating a background flow of non-symbolic
>>activity.
>
>But then you should be concerned about that non-symbolic activity,
I am, I think.
>and the methods we use to introduce symbols as a way of solving
>non-symbolic problems. But whenever I try to discuss those things,
Again, I think I am. But that involves using the symbols as
representations.
> But whenever I try to discuss those things,
>you try to turn everything back into symbolic activity.
I don't think that is right. It is rather that I don't think this
can be addressed at the sub-personal level, the level of the
brain's eye view.
E.g. you spoke about the axioms of Euclidean geometry as a set of
rules for the interpretation of measurement results. That speaks about
the activities of human beings in a world with other people, with
measuring devices, and with certain public symbols functioning as
rules in the context of this activity. The practices can be taught,
and new learners can be trained and tested in the operation of this
symbolic technique.
In that sense, the symbols that serve as an expression of a rule here
would be cultural or social objects. They have their being or function
by virtue of their role in a social medium. That seems to me to be the
right level at which to speak about rules, processes of measurement and
techniques of application. (Quinean worries about the
analytic/synthetic disctinction aside).
>>You used the analog/digital distinction yourself and remarked on pragmatic
>>reasons for choosing one over the other. Evidently you do think
>>there is a difference.
>
>I think one should examine the problem. Discrete problems call for
>discrete mathematics, including logic, in the methods of solution.
>Continuous problems call for the methodologies of continuous
>mathematics (such as geometry) for their solution. Ultimately, at
This dichotomy seems sloppy to me -- it is not as though you could do
geometry or analysis without using logic.
>I think you are drawing the wrong conclusions. We have to
>distinguish the nature of the problem from the nature of the
>solution. The trouble with much of GOFAI and much of cognitive
>science, is that it has supposed that there is a symbolic or discrete
>problem to be solved. And it has then attempted to use the methods
>of discrete mathematics to solve this presumed discrete problem. But
>if the problem to be solved is primarily a continuous one, then the
>analysis is wrong. We should be looking at the methods of solving
>continuous problems.
Again, note you can solve "continuous problems" symbolically using
the maligned "logic" (plus non-logical axioms, of course).
Part of the difficulty here, I think, is an ambiguity in what we mean
by "solving a problem". *Designing* something to, say, amplify a signal
might be a problem for an engineer, in which some reasoning is
involved, even if it is only to make trial and error attempts, and for
which representations are likely to be deployed.
The amplifier itself, whether analog or digital, does not have to solve
a problem in this sense, it just has to amplify and so embody or
instantiate a solution to the different problem of "how to amplify".
But then it is not itself using continuous mathematics as opposed to
discrete "logic", it is rather designed to exploit such properties as
continuous mathematics is appropriate for. But it is *we* who use
continuous mathematics in analyzing and explaining it.
>I take it that the brain is using continuous methods to solve
>continuous problems, but is using a digital implementation of those
>methods of solution. So this would not contradict the physical
I wonder why you think the second?
> So this would not contradict the physical
>symbol hypothesis at the implementation level, but would suggest that
>it should not be relied on at the analysis level.
Hmmm. I expect the PSS as intended involved some further assumptions
about the sorts of semantics that the symbols have. E.g. if the digital
computations involved in chess playing do not directly represent chess
positions but rather represent acoustic properties of words describing
chess positions, then I think the PSS is undermined in spirit.
And it certainly suggests that the digital symbols are not *necessary*
for intelligent behavior (for that is only an implementation choice,
others might be possible), so I think it undermines the letter as well.
Neil Rickert
In <5j3v2b$3...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5j37ho$g...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>>If a device structurally similar to a calculator fell from the
>>>sky, it might be useful to interpret it as a calculator, but we could
>>>find out it is not really a calculator (although usable as such) but
>>>something else, since it was produced by the Martians with different
>>>intentions.
>>As long as we were using it as a calculator, it was a calculator.
>Perhaps if we began to use it as a calculator it would *become* a
>calculator by virtue of being caught up in our practices. But as I
>have been trying to argue against Pindor, there is still the
>anthropologist's question "what is it" said of an *artifact*, which
>is not just a matter of its being usefully interpreted as one by us.
In my view, you are relying on sloppy thinking.
>>I see nothing wrong with such an interpretation. We may later
>>discover an even better interpretation. I think the standard for
>>interpretation is one of usefulness, rather than right or wrong.
>That is a standard for one kind of interpretation, but it is the wrong
>one if you are trying to speak about an artifact.
Then what is your definition of "artifact"? Is a bird nest an
artifact? Is a beaver's dam an artifact? Perhaps we are just wrong
about what is a bird's nest, and we should try to join the society of
birds, to see how such an artifact is used in the customs of that
society.
A Martian, examining humans, might think that our computers are not
artifacts, but are just as much a part of nature as we think bird's
nests to be.
>>However, there is a big difference. The sort of problem that has to
>>be solved by the CD sound system is really a continuous problem,
>>rather than a discrete problem. Logic is primarily the tool for the
>>analysis of discrete problems. Entirely different methods are
>>available for analyzing continuous phenomena.
>I think this use of "logic" is threatening to get a little too broad.
>After all, it is not as if the algorithms employed by a
>digital signal processor are not "logical" or lack semantics.
The word "logical" is used either to mean something like reasoning by
form or syntax, or alternatively to mean "sensible." I have been
using it in the first sense. Now I'll grant that CD systems use
logic chips, so they are using logic at the implementation level.
But the analysis I was talking about is that which went into
designing the system. That was largely an analysis based on
continuous methods, including Fourier analysis.
>> The CD sound system in
>>some sense is built upon the type of analysis appropriate for
>>continuous systems. The digital processing is merely used as a
>>convenience at the implementation level because of the reliability
>>and precision that it provides.
>OK, I see what you mean. That particular choice is at the implementation
>level with respect to the continuous problem to be solved, yes.
>Perhaps I ought to rest on the distinction between "symbolic"
>computation and non-symbolic (but possibly digital) computation,
>depending on whether the discrete symbols have semantics at the level
>of the cognitive domain.
You are hopelessly unscientific and subjectivist. There is no reason
to take symbols at the cognitive level as any more special than any
other symbols. No wonder you oppose reductionism, for it is
inconsistent with your subjectism. Scientists ought to be able to
look at human use of symbols from a detached perspective, and see how
they are used to solve problems. Looked at in that way, I suggest
that the cognitive level symbols are best seen as part of a strategy
for solving continuous problems.
>>Symbols are discrete entities. Symbol processing is digital, in any
>>important sense of the term.
>If I write discrete symbols on some discrete cards and juggle them, am
>I doing something 'digital'?
Are you juggling the symbols? Or are you juggling the cards, and the
symbols are merely going along for a ride?
>>Most of what faces us in our daily activity involves continuous
>>components. Logic is not the best tool to handle those
>>circumstances. This is perhaps why I am so cynical about the way
>This is a little misleading -- it suggests that logic is a kind of tool
>for solving problems to be discarded when it is not useful.
Quite right.
> I don't
>think of it that way at all.
I don't doubt that at all. You have made logic into a religion.
>It might be better to say that representation and reasoning are not the
>best approaches to practical problems, e.g. walking to the next room,
>or catching a fly ball.
I would not suggest that for one moment. When I walk to the next
room, my neurons are working with representations. I expect that a
baseball player is reasoning when deciding which way to run in order
to catch that fly ball. Perhaps the player does not attach a string
of words to his reasoning, and that makes it incompatible with your
religion of logic. But it is reasoning, nonetheless.
> But that is not because one is somehow being
>"illogical" when one does that, it is rather because walking or
>catching a fly ball is a bodily skill, not a cognitive problem.
Again, you exhibit your limited binary thinking. To say that one is
not using logic is not to say that one is illogical.
>> Induction, for example, is posed as a
>>discrete symbolic operation. Thus we hear of the logic of induction,
>>when it should be the illogic of induction. The methods for
>>advancing science depend much more on finding continuity, and using
>>interpolation and a limited degree of extrapolation from measurements
>>of continuous data. The importance of such interpolation and
>>extrapolation seems to be missed in the usual philosophical
>>discussions of scientific discovery.
>Well that sure can't solve the "problem of induction", if there is one,
>since it can't get you a reasonable inference from what has
>been effective in the past to what will continue to be effective
>in the future without relying on induction. So I am not sure what
>you are getting at.
Going from the past to the future is an example of extrapolation.
Induction treats that as a discrete problem, and thus assumes that
one can do extrapolation willy nilly. It is not so. We can
reasonably expect the sequence of day and night in the future to be
very much like what it has been in the past. We cannot be nearly so
sure about the sequence of weather changes. The difference is that
there is a great deal more continuity in the sequence of day and
night than in the sequence of weather changes.
>It is true, I think, that we don't walk confidently without thinking
>because we have ourselves inductively confirmed that the ground
>is generally solid. We generally just walk and are in a world in
>which that habit works well enough.
More nonsense arising from discrete thinking. We walk confidently
because we have learned effective ways of judging how solid the
ground is. We do not walk as confidently on soft mushy ground as we
do on solid ground. When there is ice on the ground, we take a great
deal more care about how we walk.
>>But then you should be concerned about that non-symbolic activity,
>I am, I think.
I would not guess it from what you write.
>E.g. you spoke about the axioms of Euclidean geometry as a set of
>rules for the interpretation of measurement results. That speaks about
>the activities of human beings in a world with other people, with
>measuring devices, and with certain public symbols functioning as
>rules in the context of this activity. The practices can be taught,
>and new learners can be trained and tested in the operation of this
>symbolic technique.
>In that sense, the symbols that serve as an expression of a rule here
>would be cultural or social objects. They have their being or function
>by virtue of their role in a social medium.
More mushy thinking. The symbols and measuring rules have their
being or function because they have been shown to have a high degree
of reliability (or repeatability of measurement), and have been found
useful. That would be true, regardless of whether there was a
society. Doubtless we have a richer set of measuring abilities
available because they can be transmitted through the culture. But
what makes them importance is not the cultural role, but their
reliability and effectiveness.
> That seems to me to be the
>right level at which to speak about rules, processes of measurement and
>techniques of application.
No doubt it is the right level for mushy thinkers who are only
interested in concocting nice sounding "Just So" stories.
>>I think one should examine the problem. Discrete problems call for
>>discrete mathematics, including logic, in the methods of solution.
>>Continuous problems call for the methodologies of continuous
>>mathematics (such as geometry) for their solution. Ultimately, at
>This dichotomy seems sloppy to me -- it is not as though you could do
>geometry or analysis without using logic.
You miss the point. One can use logic in pure mathematics or in
theoretical physics, without any difficulty. This is because pure
mathematics and theoretical physics deals with an idealized
theoretical world, rather than the real world. The idealization was
done so as to make it safe for logic. But when we deal with ordinary
real world problems, the use of logic must be greatly constrained.
Otherwise problems such as Zeno's paradox, and the sorites paradoxes
pop up all over the place. In such circumstances, a reliance on
logic would lead to ridiculous conclusions. The field of numerical
analysis, within mathematics, is based on finding ways of controlling
the errors that can arise, so that one can have reliable
methodologies.
>Again, note you can solve "continuous problems" symbolically using
>the maligned "logic" (plus non-logical axioms, of course).
You can solve idealized continuous problems using logic. You can
also use logic as part of the methodology for solving actual real
world problems, but you cannot rely on that logic unless you take
steps to control the errors that can arise.
>Part of the difficulty here, I think, is an ambiguity in what we mean
>by "solving a problem". *Designing* something to, say, amplify a signal
>might be a problem for an engineer, in which some reasoning is
>involved, even if it is only to make trial and error attempts, and for
>which representations are likely to be deployed.
>The amplifier itself, whether analog or digital, does not have to solve
>a problem in this sense, it just has to amplify and so embody or
>instantiate a solution to the different problem of "how to amplify".
The amplifier itself is part of the implementation detail of the
solution. The important analysis of the problem had to occur at the
design phase. Whether digital or analog methods are used in the
implementation is a choice that the designer can make. But a failure
to analyze the problem properly will likely result in a failed
'solution'.
>But then it is not itself using continuous mathematics as opposed to
>discrete "logic", it is rather designed to exploit such properties as
>continuous mathematics is appropriate for. But it is *we* who use
>continuous mathematics in analyzing and explaining it.
During our learning, our neurons are using continuous methodologies
to analyze the world and construct neural structures that allow us to
deal with that world.
>>I take it that the brain is using continuous methods to solve
>>continuous problems, but is using a digital implementation of those
>>methods of solution.
>I wonder why you think the second?
I would have thought that the evidence spoke for itself. Neurons are
rather imprecise devices. When you build an analog system out of
imprecise devices, you add to the degree of imprecision at every
step. You simply could not get the degree of control that we (or
other mammals) have over our behavior with purely analog methods.
But if we use imprecise analog devices for digitizing, then we can
build up as much precision as is needed by doing something equivalent
to adding more digits of precision. The type of problem the brain
has to solve, and the biochemical hardware available, demands a
digital implementation. Beyond that, the evidence from neurobiology
is one of neurons making discrete decisions.
Anders N Weinstein
In article <5j5m2n$h...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5j3v2b$3...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>>As long as we were using it as a calculator, it was a calculator.
>
>>Perhaps if we began to use it as a calculator it would *become* a
>>calculator by virtue of being caught up in our practices. But as I
>>have been trying to argue against Pindor, there is still the
>>anthropologist's question "what is it" said of an *artifact*, which
>>is not just a matter of its being usefully interpreted as one by us.
>
>In my view, you are relying on sloppy thinking.
Care to explain why? Don't archeologists ask such questions about
the artifacts they dig up? Are they asking only: what can this be
used (by us now) as? Not at all. In that sense they take themselves
to be inquiring after a fact about the find. Do you really deny
this?
>>>I see nothing wrong with such an interpretation. We may later
>>>discover an even better interpretation. I think the standard for
>>>interpretation is one of usefulness, rather than right or wrong.
>
>>That is a standard for one kind of interpretation, but it is the wrong
>>one if you are trying to speak about an artifact.
>
>Then what is your definition of "artifact"? Is a bird nest an
>artifact? Is a beaver's dam an artifact? Perhaps we are just wrong
>about what is a bird's nest, and we should try to join the society of
>birds, to see how such an artifact is used in the customs of that
>society.
In one way that is certainly exactly what we do. Of course we do not have to
*join* a society we just have to find out about it. And in the case of
animal artifacts, we certainly do have to observe the normal pattern
of, say, bird life in in its normal habitat. Unless we do that, we don't
no what such a thing *is*.
So as far as the metaphysics goes, those things are very similar to the
artifacts we cultural creatures produce, in that they are identified as
being what they are by virtue of playing a functional role in the
animal's characteristic manner of living in its environment. Not any
pile of twigs is a *nest*; on Mars there might be creatures who built
exactly the same sorts of piles of twigs, but nevertheless those are
not nests, they are, say, traps for some of their prey that get easily
snared in them.
So to find out if something really is a nest, one has to look to the
context in which it was produced, not the intrinsic physical structure
of the object.
So in that way they are just like our artifacts -- a naturalist coming
across a pile of twigs might ask "what is it" in a similar sense in
which an archaeologist asks it of a find -- and expect an objective
answer, not just a matter of what he, the naturalist, finds it useful
to interpret or use the thing for *now*.
I would say there is an important difference in species within this
genus, though, insofar as our artifacts are reproduced via cultural
transmission and mediated by linguistic representation and intention.
There is an interesting passage on the differences between human and
animal production in Marx's 1844 manuscript on "Alienated Labor". He
suggests there that that human beings by nature have the capacity for
"free conscious activity" insofar as they have the possibility of
building a thing "in the mind" (i.e. by representing it to themselves)
before building it, while he suggests no spider ever does this, no
matter how subtle the design of its webs.
>A Martian, examining humans, might think that our computers are not
>artifacts, but are just as much a part of nature as we think bird's
>nests to be.
Well first, they are *both* artifacts and part of nature, these are not
exactly exclusive categories. It is certainly the most natural natural
thing in the world for human beings to produce artifacts, it is a
feature of our natural history. Second, I do think they are analogous
to bird's nests in the above respect.
But thirdly, if the Martian failed to recognize the computers as
artifacts then the Martian would just be blind and wrong about what
they are.
>The word "logical" is used either to mean something like reasoning by
>form or syntax, or alternatively to mean "sensible." I have been
>using it in the first sense. Now I'll grant that CD systems use
>logic chips, so they are using logic at the implementation level.
>But the analysis I was talking about is that which went into
>designing the system. That was largely an analysis based on
>continuous methods, including Fourier analysis.
Right. But my point was that the mathematics, including the theory
of Fourier analysis, can of course be formalized in a first-order language,
mathematical proofs of the requisite theorems can be constructed
in this medium.
So if you are theorizing *about* the system, don't you have to
use 'logic' in everything you do?
>>>Symbols are discrete entities. Symbol processing is digital, in any
>>>important sense of the term.
>
>>If I write discrete symbols on some discrete cards and juggle them, am
>>I doing something 'digital'?
>
>Are you juggling the symbols? Or are you juggling the cards, and the
>symbols are merely going along for a ride?
The latter, I expect, in that the symbolic function is irrelevant.
But the point was to question exactly what you were claiming.
And what about the "slab" game -- the players are not really doing
"digital" symbol transformations, even though the symbols are
discrete. Yet within the context of that activity, W suggests, the
discrete symbols have referential semantics, they stand for objects in
the world -- "Slab" denotes the slabs, "Pillar" denotes the pillars,
and so on.
>>It might be better to say that representation and reasoning are not the
>>best approaches to practical problems, e.g. walking to the next room,
>>or catching a fly ball.
>
>I would not suggest that for one moment. When I walk to the next
>room, my neurons are working with representations. I expect that a
Sure, but you are not. You are not "solving a problem" at all. If we
had to solve the problem of walking by reasoning it through, most
people would be unable to do it (perhaps *you* might).
>baseball player is reasoning when deciding which way to run in order
>to catch that fly ball. Perhaps the player does not attach a string
There normally is no such reasoning, although of course there can be.
I recall being advised that in order to sharpen my instincts when
playing the outfield, I ought to try forcing myself to move in one
direction or other immediately as soon as the ball leaves the bat. The
idea was that with practice, my reflexes would be developed so that I
would recognize whether to charge in or run back appropriately,
*without* any conscious thought or judgment about the matter.
That is a case in which there is no mental process of reasoning at all,
then, rather it is simply the exercise of a cultivated disposition.
Of course there is a story to be told at the level of neural circuitry
that explains how this disposition is set up. But it is not a story of
any *reasoning* processes I conduct. The whole thing is a little
uncanny from the point of view of the subject -- he is told, in
effect, *don't* think about where to run, just move, and you will
probably wind up better adapted to the task (by virtue of processes
happening in your neurons). Almost like "use the force".
>>E.g. you spoke about the axioms of Euclidean geometry as a set of
>>rules for the interpretation of measurement results. That speaks about
>>the activities of human beings in a world with other people, with
>>measuring devices, and with certain public symbols functioning as
>>rules in the context of this activity. The practices can be taught,
>>and new learners can be trained and tested in the operation of this
>>symbolic technique.
>
>>In that sense, the symbols that serve as an expression of a rule here
>>would be cultural or social objects. They have their being or function
>>by virtue of their role in a social medium.
>
>More mushy thinking. The symbols and measuring rules have their
>being or function because they have been shown to have a high degree
>of reliability (or repeatability of measurement), and have been found
>useful. That would be true, regardless of whether there was a
>society. Doubtless we have a richer set of measuring abilities
On a side note, is it your view that all such procedures are maintained
in existence because they have been critically tested and proven
themselves to be empirically useful? What about the tests for
witchood?
>available because they can be transmitted through the culture. But
>what makes them importance is not the cultural role, but their
>reliability and effectiveness.
I don't think I need to disagree with this. The original point was:
what is it for a symbol to function as a *rule*? This includes rules
for measurement that are to be judged by their effectiveness. The idea
was that this property -- functioning as a rule -- only occurs
in a social practical context.
>>This dichotomy seems sloppy to me -- it is not as though you could do
>>geometry or analysis without using logic.
>
>You miss the point. One can use logic in pure mathematics or in
>theoretical physics, without any difficulty. This is because pure
>mathematics and theoretical physics deals with an idealized
>theoretical world, rather than the real world. The idealization was
>done so as to make it safe for logic. But when we deal with ordinary
>real world problems, the use of logic must be greatly constrained.
I thought you said the idealizations are applied by being taken to
function as rules governing measurements, e.g. the axioms of
geometry. That does not seem to be constraining the use of logic,
does it? Don't you have to allow yourself the use of logic to
apply the rules?
>>But then it is not itself using continuous mathematics as opposed to
>>discrete "logic", it is rather designed to exploit such properties as
>>continuous mathematics is appropriate for. But it is *we* who use
>>continuous mathematics in analyzing and explaining it.
>
>During our learning, our neurons are using continuous methodologies
>to analyze the world and construct neural structures that allow us to
>deal with that world.
Possibly. But we can't do anything with the 'analyses' made by our
neurons -- for they simply don't communicate these results to us.
So if we want to understand the world explicitly, we have to make
analyses of our own. And at that level culturally transmitted
techniques of operating with symbols plays a crucial role.
Jim Balter
Neil Rickert wrote:
>
> In <5j3v2b$3...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
> >It is true, I think, that we don't walk confidently without thinking
> >because we have ourselves inductively confirmed that the ground
> >is generally solid. We generally just walk and are in a world in
> >which that habit works well enough.
>
> More nonsense arising from discrete thinking. We walk confidently
> because we have learned effective ways of judging how solid the
> ground is. We do not walk as confidently on soft mushy ground as we
> do on solid ground. When there is ice on the ground, we take a great
> deal more care about how we walk.
I wouldn't be surprised to learn that Weinstein is a klutz,
"generally just walk"ing and counting on the world being such that
"that habit works well enough".
OTOH, I wouldn't be entirely surprised to learn that Weinstein
is a troll who is very good at imagining what someone *might*
say who was so locked into a linguistic ideology that he would
make even the most absurd counter-evidentiary claims under the banner
of "It is true, I think".
--
<J Q B>
Jim Balter
Neil Rickert wrote:
> >So if you are theorizing *about* the system, don't you have to
> >use 'logic' in everything you do?
>
> You can call everything 'logic' if you wish. One might say that the
> Church-Turing thesis is the thesis that everything can be given a
> description as logic. But that is using the term too broadly for it
> to be useful.
It might be useful to consider what one logician has to say about it.
The following is by Raymond Smullyan, from his book _This Book Needs NO
Title_, Simon & Schuster, ISBN 0-671-62831-3. If you like this example
of his writing, you should go out and buy his books. (There, that
brings this under fair use, so the copyright police won't get me).
I think the really funny thing about this story is that most people will
find in it support for their own position, whatever it may be.
Is Peekaboo a Machine?
This story--though completely untrue--will, I hope, convince my
readers of the value of mathematical analysis.
Peekaboo is one of my dogs. Is she a mere machine? If she is, it is
extremely difficult to find out how she works! I have often wondered,
"Just how does this remarkable machine work?" Inorganic machines like
automobiles or, for that matter, even the most advanced electronic
equipment are often quite complicated, but to discover their /modus
operandi/ is child's play compared to Peekaboo!
How /does/ Peekaboo work, anyway? This problem vexed me for a long
time, until I finally found the answer! And here is where Science and
Mathematics come in. Without these two disciplines, I would have
remained in the dark forever. But now I know!
It happened this way. I realized that no mere armchair philosophical
theorizing would solve the problem; what was needed were /experiments/.
And so I hired one hundred of the world's leading experimental
psychologists to keep Peekaboo under observation for several months, and
to most carefully record all relevant data. Finally I had sufficient
data to solve the problem completely, if I could interpret them
correctly! I finally boiled down the correct interpretation to the
matter of simultaneously solving a system of 105 partial differential
equations. Many of these equations are extremely long--indeed, one of
them fills several volumes. The task of solving them is
prodigious--even the most facile mathematician would require several
lifetimes--assuming the equations could be solved at all! If I had
lived twenty years ago, the solution would be impossible to obtain. But
now we have high-speed computers! So I rented the fastest computer in
the world, programmed in all the information, and waited. It took
several months, but at last the day arrived when the equations were
solved! Now I have the whole key, and Peekaboo's behavior is no longer
a mystery to me. I know now /exactly/ how she works. I can predict her
every action to a T. Given /any/ stimulus whatsoever, I know exactly
her response.
Take, for example, the question of obedience. In the past, when I
gave here a command, I had absolutely no way of knowing how she would
respond. Her responses seemed to me so varied that I could find no
general law which would govern them all. Now I have such a law. But
remember, I could never have found out this law without mathematical
analysis.
What is this law? I will tell you. Every time I have ever given her
a command, she has /always/ responded in exactly the same way, only I
was not bright enough to recognize what the way is. Unaided by
mathematics, I kept looking at the /differences/ of the responses and
was totally blind to the similarities. But now I know! Whenever I give
her a command, there is only one thing she ever does, and every time it
is the same thing! Either she obeys it or she doesn't.
--
<J Q B>
Neil Rickert
In <5j60u4$b...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5j5m2n$h...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>Care to explain why? Don't archeologists ask such questions about
>the artifacts they dig up? Are they asking only: what can this be
>used (by us now) as? Not at all. In that sense they take themselves
>to be inquiring after a fact about the find. Do you really deny
>this?
Presumably they are asking how it was used. There is nothing wrong
with that. They cannot really be asking what it is. At best there
is only the question of what they choose to say it is. There could
be no answer (except, perhaps, a theological one), to the question of
what it really is.
>>Then what is your definition of "artifact"? Is a bird nest an
>>artifact? Is a beaver's dam an artifact? Perhaps we are just wrong
>>about what is a bird's nest, and we should try to join the society of
>>birds, to see how such an artifact is used in the customs of that
>>society.
>In one way that is certainly exactly what we do. Of course we do not have to
>*join* a society we just have to find out about it. And in the case of
>animal artifacts, we certainly do have to observe the normal pattern
>of, say, bird life in in its normal habitat. Unless we do that, we don't
>no what such a thing *is*.
But that is only enough to determine what we say it is. The birds
may have different ideas about that, which are inaccessible to us.
>I would say there is an important difference in species within this
>genus, though, insofar as our artifacts are reproduced via cultural
>transmission and mediated by linguistic representation and intention.
You don't know to what extent there is cultural transmission between
birds, nor whether they manage to communicate things that we cannot
be aware of.
>>The word "logical" is used either to mean something like reasoning by
>>form or syntax, or alternatively to mean "sensible." I have been
>>using it in the first sense. Now I'll grant that CD systems use
>>logic chips, so they are using logic at the implementation level.
>>But the analysis I was talking about is that which went into
>>designing the system. That was largely an analysis based on
>>continuous methods, including Fourier analysis.
>Right. But my point was that the mathematics, including the theory
>of Fourier analysis, can of course be formalized in a first-order language,
>mathematical proofs of the requisite theorems can be constructed
>in this medium.
You can use logic in pure mathematics in ways that cannot be properly
used in applications.
But theories such as Fourier analysis illustrate the more important
point. Although you can give a formal construction of mathematics,
through Fourier analysis, starting with the axioms of set theory,
that is not how it happened. Nobody, starting with the axioms of set
theory, and limiting themselves to the methods available from logic,
would have thought about Fourier analysis. That theory arose instead
from the methods of continuous mathematics applied to solving real
world problems (actually heat flow problems). The formal
construction from set theory came much later.
>So if you are theorizing *about* the system, don't you have to
>use 'logic' in everything you do?
You can call everything 'logic' if you wish. One might say that the
Church-Turing thesis is the thesis that everything can be given a
description as logic. But that is using the term too broadly for it
to be useful.
>> I expect that a
>>baseball player is reasoning when deciding which way to run in order
>>to catch that fly ball.
>There normally is no such reasoning, although of course there can be.
Of course there normally is such reasoning. But I expected you to
deny it. You are, after all, committed to the intellectualist
legend, so you must deny that reasoning exist when you cannot fit it
into your intellectualist ideas.
>On a side note, is it your view that all such procedures are maintained
>in existence because they have been critically tested and proven
>themselves to be empirically useful? What about the tests for
>witchood?
I don't have any doubt that witchood was tested for reliability and
usefulness. Perhaps it was useful mainly to those who used it to
maintain political power, or to those who aspired to such power.
>>You miss the point. One can use logic in pure mathematics or in
>>theoretical physics, without any difficulty. This is because pure
>>mathematics and theoretical physics deals with an idealized
>>theoretical world, rather than the real world. The idealization was
>>done so as to make it safe for logic. But when we deal with ordinary
>>real world problems, the use of logic must be greatly constrained.
>I thought you said the idealizations are applied by being taken to
>function as rules governing measurements, e.g. the axioms of
>geometry. That does not seem to be constraining the use of logic,
>does it? Don't you have to allow yourself the use of logic to
>apply the rules?
But the axioms of geometry describe an idealization of the process of
taking measurements. Thus you have infinitely thin lines,
infinitesimal points, exact length matches. Actual measurement is
messy, and does not behave as well. As I had said, the idealizations
served to make it safe to use logic with Euclidean geometry. But in
real world problem solving you must take care of the messiness
involved with actual measurements, and the likelihood of
contradictions if you put too much reliance on logic.
>>During our learning, our neurons are using continuous methodologies
>>to analyze the world and construct neural structures that allow us to
>>deal with that world.
>Possibly. But we can't do anything with the 'analyses' made by our
>neurons -- for they simply don't communicate these results to us.
Oh, nonsense. That is where most of our knowledge is developed.
That is where our conceptualizations, and most of our scientific
discoveries arise.
>So if we want to understand the world explicitly, we have to make
>analyses of our own.
So you can pretend. But you are only capable of making these
analyses of your own because your neurons have done their homework.
Jim Balter
Anders N Weinstein wrote:
>
> In article <5j5m2n$h...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
> >In <5j3v2b$3...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
> >>>As long as we were using it as a calculator, it was a calculator.
> >
> >>Perhaps if we began to use it as a calculator it would *become* a
> >>calculator by virtue of being caught up in our practices. But as I
> >>have been trying to argue against Pindor, there is still the
> >>anthropologist's question "what is it" said of an *artifact*, which
> >>is not just a matter of its being usefully interpreted as one by us.
> >
> >In my view, you are relying on sloppy thinking.
>
> Care to explain why? Don't archeologists ask such questions about
> the artifacts they dig up? Are they asking only: what can this be
> used (by us now) as? Not at all. In that sense they take themselves
> to be inquiring after a fact about the find. Do you really deny
> this?
The sloppy thinking is seen here again, when you restrict the options to
"what is it" "and what can we use it for". The question the
archaeologist asks is rather "what role could this have played". They
don't really care "what it is", although I suppose there is a naive
conception of archaeology as just digging things, up, figuring out what
they are, and then putting them on the right shelf in the museum. But,
like any scientist, they want to build a model, in this case a model of
a civilization. An archaeologist who finds a rhinoceros horn embedded
in a skeleton doesn't just say "I've got it: it's a rhinoceros horn",
and hands it to the zoologist. Nor does he say "I've got it: it's not a
rhinoceros horn, it's a weapon". That's where "what is it" sloppiness
can lead you. The archaeologist is interested in the functional role of
the horn in this culture, and in the causal relationships that could
have led to having rhinoceri around or how having rhinoceri around might
have affected this culture. Perhaps the archaeologist has reason to
suspect that rhinoceros horns were sacred in this culture and discovers
that the rhinoceros horn had been inserted into the skeleton after the
person had died. It would be dumb to say "I've got it: it's not a
rhinoceros horn, it's not a weapon, it's a religious item!" Such
sloppiness comes of a "what is it" mentality.
One way to be a sloppy thinker is to only look at examples that provide
support for one's preconceptions, and only look at them as far as they
do. For instance, a sloppy thinker might only consider examples such as
someone finding some object and saying "it's a basket", only to discover
later that it is *really* a hat, and then go into a litany about "by
virtue of being caught up in our practices" and so on. But this is more
sloppiness. In this case "it's a basket" carries with it connotations
of appearance and possible use; the question of whether it "really" is a
basket is nonsensical, because the word "basket" *always* carries these
connotations with it and cannot be separated from them. Failing to note
this leads to false dichotomies, to claims that it *really* is a hat as
a consequence of its artifactual history, it isn't *really* a basket at
all, its just a mistake to call it one. But if "it's a basket" often
means that it looks like a basket and could be used as one, and only
sometimes also means that it actually has been used as one, then there
is no categorical answer to the question "is it *really* a basket". And
perhaps more evidence would reveal that, although it was usually worn as
a hat, it was also sometimes used as a basket as well. Does this then
mean that it really *is* a basket after all? Only if you indulge in
sloppy "what is it" thinking; otherwise it simply means that it was used
as a basket; and as a hat; and it looks like a basket; and it could be
used as one.
So there is a difference between baskets and rhinoceros horns, because
they carry different sorts of connotations. The connotation of "basket"
is much more artifactual than "rhinoceros horn"; "basket" is primarily a
thing constructed of certain materials with a certain intent, and only
secondarily a matter of physical or functional capacity, whereas
"rhinoceros horn" is primarily a matter of its structure and causal
origin, and only incidentally artifactual. But suppose it turns out
that the rhinoceros horn was constructed by an alien race that is
unfamiliar with rhinoceri, for some purpose we can't even fathom. Does
this mean we were *wrong* to claim that it was a rhinoceros horn? Well,
to a sloppy "what is it" thinker we would be, but it depends upon
whether one takes "it's a rhinoceros horn" to mean that it definitely
was once attached to a rhinoceros, or if it is taken to mean that it has
the appearance and structure of the other objects that we have seen
attached to rhinoceri. Sloppy "what is it" thinking creates sloppy
inappropriate right/wrong dichotomies.
But it would be sloppy to stop there, considering only examples where we
might be right or wrong about something because of a lack of historical
knowledge. Of course, a sloppy thinker doesn't want to push on to
examples that might even further challenge his sloppy thinking. But
let's not be sloppy; let's consider whether there are other sorts of
connotations even further removed from historical consideration.
Suppose the archaeologist is wandering through a dig with a compass, and
notices it suddenly going wild. He triangulates a bit, finds the
location of the source, and digs up a rock that causes the compass
needle to always point at it. Now, he can speculate on how this rock
was used, and consider how we might use such rocks, how they might get
"caught up in our practices"; he might speculate on whether the rock was
made by aliens for some obscure purpose. But what is beyond speculation
is whether the rock is a magnet. Even if the rock was constructed by
some alien race for some unfathomable purpose out of some material that
we have never before encountered. Even if it slipped into our universe
from some other where there is no magnetism. If the thing attracts that
needle, its a magnet.
What is different about magnets from baskets and rhinoceros horns?
Well, "basket" can have an artifactual connotation that involves
history; something is a basket by virtue of being used as one or
designed as one. And "rhinoceros horn" can have a causal connotation
that involves history; something is a rhinoceros horn by virtue of
having developed as part of a rhinoceros. But "magnet" has a
*functional* connotation; something is a magnet by virtue of what it
*can do*, regardless of its history, and so we cannot be wrong about
whether something is a magnet by virtue of being misinformed about
history (unless our belief that it acts like a magnet is itself based on
history, rather than more direct observation).
It's just sloppy to think that all words have the same sorts of
connotations and thus can have the same mantra repeated about them, or
that things have only one or even some definite number of "what is it".
The whole notion of "what is it" is conceptually sloppy, as though "what
is it" were not itself a linguistic construct, but rather some
metaphysical law that things have true Platonic natures, despite the
fact that "true" and "Platonic" and "nature" are themselves linguistic
constructs that cannot be the pathways to "what is it" without being
circular and thus undemonstrable, a mere matter of dogma (which
complaint is not a case of verificationism, a term abused by people who
want to be able to make unsubstantiated claims and not be held
accountable for them).
So, for each word, we need to *carefully* consider how that word is
used, what sorts of connotations we attach to it. Like the word
"calculator". When we call something a calculator, are we primarily
referring to its role as an artifact in some culture, or its history, or
the materials from which it is built, or its location, or perhaps to
something else? Perhaps to what it can do.
Now of course one can sloppily point out that it can do more than one
thing; for instance, it can stop doors. But we weren't asking whether
it is a door stop, we were asking whether it is a calculator. Only
through sloppy "what it is" thinking does one interfere with the other.
We could point out that "door stop" usually carries with it a
connotation of something durable, something that wouldn't cease to
function in some other capacity through the use of it as a door stop,
and that that is why it seems odd to call a calculator a door stop, but
it is sloppy thinking that would lead to a debate as to whether it
"really is" a door stop as opposed to a calculator, even if it turns out
that the thing was made by aliens who find door stops with pushbuttons
aesthetically appealing and it was only by an amazing coincidence that
it can be used as a calculator; that it can do what calculators can do.
It is (poorly) functionally a door stop, and perhaps it is artifactually
a door stop in some alien culture. And it is functionally a calculator,
independently of the fact that it is artifactually a calculator in our
culture. It is only the most sloppy sort of "what is it" thinking by
which the functional nature of calculators as calculators gets negated
by the artifactual nature of calculators, in our culture or any other.
--
<J Q B>
Anders N Weinstein
In article <5j68a2$1...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5j60u4$b...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>In article <5j5m2n$h...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>
>>Care to explain why? Don't archeologists ask such questions about
>>the artifacts they dig up? Are they asking only: what can this be
>>used (by us now) as? Not at all. In that sense they take themselves
>>to be inquiring after a fact about the find. Do you really deny
>>this?
>
>Presumably they are asking how it was used. There is nothing wrong
>with that.
Yes, that is the point.
> They cannot really be asking what it is. At best there
>is only the question of what they choose to say it is. There could
>be no answer (except, perhaps, a theological one), to the question of
>what it really is.
All I need is that there can in most cases be a right answer to whether
it is an artifact and if so, what kind of artifact. Their "choosing" to
take it as, say, a table, cannot make such a claim correct.
It is not up to them what to say on such a matter, rather it is
objective, not a matter of subjective whim or caprice.
Your formulation misleadingly suggests they might say whatever they
want on the question and no one could prove them wrong -- which is
clearly an error.
>>I would say there is an important difference in species within this
>>genus, though, insofar as our artifacts are reproduced via cultural
>>transmission and mediated by linguistic representation and intention.
>
>You don't know to what extent there is cultural transmission between
>birds, nor whether they manage to communicate things that we cannot
>be aware of.
This is just idle and useless speculation until you provide
some evidence for the claims. Maybe the rocks and trees are communicating
too. Unless we can interpret them, we can do nothing with this
supposed hypothesis.
>>So if you are theorizing *about* the system, don't you have to
>>use 'logic' in everything you do?
>
>You can call everything 'logic' if you wish. One might say that the
I do not wish to call everything 'logic'. I wish to say that all
explicit sequential proof or reasoning requires employment of some
norms distinguishing between correct and invalid transitions, and that
is its 'logic'. It might in fact be quite informal in particular
contexts, as long as some distinction in practice is made between what
follows and what doesn't.
>>> I expect that a
>>>baseball player is reasoning when deciding which way to run in order
>>>to catch that fly ball.
>
>>There normally is no such reasoning, although of course there can be.
>
>Of course there normally is such reasoning. But I expected you to
>deny it. You are, after all, committed to the intellectualist
>legend, so you must deny that reasoning exist when you cannot fit it
>into your intellectualist ideas.
Hmm, I thought the "intellectualist legend" was to treat exercises
of skill as if they involved reasoning from theories and the like.
All I meant is that there is no conscious and explicit
reasoning done by the person, something I thought you agreed with --
the person does not think through how to move.
I don't really believe in the concept of unconscious *mental*
processes -- if they're not conscious, that means they're just
not *mental* processes, because the agent is not responsible for their
goodness or badness, and the mental ought to be defined by reference
to the agent's subjective point of view.
>>On a side note, is it your view that all such procedures are maintained
>>in existence because they have been critically tested and proven
>>themselves to be empirically useful? What about the tests for
>>witchood?
>
>I don't have any doubt that witchood was tested for reliability and
>usefulness. Perhaps it was useful mainly to those who used it to
>maintain political power, or to those who aspired to such power.
Right, but isn't that a very different sort of usefulness than it
actually purported to have, e.g. empirical efficacy? If it could be
shown that present-day science is only useful in the same way, that
would undermine it, not support it.
>>>During our learning, our neurons are using continuous methodologies
>>>to analyze the world and construct neural structures that allow us to
>>>deal with that world.
>
>>Possibly. But we can't do anything with the 'analyses' made by our
>>neurons -- for they simply don't communicate these results to us.
>
>Oh, nonsense. That is where most of our knowledge is developed.
>That is where our conceptualizations, and most of our scientific
>discoveries arise.
This claim does not explain how I can do something with an analysis
my neurons have made for me but not communicated to me. Do they
convey them to me in a dream, or what? Perhaps I get a hunch? But
then if I tell someone else about it, and they ask me why they
should believe it, what can I offer as justification? Ask my
neurons? At the epistemological level, it would just have arisen
as a hunch, and so would be without grounds.
>>So if we want to understand the world explicitly, we have to make
>>analyses of our own.
>
>So you can pretend. But you are only capable of making these
>analyses of your own because your neurons have done their homework.
I can agree with that.
Neil Rickert
In <5j8nd6$l...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5j68a2$1...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>>Care to explain why? Don't archeologists ask such questions about
>>>the artifacts they dig up? Are they asking only: what can this be
>>>used (by us now) as? Not at all. In that sense they take themselves
>>>to be inquiring after a fact about the find. Do you really deny
>>>this?
>>Presumably they are asking how it was used. There is nothing wrong
>>with that.
>Yes, that is the point.
>> They cannot really be asking what it is. At best there
>>is only the question of what they choose to say it is. There could
>>be no answer (except, perhaps, a theological one), to the question of
>>what it really is.
>All I need is that there can in most cases be a right answer to whether
>it is an artifact and if so, what kind of artifact. Their "choosing" to
>take it as, say, a table, cannot make such a claim correct.
But there usually is not a right answer, except in the imaginary
world of binary thinkers.
>Your formulation misleadingly suggests they might say whatever they
>want on the question and no one could prove them wrong -- which is
>clearly an error.
You can draw that silly conclusion if you wish. You only draw
attention to your own confusion.
>>>I would say there is an important difference in species within this
>>>genus, though, insofar as our artifacts are reproduced via cultural
>>>transmission and mediated by linguistic representation and intention.
>>You don't know to what extent there is cultural transmission between
>>birds, nor whether they manage to communicate things that we cannot
>>be aware of.
>This is just idle and useless speculation until you provide
>some evidence for the claims. Maybe the rocks and trees are communicating
>too.
Maybe they are. As I understand it, there is evidence of
communication between trees (by plant hormone emission).
>>>So if you are theorizing *about* the system, don't you have to
>>>use 'logic' in everything you do?
>>You can call everything 'logic' if you wish. One might say that the
>I do not wish to call everything 'logic'. I wish to say that all
>explicit sequential proof or reasoning requires employment of some
>norms distinguishing between correct and invalid transitions, and that
>is its 'logic'.
No doubt that is exactly what is required for the purely syntactic
world of binary thinkers such as yourself and Searle. In real life
it is not so. Only in worlds of discrete entities can you have
simple binary "correct or invalid" transitions. In the real world of
continuous phenomena, you have to make do with "near enough, given
the precision available." Now, as a binary thinker, you might want
to say that it is either true or false that the answer is "near
enough, given the precision available." But that is not so. You
might have to make do with "it is near enough to true." At best you
can have probability distributions giving some indication of the
degree of uncertainty.
>>>> I expect that a
>>>>baseball player is reasoning when deciding which way to run in order
>>>>to catch that fly ball.
>>>There normally is no such reasoning, although of course there can be.
>>Of course there normally is such reasoning. But I expected you to
>>deny it. You are, after all, committed to the intellectualist
>>legend, so you must deny that reasoning exist when you cannot fit it
>>into your intellectualist ideas.
>Hmm, I thought the "intellectualist legend" was to treat exercises
>of skill as if they involved reasoning from theories and the like.
I take the "intellectualist legend" to be the claim that the only
reasoning which counts is that which is accompanied by conscious
language use (whether internal or external).
>All I meant is that there is no conscious and explicit
>reasoning done by the person, something I thought you agreed with --
>the person does not think through how to move.
I was arguing that there is conscious reasoning. No doubt you would
not call it explicit if the reasoning does not have an adequate
linguistic accompaniment.
>I don't really believe in the concept of unconscious *mental*
>processes -- if they're not conscious, that means they're just
>not *mental* processes, because the agent is not responsible for their
>goodness or badness, and the mental ought to be defined by reference
>to the agent's subjective point of view.
I haven't suggested that the processes are not conscious. They at
least have a conscious component. Regardless. Let's even assume
that they are unconscious. The baseball player is still held
responsible for their goodness or badness -- did he catch that fly
ball, or did he fumble it?
>>>On a side note, is it your view that all such procedures are maintained
>>>in existence because they have been critically tested and proven
>>>themselves to be empirically useful? What about the tests for
>>>witchood?
>>I don't have any doubt that witchood was tested for reliability and
>>usefulness. Perhaps it was useful mainly to those who used it to
>>maintain political power, or to those who aspired to such power.
>Right, but isn't that a very different sort of usefulness than it
>actually purported to have, e.g. empirical efficacy?
I presume that the politicians involved would consider that they have
evidence of empirical efficacy.
> If it could be
>shown that present-day science is only useful in the same way, that
>would undermine it, not support it.
As far as I know, there was never much serious scientific theorizing
about witchhood.
>>>Possibly. But we can't do anything with the 'analyses' made by our
>>>neurons -- for they simply don't communicate these results to us.
>>Oh, nonsense. That is where most of our knowledge is developed.
>>That is where our conceptualizations, and most of our scientific
>>discoveries arise.
>This claim does not explain how I can do something with an analysis
>my neurons have made for me but not communicated to me.
You probably can't. After all, you live in a purely syntactic binary
world. Your neurons have probably long since given up trying. Every
time they have tried to communicate, they have found you otherwise
engaged in counting angels dancing on heads of pins.
As far as I am concerned, what my neurons have done and are doing is
never very far beneath the surface. There isn't any deeply complex
opaque algorithm being implemented by my neurons.
> Do they
>convey them to me in a dream, or what?
They are not making binary decisions. Therefore anything your
neurons have to say is incommensurable with your philosophical
predispositions.
> But
>then if I tell someone else about it, and they ask me why they
>should believe it, what can I offer as justification? Ask my
>neurons? At the epistemological level, it would just have arisen
>as a hunch, and so would be without grounds.
When Galileo was asked for justification, he didn't provide any.
Instead, he asked the inquirers to look through the telescope
themselves, so that their neurons could make the same discoveries for
them that his neurons had made for him.
H. M. Hubey
>> >In <5j3v2b$3...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>> >>Perhaps if we began to use it as a calculator it would *become* a
>> >>calculator by virtue of being caught up in our practices. But as I
>> >>have been trying to argue against Pindor, there is still the
>> >>anthropologist's question "what is it" said of an *artifact*, which
>> >>is not just a matter of its being usefully interpreted as one by us.
If we throw some grains on the ground and some birds start to
peck at them, are they asking themselves "what is it?".
If they are not doing that, then are they pecking at everything in
the world, or are they pecking at things that look like grains?
When a mosquito lands on your skin, it somehow "knows" that
there's blood under the surface. Does it try to do that when
it lands on trees or rocks? Is it saying to itself "what is it?"
Of course, I do not mean literally that either a bird or a
mosquito can understand human language. Do they have some means
of sensing their environment so that the equivalent question of
"what is it?" can be answered by them? at least to their own
satisfaction?
Is it possible that "meaning" and "interpretation" of things
are only meaningful to intelligent things including machines?
Why can we allow them for birds and insects but not machines?
How can we tell that a chicken mistook a pebble for rice
except by its observable behavior? If we attribute intention
to humans and animals based on its behavior, why can't we
do the same for machines?
--
Mark Hubey ---------------------------------------------------------
http://www.csam.montclair.edu/Faculty/Hubey.html
hu...@pegasus.montclair.edu hub...@alpha.montclair.edu
--------------------------------------------------------------------------
Philip Jackson
H. M. Hubey wrote:
>
> If we throw some grains on the ground and some birds start to
> peck at them, are they asking themselves "what is it?".
>
> If they are not doing that, then are they pecking at everything in
> the world, or are they pecking at things that look like grains?
>
> When a mosquito lands on your skin, it somehow "knows" that
> there's blood under the surface. Does it try to do that when
> it lands on trees or rocks? Is it saying to itself "what is it?"
>
> Of course, I do not mean literally that either a bird or a
> mosquito can understand human language. Do they have some means
> of sensing their environment so that the equivalent question of
> "what is it?" can be answered by them? at least to their own
> satisfaction?
>
> Is it possible that "meaning" and "interpretation" of things
> are only meaningful to intelligent things including machines?
> Why can we allow them for birds and insects but not machines?
> How can we tell that a chicken mistook a pebble for rice
> except by its observable behavior? If we attribute intention
> to humans and animals based on its behavior, why can't we
> do the same for machines?
>
Yes, and why did the chicken cross the road, by the way? Oh wait, that
was another thread, for the disputative...
Phil Jackson
--------------------------------------------------------------------
"...for the word is the sole sign and the only certain mark of the
presence of thought hidden and wrapt up in the body..." -- Descartes
--------------------------------------------------------------------
Standard Disclaimers. <pjac...@ic.net>
David Longley
In article <3359F8...@mail.ic.net>
pjac...@mail.ic.net "Philip Jackson" writes:
>
> Yes, and why did the chicken cross the road, by the way? Oh wait, that
> was another thread, for the disputative...
>
> Phil Jackson
There was something in "Philadelphia" about discrimination on the
basis of class memberhip being unconstutional. Did anyone inform
the insurance companies?
--
David Longley
pjac...@ic.net
In article <861538...@longley.demon.co.uk>,
Da...@longley.demon.co.uk wrote:
>
>
> In article <3359F8...@mail.ic.net>
> pjac...@mail.ic.net "Philip Jackson" writes:
> >
> > Yes, and why did the chicken cross the road, by the way? Oh wait, that
> > was another thread, for the disputative...
> >
> basis of class memberhip being unconstutional. Did anyone inform
> the insurance companies?
Yes. Every insurance company will now insure chickens who can fork over
enough cash (or chicken feed) to buy a policy.
Phil Jackson
--------------------------------------------------------------------
"...for the word is the sole sign and the only certain mark of the
presence of thought hidden and wrapt up in the body..." -- Descartes
--------------------------------------------------------------------
Standard Disclaimers. <pjac...@ic.net>
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
David Longley
In article <8615471...@dejanews.com> pjac...@ic.net writes:
>
> In article <861538...@longley.demon.co.uk>,
> Da...@longley.demon.co.uk wrote:
> >
> >
> > In article <3359F8...@mail.ic.net>
> > pjac...@mail.ic.net "Philip Jackson" writes:
> > >
> > > Yes, and why did the chicken cross the road, by the way? Oh wait, that
> > > was another thread, for the disputative...
> > >
> > There was something in "Philadelphia" about discrimination on the
> > basis of class memberhip being unconstutional. Did anyone inform
> > the insurance companies?
>
> Yes. Every insurance company will now insure chickens who can fork over
> enough cash (or chicken feed) to buy a policy.
>
Is that just for US born chickens? What about non grren card
carrying migrant chickens, can they get coverage too?
Does the constitution and its ammendments in general apply to
such chickens? I'm still intrigued by what Denzel Washington said
in "Philadelphia" that "discrimination" on the basis of group
memberhip is deemed unconstitutional...but maybe my memory is
failing me, maybe he didn't actually "say that"....
--
David Longley
Daryl McCullough
Jim Balter says...
>Perhaps you are under the false impression that the Church-Turing thesis
>claims that algorithms can be implemented on a Turing machine,
That's my impression, too. The way I remember it (it has certainly been
a while since I read anything about it, so maybe my memory is playing
tricks on me), Turing had his notion of computability (in terms
of Turing machines), and Church had his notion (in terms of the lambda
calculus) and Post (in terms of Post machines, whatever they are) and
Godel had his notion (in terms of recursive functions). Although they
were all different notions of computability, it turned out that they
were all mathematically equivalent---they all computed exactly the
same functions. The Church-Turing thesis was that *all* sensible notions
of computability would turn out to be equivalent to Turing machine
computability.
>rather than claiming that anything computable by a human is computable
>by a machine.
I think the original work on computability was really about *human*
computability (there *were* no machines in those days). So I would
characterize the thesis as "computable implies Turing-computable".
Maybe I have that wrong, but that's the way I remember it.
Here's the way it is put in _Elements of the Theory of Computation_,
by Lewis & Papadimitriou:
The principle that Turing machines are formal versions of
algorithms and that no computational procedure will be
considered an algorithm unless it can be presented as a
Turing machine is known as Church's Thesis or the Church-Turing
Thesis.
Strictly speaking, to have noncomputable human behavior would not
violate Church's Thesis, if that behavior could not be characterized
as "following an algorithm". An algorithm has to be guaranteed to
lead to the correct result, so a procedure that has a step like "Wait
for a brilliant insight" wouldn't be an algorithm.
The principle that "Anything a human can do (whether algorithmic or not)
can be done by a Turing machine" is not really Church's Thesis, I don't
think.
Daryl McCullough
CoGenTex, Inc.
Ithaca, NY
Oliver Sparrow
(H. M. Hubey) writes:
> If we throw some grains on the ground and some birds start to
> peck at them, are they asking themselves "what is it?".
>
> If they are not doing that, then are they pecking at everything in
> the world, or are they pecking at things that look like grains?
I am probably being literal-mined again, but we know the answer to
this one. Chicks peck randomly at anything in the right general size
range, but quickly learn (and do so more quickly when socialised) what
is food. The classes of things that are food depend on what they are
fed: grain is ignored by pellet fed birds, until a few brave souls
experiment. (Yes, that too seems to be built in, although randomly
distributed across the population and expressed most when risks are
perceived to be low.) Chickens share with humans the 'what' and
'where' division in post-visual cortical processing. That is, a
percept is streamed off to the limbic system for classification
and to near the motor cortex for physical mapping into the map of
the locale.
Why don't pocket calculator do the same. Because they are dead,
simply chunks of junk when compared to the CNS of a chicken, is why.
Philosowaffle tends to take a loose concept useful to complex
structures - meteorology, let us say - and then ask: why is this
ripple in my tea cup not describable as a 'storm'? Indeed, might
there not be tiny bacterial poets at this moment.....? Babble.
_________________________________________________
Oliver Sparrow
oh...@chatham.demon.co.uk
David Longley
In article <233877...@chatham.demon.co.uk>
oh...@chatham.demon.co.uk "Oliver Sparrow" writes:
Yep...the above is the thing I zoomed in on when doing my PhD in
the 80s, except I just worked on rats, not chickens or pigeons.
It's reassuring that bird brained animals manifest neophoba, as
it rules out the hippocampus as being necessary for "novelty
detection" and therefore "learning". I was looking at how things
"acquired meaning" (in my terms - become conditioned stimuli or
signals) - It made me become a S.O.M kind of guy (before SOM was
published of course) - that is, it made me seriously question the
localizationist ideas of learning, and believe in a levels of
learning/adaptatation approach where animals become more
integrated and adapt in their responses to signals of events as
they mature and experience the world and its contingent
relations.
Fragments of Behaviour: The Extensional Stance
http://www.uni-hamburg.de/~kriminol/TS/tskr.htm
The following was a 10 minute presentation with slides, covering
nearly 2 years work. Rules of the BPS dictate that authors must
be specified in alphabetical order, with the senior
author/presenter asterisked. The abstract is phrased as it is as
it was for a pharmacology audience. Reference to saline, is
isotonic saline, ie the vehicle which naloxone was dissolved in.
Further studies, delivered naloxone intra-cerebro-ventricularly
to confirm that it was a CNS phenomenon. As a behavioural
experiment, the results are very clear.
In the NIMR annual report, the work was summarised as indicating
that endogenous opioid peptides are involved in the processing of
novel stimuli. Rather than regard naloxone as 'enhancing'
neophobia it is better to take the complement of that notion and
say that it retards the decline of neophobia. That decline is, I
have argued, co-extensive with learning, or habit formation. The
pre-requisite to learning, or if you prefer, the constraint on
learning, is a variant of the UCR of neophobia. Opioids must be
initially OFF (theoretical elaboration on that point takes one
back to a neo-Hullian model of learning/habit formation).
I have not seen anything in the literature since this was first
reported, and by behavioural experiment standards, it is SEEMS a
dramatic effect. Has anyone the time, and resources to try to rep
licate the work? The findings stand for PVG and Sprague-Dawley
rats, male and female.
J.F.W. DEAKIN & D.C. LONGLEY*
(introduced by T.J. Crow)
National Institute for Medical Research,
Mill Hill, London, NW7 1AA
Several studies report that naloxone, an opiate receptor
antagonist, reduces deprivation induced eating and drinking.
However, in the present study, naloxone (5mg/kg,i.p.) did not
reduce food intake of rats maintained on a 22 h deprivation - 2 h
feeding schedule. In contrast, naloxone (5 mg/kg,i.p.)
progressively reduced water intake in deprived animals to 46% of
saline treated controls. No effects of naloxone (1, 5 mg/kg) on
established bar pressing for food or water were observed with
either continuous or fixed ratio schedules of reinforcement.
However, naloxone (5mg/kg) accelerated extinction of responding
when food and water were no longer available.
Animals treated with naloxone (5mg/kg) during training of the
bar-pressing ate only 26% of the pellets delivered whereas
controls ate all pellets delivered. Since the animals had not
previously experienced the pellets or the operant apparatus, the
possibilities arose that naloxone effects were due to enhanced
neophobic effects of the novel food pellets or novel apparatus
cues, or were due to conditioned taste aversion. Therefore, food
novelty, apparatus novelty and timing of injections were
independently varied in different groups of 8-10 rats treated
with saline or naloxone. Rats were maintained at 85% body weight
with 12g lab chow per day. On experimental days 46 small pellets
(Cambden instruments) were placed on a small petri dish in the
home cage of some groups or released from a pellet dispenser in
an operant box for other groups. The dependent variable was the
number of pellets eaten over 15 minutes.
Naloxone (1,5 mg/kg i.p.) injected 5 or 20 min before test almost
completely suppressed pellet eating if the animals had not been
previously exposed to the pellets (p<0.01 't' test vs saline
groups). This occurred independently of whether tests were
carried out in the home cage or novel operant box. Naloxone
induced suppression of pellet eating was almost completely
abolished in either environment if animals had been exposed to
the pellets for the five preceding days in the same or different
environment. Naloxone (5mg/kg, i.p.) administered immediately
after pellet eating tests failed to suppress subsequent pellet
eating.
Thus, naloxone suppressed pellet eating if the pellets were novel
and if naloxone was administered before eating tests. The results
suggest naloxone enhances neophobic effects of novel foods and
that suppression of novel pellet eating is not due to enhanced
effects of novelty of apparatus cues or to conditioned taste
aversion.
Reference
FRENK, H & ROGERS G.H. (1979) The suppressant effects of naloxone
on food and water intake in the rat.
Behav. Neural. Biol, 26, 23-40.
PROCEEDINGS OF THE BRITISH PHARMACOLGICAL SOCIETY (BPS)
1-3 April 1981
(Also British J Phramacology 1981)
---
David Longley
David Longley
As to "what it's for" ... I think that's quite clear. It's
impossible to pre-program all adaptive responses as other species
and the physical environment itself is always changing. Adapting
to signals which predict other events is an effective way of
dealing with that flux, and the efficiency of that plasticity is
something again which can be selected for. It is primarily via
observation categoricals (which is externalist) and not just
personal experience per se, that this process is optimised.
It's a mistake to get bogged down with the ephemeral notion of
"meaning".
--
David Longley
Anders N Weinstein
In article <335755...@netcom.com>, Jim Balter <j...@netcom.com> wrote:
>Anders N Weinstein wrote:
>>
>> In article <5j5m2n$h...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>> >In <5j3v2b$3...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>> >>>As long as we were using it as a calculator, it was a calculator.
>> >
>> >>Perhaps if we began to use it as a calculator it would *become* a
>> >>calculator by virtue of being caught up in our practices. But as I
>> >>have been trying to argue against Pindor, there is still the
>> >>anthropologist's question "what is it" said of an *artifact*, which
>> >>is not just a matter of its being usefully interpreted as one by us.
>> >
>> >In my view, you are relying on sloppy thinking.
>>
>> Care to explain why? Don't archeologists ask such questions about
>> the artifacts they dig up? Are they asking only: what can this be
>> used (by us now) as? Not at all. In that sense they take themselves
>> to be inquiring after a fact about the find. Do you really deny
>> this?
>
>The sloppy thinking is seen here again, when you restrict the options to
>"what is it" "and what can we use it for". The question the
>archaeologist asks is rather "what role could this have played". They
While they might at first ask "what role *could* this have played?" they
don't *just* ask that. Sure they will speculate at first:
it *could* have been some kind of weapon, it *could* have been used
in a religious ritual, it *could* have been merely decorative, it
*could* have been used for eating and so on.
But they are also interested in what institutionalized role such things
*actually* played. For which they can perfectly well get evidence by
looking to the wider context than the physical structure of the thing.
>don't really care "what it is", although I suppose there is a naive
>conception of archaeology as just digging things, up, figuring out what
>they are, and then putting them on the right shelf in the museum. But,
>like any scientist, they want to build a model, in this case a model of
>a civilization. An archaeologist who finds a rhinoceros horn embedded
Maybe you misunderstand me -- what I mean by finding out "what it is"
and what you mean by "building a model of a civilization" are basically
the same thing, as far as I can tell. E.g. to discover that it was an
ordinary drinking vessel commonly used by the upper classes, one would
have to know rather a lot about the institutionalized forms of life in
that civilization.
Compare: a foreigner sees Jones moving his arm in such and such a way,
so as to perform a certain gesture. "What's he doing"? the foreigner
asks -- a way of asking "what is it?". I might be able to give an
answer via a redescription, by furnishing a "thick description" -- he
is waving hello, he is raising the bid [at the auction], he is
exercising, and so on.
Now some of these might require further explanation of the relevant
context to understand -- e.g. explaining the full details of the practice of
auctions, and what they're for, and the like. If that is not in place,
I will also have to transmit that to the foreigner.
But if it is in place, then one kind of explanation is achieved
precisely via *redescription*, not explanatory model-building.
There is another sort of question we can ask -- "why is he doing that
[i.e. raising the bid]", seeking a reason. But we must be able to
redescribe it in thick terms first, to answer the "what is it?"
question, before we are in a position to ask "why is it happening?".
>can lead you. The archaeologist is interested in the functional role of
>the horn in this culture, and in the causal relationships that could
>have led to having rhinoceri around or how having rhinoceri around might
>have affected this culture. Perhaps the archaeologist has reason to
>suspect that rhinoceros horns were sacred in this culture and discovers
>that the rhinoceros horn had been inserted into the skeleton after the
>person had died. It would be dumb to say "I've got it: it's not a
>rhinoceros horn, it's not a weapon, it's a religious item!" Such
>sloppiness comes of a "what is it" mentality.
I would say the only sloppiness is in your uncharitable reading of
what I wrote. This is exactly the sort of thing I meant by
answering the "what is it" question.
Recall my original *point* -- that this is an objective question about the
socio-functional role of an artifact, not merely a matter of *our*
interpretation or what we find it useful to use, or what, objectively
speaking, it is well suited to be used for ("functionally a calculator").
>all, its just a mistake to call it one. But if "it's a basket" often
>means that it looks like a basket and could be used as one, and only
>sometimes also means that it actually has been used as one, then there
>is no categorical answer to the question "is it *really* a basket". And
I am skeptical about this as a claim about ordinary usage. But even if
I grant it, it would seem rather easy to explicitly disambiguate the two
senses of "it's a basket". All I need is that there's *one* sense in which
it is not merely a matter of present interpretation.
>origin, and only incidentally artifactual. But suppose it turns out
>that the rhinoceros horn was constructed by an alien race that is
>unfamiliar with rhinoceri, for some purpose we can't even fathom. Does
>this mean we were *wrong* to claim that it was a rhinoceros horn? Well,
Of course. Rhinoceros horns are basically on a par with artifacts insofar
as their individuation is "etiology laden". They differ in that the
etiological process is non-intentional and natural.
>to a sloppy "what is it" thinker we would be, but it depends upon
>whether one takes "it's a rhinoceros horn" to mean that it definitely
>was once attached to a rhinoceros, or if it is taken to mean that it has
>the appearance and structure of the other objects that we have seen
>attached to rhinoceri.
I find it very hard to imagine anyone really using the latter meaning.
But anyway, we have no problem disambiguating.
>Suppose the archaeologist is wandering through a dig with a compass, and
>notices it suddenly going wild. He triangulates a bit, finds the
>location of the source, and digs up a rock that causes the compass
>needle to always point at it. Now, he can speculate on how this rock
>was used, and consider how we might use such rocks, how they might get
>"caught up in our practices"; he might speculate on whether the rock was
>made by aliens for some obscure purpose. But what is beyond speculation
>is whether the rock is a magnet. Even if the rock was constructed by
>some alien race for some unfathomable purpose out of some material that
>we have never before encountered. Even if it slipped into our universe
>from some other where there is no magnetism. If the thing attracts that
>needle, its a magnet.
OK, then you are using "magnet" to denote an object considered as a
physical object, not as a socio-functional object, a player of an
institutionalized role within a community. That is perfectly fine.
Objects have both sorts of aspects, I think. Chess figures are also
shaped pieces of wood which have physical properties apart from their
role in chess games. And so on.
I don't know why you think I am not allowing for this. You seem to
be going to a lot of trouble to note that there are different senses
of the words we can use.
>So, for each word, we need to *carefully* consider how that word is
>used, what sorts of connotations we attach to it. Like the word
>"calculator". When we call something a calculator, are we primarily
>referring to its role as an artifact in some culture, or its history, or
>the materials from which it is built, or its location, or perhaps to
>something else? Perhaps to what it can do.
You could of course invent such a concept if there were some point to it.
I have no problem with that.
You could perhaps try to individuate brains in terms of "what they can
do" in this "narrow functional" sense. Dennett had a go at such an idea
in his "Beyond Belief". It is actually a little difficult, insofar as
most functional explanation requires some reference to the context of
operation, e.g. connections to a body of such and such a form and with
such and such biological needs in such and such an environment. So it
is not just a matter of what uninterpreted input-output pattern it can
produce, it also normally involves a teleological element, e.g. these
computations are for isolating information about likely *edges* in the
distal environment.
Again, if you want to invent these "purely functional" concepts of
"what a thing can do", go ahead. I would say they have nothing much to
do with psychology or biology or the mental or with philosophy of mind
or with cognitive science as actually practiced, but you are perfectly
free to employ them if you find some context in which they are
important.
Anders N Weinstein
In article <5j8vga$6...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5j8nd6$l...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>
>>> They cannot really be asking what it is. At best there
>>>is only the question of what they choose to say it is. There could
>>>be no answer (except, perhaps, a theological one), to the question of
>>>what it really is.
>
>>All I need is that there can in most cases be a right answer to whether
>>it is an artifact and if so, what kind of artifact. Their "choosing" to
>>take it as, say, a table, cannot make such a claim correct.
>
>But there usually is not a right answer, except in the imaginary
>world of binary thinkers.
>>Your formulation misleadingly suggests they might say whatever they
>>want on the question and no one could prove them wrong -- which is
>>clearly an error.
>
>You can draw that silly conclusion if you wish. You only draw
>attention to your own confusion.
And you only draw attention to your own non-responsiveness. Sometimes I
wonder why you bother with useless abuse, at these times you might
as well be Longley for all the point there is in trying to have
any discussion.
What are you disagreeing with me about here? Earlier you said there was
a perfectly good question about how an artifact was used in a culture.
These would be social facts or facts about socio-functional roles and I
say they are often objective matters of fact, not matters of
interpretation in the sense of "whatever *seems* right to the
archaeologist". That is, they support a distinction between what it is
right to say and what someone takes it to be right to say on the matter
-- my criterion of "objectivity".
So where do you disagree and why?
>>All I meant is that there is no conscious and explicit
>>reasoning done by the person, something I thought you agreed with --
>>the person does not think through how to move.
>
>I was arguing that there is conscious reasoning. No doubt you would
>not call it explicit if the reasoning does not have an adequate
>linguistic accompaniment.
Well can you give me some idea of what the contents of consciousness
are in the case of the "conscious reasoning" done in catching a fly
ball?
>> But
>>then if I tell someone else about it, and they ask me why they
>>should believe it, what can I offer as justification? Ask my
>>neurons? At the epistemological level, it would just have arisen
>>as a hunch, and so would be without grounds.
>
>When Galileo was asked for justification, he didn't provide any.
>Instead, he asked the inquirers to look through the telescope
>themselves, so that their neurons could make the same discoveries for
>them that his neurons had made for him.
That seems a comical way of characterizing it to me -- he wanted them
to look so they would, e.g. see the moons orbiting Jupiter *as* moons
orbiting Jupiter and so make the discovery themselves. But their
preconceptions prevented them from even bothering to try to make such a
discovery -- and these would have prevented them even if they had looked.
Most of the work needed would rather be in conveying to them the
conceptual understanding required to see the moons as moons, I think.
Without that, they might peer till they die without making any
advance.
Anders N Weinstein
In article <335BD9...@ace.acadiau.ca>,
Mark Young <MYo...@ace.acadiau.ca> wrote:
>Neurons do not communicate *to* their owners. Neurons are parts of
>their owner, and so anything going on in them is already there *at* the
>owner. Likewise I don't communicate *to* my fingers that they should
>move in certain ways over the keyboard; I simply move my fingers.
>Now it's possible to say that my brain communicates to my fingers that
>they should move in a certain way---my brain and my fingers are not in a
>part/whole relationship. Likewise one part of the computer may
>communicate to another part of the computer "telling" it what it should
>do.
Right, this is exactly my position. I was making fun of the idea.
> But the computer as a whole plays the same role to its parts as we
>do to ours. And in Strong AI it is the *computer* that is said to have
Only if, for the purposes of this sort of explanation, we *are* computers.
That is something I am strongly inclined to deny. I think we are better
thought of as human beings, a kind of organism with the natural capacity for
leading a special sort of life.
Of course computing is *one* of the many things we can do, e.g. when we
do sums or long division.
> And in Strong AI it is the *computer* that is said to have
>mental states in virtue of its parts acting in certain ways (under the
>control of software), just as we humans are said to have mental states
>in virtue of our parts acting in certain ways.
I am not sure Strong AI makes the same dualism of explanatory
properties discernable at different levels that I want to make. As
defined, Strong AI is the thesis that a system can think (have
intentional states) *in virtue of* instantiating the right program,
i.e. in virtue of being the right sort of *computer*. I do not think we
can be said to understand a language in virtue of instantiating a
program -- non-computational interactions with the environment are
necessary, as is much else, before the concept of understanding gets a
non-metaphorical grip. And then the inner program can pretty much
drop out of the account.
>IF (and I'm not sure it's even possible) "you" (your consciousness, I'm
>assuming) get a message "from" your neurons---as a hunch or dream or
>whatever---then they *have* communicated with "you". If they don't
>communicate them to "you", then you are simply unaware of them, just as
>you are unaware of countless other activities going on inside your body.
Right, but again, this shows they are not part of my mind as I demarcate
the mental -- they are not involved in giving my subjective perspective
on the world. Without some relation to a subjective point of view, I think
there is no way to delimit the mental even loosely -- events in
my adrenal glands have effects in my psychological state, but are not
themselves part of my mind. Similarly for events in my neurons, even if
at their own level they are processing information and "consulting"
representations I am wholly unaware of.
Mark Young
Anders N Weinstein wrote:
> I don't really believe in the concept of unconscious *mental*
> processes -- if they're not conscious, that means they're just
> not *mental* processes, because the agent is not responsible for their
> goodness or badness, and the mental ought to be defined by reference
> to the agent's subjective point of view.
Ah, but 'mental' *isn't* defined by reference to the agent's subjective
point of view. It's defined by reference to the subject's mind, which
includes the subconscious.
[...]
> This claim does not explain how I can do something with an analysis
> my neurons have made for me but not communicated to me.
Neurons do not communicate *to* their owners. Neurons are parts of
their owner, and so anything going on in them is already there *at* the
owner. Likewise I don't communicate *to* my fingers that they should
move in certain ways over the keyboard; I simply move my fingers.
Now it's possible to say that my brain communicates to my fingers that
they should move in a certain way---my brain and my fingers are not in a
part/whole relationship. Likewise one part of the computer may
communicate to another part of the computer "telling" it what it should
do. But the computer as a whole plays the same role to its parts as we
do to ours. And in Strong AI it is the *computer* that is said to have
mental states in virtue of its parts acting in certain ways (under the
control of software), just as we humans are said to have mental states
in virtue of our parts acting in certain ways.
> Do they convey them to me in a dream, or what? Perhaps I get a hunch?
IF (and I'm not sure it's even possible) "you" (your consciousness, I'm
assuming) get a message "from" your neurons---as a hunch or dream or
whatever---then they *have* communicated with "you". If they don't
communicate them to "you", then you are simply unaware of them, just as
you are unaware of countless other activities going on inside your body.
...mark young
Neil Rickert
In <5jg1d7$2...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>>All I need is that there can in most cases be a right answer to whether
>>>it is an artifact and if so, what kind of artifact. Their "choosing" to
>>>take it as, say, a table, cannot make such a claim correct.
>>But there usually is not a right answer, except in the imaginary
>>world of binary thinkers.
>>>Your formulation misleadingly suggests they might say whatever they
>>>want on the question and no one could prove them wrong -- which is
>>>clearly an error.
>>You can draw that silly conclusion if you wish. You only draw
>>attention to your own confusion.
>And you only draw attention to your own non-responsiveness.
What non-responsiveness? Just because you failed to quote the more
detailed parts of my response, you cannot use that failure to claim
that I was non-responsive.
I'll respond again, with different wording and more detail.
If you ask me to write down the decimal expansion of the square root
of 2, there is no answer I can write down that will be correct. But
it does not follow that I can say whatever I want to. Even though I
cannot right down an answer that is correct, I can see that 1.4 is a
better answer than 1., and 1.4142 is an even better answer. If you
ask me to write down Planck's constant, this is even more difficult
since there is no known mathematically perfect answer. Given any two
answers A and B, I can sometimes say that A is a better answer than
B. But only someone confused by binary thinking would claim that,
just because I cannot say whether an answer is right, therefore I can
say whatever I please.
>What are you disagreeing with me about here? Earlier you said there was
>a perfectly good question about how an artifact was used in a culture.
It does not follow that such a question has a definite answer.
>These would be social facts or facts about socio-functional roles and I
>say they are often objective matters of fact, not matters of
>interpretation in the sense of "whatever *seems* right to the
>archaeologist".
The society, within which they perhaps were social facts, no longer
exists. There is no certain standard as what was that social fact,
if indeed they were social facts. No matter what, it is a question
of interpetation. Getting from such uncertainty to a completely
subjective "whatever *seems* right to the archaeologist" strikes me
as nonsensical.
>>>All I meant is that there is no conscious and explicit
>>>reasoning done by the person, something I thought you agreed with --
>>>the person does not think through how to move.
>>I was arguing that there is conscious reasoning. No doubt you would
>>not call it explicit if the reasoning does not have an adequate
>>linguistic accompaniment.
>Well can you give me some idea of what the contents of consciousness
>are in the case of the "conscious reasoning" done in catching a fly
>ball?
You are consciously deciding in which direction to run, and in which
direction to stretch out your hands. Now perhaps that is only the
way I do it. Perhaps when you play baseball, you do so as a zombie
and your fielding is all done without any consciousness. But somehow
I doubt that you could play as such a zombie.
>>> But
>>>then if I tell someone else about it, and they ask me why they
>>>should believe it, what can I offer as justification? Ask my
>>>neurons? At the epistemological level, it would just have arisen
>>>as a hunch, and so would be without grounds.
>>When Galileo was asked for justification, he didn't provide any.
>>Instead, he asked the inquirers to look through the telescope
>>themselves, so that their neurons could make the same discoveries for
>>them that his neurons had made for him.
>That seems a comical way of characterizing it to me -- he wanted them
>to look so they would, e.g. see the moons orbiting Jupiter *as* moons
>orbiting Jupiter and so make the discovery themselves.
I am not making the comic characterization that you presume. If they
looked, they would not see the moons orbiting Jupiter. Have you read
"Do you see through a microscope" (Ian Hacking)? It is reproduced in
his book "Representing and Intervening." You cannot just look and
see. You have to learn how to see through a microscope. Similarly,
you have to learn how to see through a telescope. So Galileo could
not have provided any justification. At best, he could have provided
the opportunity for his critics to learn for themselves how to see
through a telescope. Once they had so learned, they would be able to
look toward Jupiter and see for themselves.
> But their
>preconceptions prevented them from even bothering to try to make such a
>discovery -- and these would have prevented them even if they had looked.
Perhaps in the same way that your preconceptions prevent you from
seeing what I am trying to describe to you. But Galileo did find
some people willing to take the time to learn to see through that
telescope.
>Most of the work needed would rather be in conveying to them the
>conceptual understanding required to see the moons as moons, I think.
No amount of talk or cultural transmission of concepts could have
conveyed that. They had to learn for themselves how to see through
the telescope. But you will not agree. You are so committed to your
own conceptual misunderstandings, that you have no ability to learn
anything that would contradict them.
>Without that, they might peer till they die without making any
>advance.
If they only looked at Jupiter, they might peer till death without
learning anything. They instead would have to use the telescope in
many ways so as to learn how to correlate the signals they receive
through the telescope with those obtainable in other ways. Only in
that way could they learn sufficiently that they might be able to
appreciate what they say when looking toward Jupiter.
Anders N Weinstein
In article <5jgnfm$7...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5jg1d7$2...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>In article <5j8vga$6...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>>You can draw that silly conclusion if you wish. You only draw
>>>attention to your own confusion.
>
>>And you only draw attention to your own non-responsiveness.
>
>What non-responsiveness? Just because you failed to quote the more
>detailed parts of my response, you cannot use that failure to claim
>that I was non-responsive.
I didn't and don't see how that detail -- which was further down in
the post -- provides "support" for this.
>If you ask me to write down the decimal expansion of the square root
>of 2, there is no answer I can write down that will be correct. But
>it does not follow that I can say whatever I want to. Even though I
>cannot right down an answer that is correct, I can see that 1.4 is a
>better answer than 1., and 1.4142 is an even better answer. If you
>ask me to write down Planck's constant, this is even more difficult
>since there is no known mathematically perfect answer. Given any two
>answers A and B, I can sometimes say that A is a better answer than
>B. But only someone confused by binary thinking would claim that,
>just because I cannot say whether an answer is right, therefore I can
>say whatever I please.
Fine, now how does this apply to the case at issue?
Note also that to a great many questions, in a great many contexts, there
may be exactly one right answer. Not a continuum of better or worse with
no best, but just one. Everyday yes-no questions are usually like this.
So if an archaeologist inquires whether this artifact was used as a weapon
they *may* get the answer "it depends" with some gloss ("it began as
a soldier's multi-purpose tool, usable as a weapon but also for
cutting food, and later evolved into a largely ornamental device, although
occasionally used in such and such a way..."). That is fine, it is
still to the question.
But I think you are only kicking up dust here if all you are saying is
that *some* questions are subject only to a better or worse with no best.
Sure. So what?
>>What are you disagreeing with me about here? Earlier you said there was
>>a perfectly good question about how an artifact was used in a culture.
>
>It does not follow that such a question has a definite answer.
No, it doesn't, but why should I think it does not in most cases have
a perfectly definite answer? Is this something archaeologists fail to
recognize when they make their inquiries? Very often it comes up with
a perfectly adequate right answer, and where not, the failure to have
a definite answer can be explained.
And anyway, what does this fact have to do with the objectivity of
the question as I defined it? That depends only on the intelligibility
of a distinction between what is the case and what is taken to be
the case.
>>These would be social facts or facts about socio-functional roles and I
>>say they are often objective matters of fact, not matters of
>>interpretation in the sense of "whatever *seems* right to the
>>archaeologist".
>
>The society, within which they perhaps were social facts, no longer
>exists. There is no certain standard as what was that social fact,
>if indeed they were social facts. No matter what, it is a question
>of interpetation. Getting from such uncertainty to a completely
>subjective "whatever *seems* right to the archaeologist" strikes me
>as nonsensical.
OK. What is the content for you of saying it is "a question of
interpretation"? When I said "it is not a matter of interpretation" I
meant that the inquiry is taken by all concerned to be objective, in
that there is a difference between what is right and what seems right
to the archaeologist.
If you go back to the original context, I in particular meant that the
question as to whether something that fell from the sky was an
artifactual computer in some society is not settled solely by its
usability or interpretability as a computer by us.
Now you want to kick up dust and say "in some cases, this question might not
have a definite answer". Sure. But in many cases it does.
>>Well can you give me some idea of what the contents of consciousness
>>are in the case of the "conscious reasoning" done in catching a fly
>>ball?
>
>You are consciously deciding in which direction to run, and in which
>direction to stretch out your hands. Now perhaps that is only the
>way I do it. Perhaps when you play baseball, you do so as a zombie
>and your fielding is all done without any consciousness. But somehow
>I doubt that you could play as such a zombie.
I might be conscious all the while. But still not be conscious of how I
am solving the problem -- my consciousness might be occupied entirely
with such thoughts as "I hope I catch it and don't make a fool of
myself" or something like that. More likely, it might be unpacked as "I
need to get *there* in time".
>>>When Galileo was asked for justification, he didn't provide any.
>>>Instead, he asked the inquirers to look through the telescope
>>>themselves, so that their neurons could make the same discoveries for
>>>them that his neurons had made for him.
>
>>That seems a comical way of characterizing it to me -- he wanted them
>>to look so they would, e.g. see the moons orbiting Jupiter *as* moons
>>orbiting Jupiter and so make the discovery themselves.
>
>I am not making the comic characterization that you presume. If they
>looked, they would not see the moons orbiting Jupiter. Have you read
>"Do you see through a microscope" (Ian Hacking)? It is reproduced in
>his book "Representing and Intervening." You cannot just look and
>see. You have to learn how to see through a microscope. Similarly,
>you have to learn how to see through a telescope.
Yes, I was thinking of just this point.
> So Galileo could
>not have provided any justification. At best, he could have provided
>the opportunity for his critics to learn for themselves how to see
>through a telescope. Once they had so learned, they would be able to
>look toward Jupiter and see for themselves.
This does not follow. He could explain, *show*, or train them in how to
see through the telescope (my "conveying the understanding needed").
He does not have to simply hand it to them and hope that they will hit
on the right way to take what they see all by themselves -- something
extremely unlikely given their mental set.
>> But their
>>preconceptions prevented them from even bothering to try to make such a
>>discovery -- and these would have prevented them even if they had looked.
>
>Perhaps in the same way that your preconceptions prevent you from
>seeing what I am trying to describe to you. But Galileo did find
>some people willing to take the time to learn to see through that
>telescope.
Of course, but this is not a matter of their neurons making some
discovery for them.
>>Most of the work needed would rather be in conveying to them the
>>conceptual understanding required to see the moons as moons, I think.
>
>No amount of talk or cultural transmission of concepts could have
>conveyed that. They had to learn for themselves how to see through
>the telescope.
But why is training someone in how to see through telescopes not a
species of cultural transmission of a cultivated capacity? In particular,
they could not see tiny moving dots as *moons* orbiting Jupiter without
a large theoretical superstructure, which is conveyed largeley through
language.
Jason Corley
Jim Balter (j...@netcom.com) wrote:
: Belief has nothing to do with it; the SR simply points out that
: Searle is arguing against a strawman. That is true whether or not
: Strong AI is valid, whether or not Searle's conclusions or premises
: are right, whether or not understanding is a matter of following
: rules. The SR makes no claim whatsoever about such things.
: All it says is "Whether the Chinese Room understands Chinese does not
: hinge upon whether the Searle homunculus understands Chinese."
: That's it; period.
I think the System reply to the Chinese room is particularly strong, but I
never thought of it in quite such a stripped-down way.
If this is really all the reply says, then it's quite possible that Searle
could quite legitimately say that this reply has gone too far in the
_opposite_ direction, to a too-_narrow_ conception of understanding.
That is, if the System reply only says that the understanding of the
Searle homunculus (which I would really like to see, by the way) is
irrelevant to the question of whether the CR understands Chinese, then the
natural question is: what _is_ relevant? There are only a few more pieces
of the CR that we can point to: the lists of characters, the rules for
manipulating them, the inputs provided to the CR system...what this
version of the System reply seems to say is not that the CR system
understands Chinese, but that the researcher that provided the CR system
with those lists of characters and the rules for manipulating them
understands Chinese, which nobody disputes, but really is only vacuously
true.
I like a more general version of the System reply, which is that clearly
we have to admit that the CR is certainly doing _something_ cognitive,
just as a loud American tourist armed with a Chinese-English dictionary
might muddle his way through social interactions in Beijing. We don't
want to say that the tourist "understands" Chinese in the same way as a
native Chinese speaker does, and certainly not in the same way as the
author of the dictionary does. This version of the Systems reply says
that Searle's argument definitely shows that our concept of understanding
is vague enough to cause us some problems when we go to use it in a
practical case, but it does not necessarily show us anything about
strictly _artificial_ intelligence...the force of the CR argument is about
intelligence and understanding _in general_. Proponents of Strong AI who
use the vague layman's definition of understanding are going to run into
the problems Searle states.
Just discovered this newsgroup, and it's really pushing my buttons. I
think I'll stick around.
--
"He means well for his country, is always an honest man, often a wise man,
but sometimes, and in some things, absolutely out of his senses."
-----Benjamin Franklin, 1783
Jason D. "cor...@tau.lpl.arizona.edu" Corley isn't John Adams.
H. M. Hubey
Oliver Sparrow <oh...@chatham.demon.co.uk> writes:
>Why don't pocket calculator do the same. Because they are dead,
>simply chunks of junk when compared to the CNS of a chicken, is why.
Sure, but there are unspoken assumptions, taken for granted
by techno people. One of them is that motion is an understood
problem. It's just that its space complexity is large. Since
we don't need machines that walk, and it's easier to make
machines that roll, fly or swim, nobody has attempted to
make such machines.
What they are concentrating is on the unexplored and
not-yet-understood parts. One of the things that's already
understood is that formalizable things, like math, as
difficult as it seems for the masses is easy for computers.
Ditto for expert systems. The bigger problem boils down to
finding efficient ways of imparting algorithmic knowledge
to machines without having to explicitly program them. IT works
for simple systems like WP, DTP, OS LAN etc but it clearly
doesn't scale well. You cannot make a machine with common
sense (even as much as a chicken's common sense) using
these methods. Connectionism is one attractive way, because
it's a step toward implicit learning, i.e. one in which the
knowledge/algorithm is not written by humans and coded into ROM
or RAM. That is why now the interest is in automatic data
analysis, data-mining etc. This is one step removed from programmming
the algorithm. We will find that there are more and more layers
of complexity involved in what "intelligence" and "knowledge"
means and what they mean for "understanding".
>Philosowaffle tends to take a loose concept useful to complex
>structures - meteorology, let us say - and then ask: why is this
>ripple in my tea cup not describable as a 'storm'? Indeed, might
>there not be tiny bacterial poets at this moment.....? Babble.
Thank you. Philosowaffle and psychobabble are what seem to
stand in the way because those who are in the position to
think about these think in different ways, in different levels
of detail and with different sets of mental tools. But as
the saying goes, babbling and waffling halt neither
science nor technology.
Neil Rickert
In <5jguoe$7...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5jgnfm$7...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>If you ask me to write down the decimal expansion of the square root
>>of 2, there is no answer I can write down that will be correct. But
>>it does not follow that I can say whatever I want to. Even though I
>>cannot right down an answer that is correct, I can see that 1.4 is a
>>better answer than 1., and 1.4142 is an even better answer. If you
>>ask me to write down Planck's constant, this is even more difficult
>>since there is no known mathematically perfect answer. Given any two
>>answers A and B, I can sometimes say that A is a better answer than
>>B. But only someone confused by binary thinking would claim that,
>>just because I cannot say whether an answer is right, therefore I can
>>say whatever I please.
>Fine, now how does this apply to the case at issue?
I don't know. It is merely a counter example which completely
refutes what you had said. I guess that means it does not apply.
And yet at other times you argue that there could not be such a thing
as meaning incommensurability.
>Note also that to a great many questions, in a great many contexts, there
>may be exactly one right answer. Not a continuum of better or worse with
>no best, but just one. Everyday yes-no questions are usually like this.
Many every day yes-no questions are not like that at all.
>But I think you are only kicking up dust here if all you are saying is
>that *some* questions are subject only to a better or worse with no best.
>Sure. So what?
I am rejecting your absurd insinuation that if there is not a
definite answer it must be purely arbitrary.
>>It does not follow that such a question has a definite answer.
>No, it doesn't, but why should I think it does not in most cases have
>a perfectly definite answer?
There is not reason at all for you to think that there might not be a
definite answer. You are, after all, committed to the folly of
binary thinking.
> Is this something archaeologists fail to
>recognize when they make their inquiries? Very often it comes up with
>a perfectly adequate right answer, and where not, the failure to have
>a definite answer can be explained.
They may come up with a perfectly adequate answer. They may, as a
matter of convention, decide to assert that this is the right
answer. But that is hardly the same thing as coming up with "a
perfectly adequate right answer."
>OK. What is the content for you of saying it is "a question of
>interpretation"? When I said "it is not a matter of interpretation" I
>meant that the inquiry is taken by all concerned to be objective, in
>that there is a difference between what is right and what seems right
>to the archaeologist.
The need to kill Bosnians was probably taken by all Serbs concerned
to be objective, in that there was a difference (in their view)
between what is right and what seemed right to an individual Serb.
Do we then conclude that it was objectively right? Perhaps you are
suggesting that might makes right?
>>You are consciously deciding in which direction to run, and in which
>>direction to stretch out your hands. Now perhaps that is only the
>>way I do it. Perhaps when you play baseball, you do so as a zombie
>>and your fielding is all done without any consciousness. But somehow
>>I doubt that you could play as such a zombie.
>I might be conscious all the while. But still not be conscious of how I
>am solving the problem
By which you mean that you cannot give an explicit verbal
description. But if the reasoning was non-verbal, we should not
expect that you could give a verbal description. I think it a
mistake to say that you are not conscious of how you are solving the
problem, merely because your solution is non-verbal.
> -- my consciousness might be occupied entirely
>with such thoughts as "I hope I catch it and don't make a fool of
>myself" or something like that. More likely, it might be unpacked as "I
>need to get *there* in time".
In that case you would probably miss the catch most of the time.
More likely you are concentrating on the ball and consciously
positioning your hands so as to meet the ball. That requires
decision making, so is reasoning even if not verbal.
>> So Galileo could
>>not have provided any justification. At best, he could have provided
>>the opportunity for his critics to learn for themselves how to see
>>through a telescope. Once they had so learned, they would be able to
>>look toward Jupiter and see for themselves.
>This does not follow. He could explain, *show*, or train them in how to
>see through the telescope (my "conveying the understanding needed").
Most of what you have to learn cannot be explained verbally.
>>Perhaps in the same way that your preconceptions prevent you from
>>seeing what I am trying to describe to you. But Galileo did find
>>some people willing to take the time to learn to see through that
>>telescope.
>Of course, but this is not a matter of their neurons making some
>discovery for them.
No, of course not. Obviously that would contradict your a priori
philosophical notions, so on the basis of dogma it must be wrong.
>>No amount of talk or cultural transmission of concepts could have
>>conveyed that. They had to learn for themselves how to see through
>>the telescope.
>But why is training someone in how to see through telescopes not a
>species of cultural transmission of a cultivated capacity?
If the telescope were left lying around, someone who found it might
learn how to see through a telescope with no cultural transmission.
Jim Balter
Anders N Weinstein wrote:
> >The sloppy thinking is seen here again, when you restrict the options to
> >"what is it" "and what can we use it for". The question the
> >archaeologist asks is rather "what role could this have played". They
>
> While they might at first ask "what role *could* this have played?" they
> don't *just* ask that. Sure they will speculate at first:
> it *could* have been some kind of weapon, it *could* have been used
> in a religious ritual, it *could* have been merely decorative, it
> *could* have been used for eating and so on.
>
> But they are also interested in what institutionalized role such things
> *actually* played. For which they can perfectly well get evidence by
> looking to the wider context than the physical structure of the thing.
I have no response to your sloppy metaphysical realism.
I'll leave my posting to others able to comprehend it.
I will simply point out that "could have" above means "could have
consistent with the evidence".
--
<J Q B>
Jim Balter
Jason Corley wrote:
>
> Jim Balter (j...@netcom.com) wrote:
>
> : Belief has nothing to do with it; the SR simply points out that
> : Searle is arguing against a strawman. That is true whether or not
> : Strong AI is valid, whether or not Searle's conclusions or premises
> : are right, whether or not understanding is a matter of following
> : rules. The SR makes no claim whatsoever about such things.
> : All it says is "Whether the Chinese Room understands Chinese does not
> : hinge upon whether the Searle homunculus understands Chinese."
> : That's it; period.
>
> I think the System reply to the Chinese room is particularly strong, but I
> never thought of it in quite such a stripped-down way.
>
> If this is really all the reply says, then it's quite possible that Searle
> could quite legitimately say that this reply has gone too far in the
> _opposite_ direction, to a too-_narrow_ conception of understanding.
>
> That is, if the System reply only says that the understanding of the
> Searle homunculus (which I would really like to see, by the way) is
> irrelevant to the question of whether the CR understands Chinese, then the
> natural question is: what _is_ relevant?
Well, it *is* irrelevant, isn't it? If Searle did happen to speak
Chinese, it would have no bearing on his carrying out the instructions.
(At least not according to Searle's scenario; Aaron Sloman points out
that humans may not be the right sorts of control systems; a Searle
homunculus that understood Chinese could interfere with the CR
understanding Chinese; for instance, a message might come in through
the slot in Chinese asking whether Searle's wife has a certain
hidden tattoo in a private place.)
> There are only a few more pieces
> of the CR that we can point to: the lists of characters, the rules for
> manipulating them, the inputs provided to the CR system...what this
> version of the System reply seems to say is not that the CR system
> understands Chinese, but that the researcher that provided the CR system
> with those lists of characters and the rules for manipulating them
> understands Chinese, which nobody disputes, but really is only vacuously
> true.
No, it says no such thing. It says that Searle's conclusion doesn't
follow because whether the CR understands Chinese doesn't hinge upon
whether the Searle homunculus understands Chinese. The Systems Reply
says to take the question "Does the CR understand Chinese?" seriously,
and stop answering irrelevant questions about "pieces" of the CR.
Searle has not addressed the question he posed.
> I like a more general version of the System reply, which is that clearly
> we have to admit that the CR is certainly doing _something_ cognitive,
> just as a loud American tourist armed with a Chinese-English dictionary
> might muddle his way through social interactions in Beijing.
But this isn't the SR, and Searle *does not* admit that the CR
is doing something cognitive; this the whole issue! You can't
rebut Searle simply by contradicting him.
> We don't
> want to say that the tourist "understands" Chinese in the same way as a
> native Chinese speaker does, and certainly not in the same way as the
> author of the dictionary does.
But the assertion that Searle sets out to disprove is Simon and Newell's
claim that they *are* the same.
> This version of the Systems reply says
This isn't the SR at all.
> that Searle's argument definitely shows that our concept of understanding
> is vague enough to cause us some problems when we go to use it in a
> practical case, but it does not necessarily show us anything about
> strictly _artificial_ intelligence...the force of the CR argument is about
> intelligence and understanding _in general_. Proponents of Strong AI who
> use the vague layman's definition of understanding are going to run into
> the problems Searle states.
I think you should reread the Searle material and the response to it
in _The Mind's I_.
> Just discovered this newsgroup, and it's really pushing my buttons. I
> think I'll stick around.
Rotsa ruck.
--
<J Q B>
Jim Balter
Daryl McCullough wrote:
>
> Jim Balter says...
>
> >Perhaps you are under the false impression that the Church-Turing thesis
> >claims that algorithms can be implemented on a Turing machine,
>
> That's my impression, too.
But that a Turing machine can realize any general recursive function
isn't in dispute, is it?
> The way I remember it (it has certainly been
> a while since I read anything about it, so maybe my memory is playing
> tricks on me), Turing had his notion of computability (in terms
> of Turing machines), and Church had his notion (in terms of the lambda
> calculus) and Post (in terms of Post machines, whatever they are) and
> Godel had his notion (in terms of recursive functions). Although they
> were all different notions of computability, it turned out that they
> were all mathematically equivalent---they all computed exactly the
> same functions. The Church-Turing thesis was that *all* sensible notions
> of computability would turn out to be equivalent to Turing machine
> computability.
But that goes beyond the claim that all algorithms can be implemented
on a Turing machine, does it not? It claims that all computations
can be realized as algorithms, a much stronger claim.
> >rather than claiming that anything computable by a human is computable
> >by a machine.
>
> I think the original work on computability was really about *human*
> computability (there *were* no machines in those days). So I would
> characterize the thesis as "computable implies Turing-computable".
> Maybe I have that wrong, but that's the way I remember it.
I gave Hofstadter's version.
> Here's the way it is put in _Elements of the Theory of Computation_,
> by Lewis & Papadimitriou:
>
> The principle that Turing machines are formal versions of
> algorithms and that no computational procedure will be
> considered an algorithm unless it can be presented as a
> Turing machine is known as Church's Thesis or the Church-Turing
> Thesis.
Ok, there definitely seem to be some terminological disagreements.
I took it that an algorithm is equivalent to a general recursive
function, not a "computation" per se, and that it is the C-T that claims
the equivalence. The C-T as given above appears to be a tautology,
not a thesis as I understand it.
> Strictly speaking, to have noncomputable human behavior would not
> violate Church's Thesis, if that behavior could not be characterized
> as "following an algorithm".
But I read the C-T as asserting that all properly described behavior
can be *modeled* as an algorithm.
> An algorithm has to be guaranteed to
> lead to the correct result, so a procedure that has a step like "Wait
> for a brilliant insight" wouldn't be an algorithm.
I wouldn't call that a proper description of a behavior, any more than
"wait for a unicorn".
> The principle that "Anything a human can do (whether algorithmic or not)
> can be done by a Turing machine" is not really Church's Thesis, I don't
> think.
"wait for a brilliant insight without actually having one"
is not a proper description of "something a human can do";
it cannot be distinguished from "wait", except by modeling
the psychological processes of the human, and we are then heading
right back to computation.
--
<J Q B>
Mark Young
Mark Young:
>> Neurons do not communicate *to* their owners. Neurons are parts of
>> their owner, and so anything going on in them is already there *at* the
>> owner. Likewise I don't communicate *to* my fingers that they should
>> move in certain ways over the keyboard; I simply move my fingers.
>
>> Now it's possible to say that my brain communicates to my fingers that
>> they should move in a certain way---my brain and my fingers are not in a
>> part/whole relationship. Likewise one part of the computer may
>> communicate to another part of the computer "telling" it what it should
>> do.
Anders N Weinstein:
> Right, this is exactly my position. I was making fun of the idea.
You're going to have to be more explicit. What is your position? What
idea were you mocking? My guesses are that your position is that our
neurons do not communicate with us (nor we with our neurons), but that
parts of us communicate with other parts. But then the notion that you
would be mocking would be that we *do* have such communication. But that
doesn't seem to fit with the context:
Neil Rickert:
] >>>During our learning, our neurons are using continuous methodologies
] >>>to analyze the world and construct neural structures that allow us
to
] >>>deal with that world.
Anders Weinstein:
] >>Possibly. But we can't do anything with the 'analyses' made by our
] >>neurons -- for they simply don't communicate these results to us.
Neil Rickert:
] >Oh, nonsense. That is where most of our knowledge is developed.
] >That is where our conceptualizations, and most of our scientific
] >discoveries arise.
Anders Weinstein:
] This claim does not explain how I can do something with an analysis
] my neurons have made for me but not communicated to me.
I interpret Neil as saying that we can deal with the world because our
neurons are in certain structures, developed in reaction to inputs from
the world. You respond that these "results" (the structures?) are not
communicated to us, so we can't do anything with them. But my point
(and I think Neil's) is that the structures *are* us (at least a very
important part of us), so there's no need for any sort of
"communication" at all. We act the way we do because we have those
structures. We *think* the way we do because of those structures.
So what is it you're mocking?
>> But the computer as a whole plays the same role to its parts as we
>>do to ours. And in Strong AI it is the *computer* that is said to have
> Only if, for the purposes of this sort of explanation, we *are* computers.
Maybe you read more into "plays the same role" than I meant. All I
meant was that we can make the same distinction between computers and
their parts as we can between humans and our parts. I said this because
I get the impression that you reject this notion: that you would say
that the computer communicates to the monitor, or that the keyboard
communicates key presses to the computer. I say that the keyboard and
monitor are parts of the computer, and so the only communication *to*
(*from*) the computer in these cases is the communication to (from) the
human using it.
>> And in Strong AI it is the *computer* that is said to have
>>mental states in virtue of its parts acting in certain ways (under the
>>control of software), just as we humans are said to have mental states
>>in virtue of our parts acting in certain ways.
> I am not sure Strong AI makes the same dualism of explanatory
> properties discernable at different levels that I want to make. As
> defined, Strong AI is the thesis that a system can think (have
> intentional states) *in virtue of* instantiating the right program,
> i.e. in virtue of being the right sort of *computer*. I do not think we
> can be said to understand a language in virtue of instantiating a
> program
That's not necessary for Strong AI to be true. "In virtue of" is a
sufficiency relationship, not a necessity.
> -- non-computational interactions with the environment are
> necessary, as is much else, before the concept of understanding gets a
> non-metaphorical grip. And then the inner program can pretty much
> drop out of the account.
I can see that interactions with the environment are necssary for
practical reasons, but I haven't seen any good arguments that they are
logically necessary. The notion that the swamp man doesn't *mean*
anything by the noises /wun twunny thri mein strit/ seems patently false
to me. It seems equally absurd to say that it wouldn't *understand* the
cabbie's response "I don't go to that part of town this late at night,
buddy."
[...]
> Right, but again, this shows they are not part of my mind as I demarcate
> the mental -- they are not involved in giving my subjective perspective
> on the world. Without some relation to a subjective point of view, I think
> there is no way to delimit the mental even loosely -- events in
> my adrenal glands have effects in my psychological state, but are not
> themselves part of my mind. Similarly for events in my neurons, even if
> at their own level they are processing information and "consulting"
> representations I am wholly unaware of.
Many concepts are fuzzy, and the mind is one such. We live in a world
that has no respect for the categories we create in order to better
understand it.
...mark young
Daryl McCullough
Jim Balter wrote:
>
> Daryl McCullough wrote:
> > Here's the way it is put in _Elements of the Theory of Computation_,
> > by Lewis & Papadimitriou:
> >
> > The principle that Turing machines are formal versions of
> > algorithms and that no computational procedure will be
> > considered an algorithm unless it can be presented as a
> > Turing machine is known as Church's Thesis or the Church-Turing
> > Thesis.
>
> Ok, there definitely seem to be some terminological disagreements.
> I took it that an algorithm is equivalent to a general recursive
> function, not a "computation" per se, and that it is the C-T that claims
> the equivalence. The C-T as given above appears to be a tautology,
> not a thesis as I understand it.
It's not a tautology. There is the informal notion of an algorithm,
which
is the notion of systematically following rules to accomplish some task.
Then there is the formal notion of an algorithm, which is defined by
Turing
machines. To claim that the formal notion captures the informal notion
is
not tautologous.
> > An algorithm has to be guaranteed to
> > lead to the correct result, so a procedure that has a step like "Wait
> > for a brilliant insight" wouldn't be an algorithm.
>
> I wouldn't call that a proper description of a behavior,
It's not a description of a behavior, it is a hypothetical instruction
to
a human about how to accomplish some task. Maybe the task is "writing
a poem" or maybe it is "proving a theorem" or maybe it is "winning at
chess".
My point is that some instructions are algorithmic, in that following
them
correctly will be guaranteed to lead to a successful conclusion, while
other
instructions are not algorithmic.
> > The principle that "Anything a human can do (whether algorithmic or not)
> > can be done by a Turing machine" is not really Church's Thesis, I don't
> > think.
>
> "wait for a brilliant insight without actually having one"
> is not a proper description of "something a human can do";
Why not? A person can certainly wait for a brilliant insight.
> it cannot be distinguished from "wait",
As an instruction to a reader, it is certainly different from wait. I'm
not
talking about describing behavior, I'm talking about *prescribing*
behavior.
An algorithm is, by definition, a prescription for behavior leading to
the accomplishment of some task. "Wait for an insight" is a perfectly
good
instruction, I think, but it doesn't necessarily terminate, which makes
it not a legitimate step in an algorithm.
Anyway, my point is not that humans can do things that computers can't.
I
don't believe they can. The question was just what Church's Thesis says
about it.
Neil Rickert
In <335D07...@ace.acadiau.ca> Mark Young <MYo...@ace.acadiau.ca> writes:
>Anders Weinstein:
>] This claim does not explain how I can do something with an analysis
>] my neurons have made for me but not communicated to me.
>I interpret Neil as saying that we can deal with the world because our
>neurons are in certain structures, developed in reaction to inputs from
>the world. You respond that these "results" (the structures?) are not
>communicated to us, so we can't do anything with them. But my point
>(and I think Neil's) is that the structures *are* us (at least a very
>important part of us), so there's no need for any sort of
>"communication" at all. We act the way we do because we have those
>structures. We *think* the way we do because of those structures.
Yes, that is just what I was saying.
Anders N Weinstein
In article <335D07...@ace.acadiau.ca>,
Mark Young <MYo...@ace.acadiau.ca> wrote:
>You're going to have to be more explicit. What is your position? What
>idea were you mocking? My guesses are that your position is that our
>neurons do not communicate with us (nor we with our neurons), but that
>parts of us communicate with other parts.
Yes. Also that we communicate with each other. Sometimes, if we are
acting as brain or cognitive scientists, we communicate ideas about the
"communication" that goes on among our parts.
> But then the notion that you
>would be mocking would be that we *do* have such communication. But that
>doesn't seem to fit with the context:
You are right that I wasn't literally charging Neil Rickert with the
mocked idea. I was trying to use its absurdity to suggest that we
could not use "analyses" or "discoveries" made by our neurons as
part of our account of scientific analysis of data and discoveries
about the world at the personal level, the level of whole communicating
organisms.
I meant: it is one thing for the neurons to "analyze data" or "make a
discovery"; another for a person to analyze data or make discoveries.
The former can only have a causal effect on the scientific analyses or
discoveries of the whole person, it may be a necessary prerequisite,
but it cannot constitute the content of the discoveries made by the
whole person. I think the acts of the person exist on their own level,
and that that is the level at which one ought to analyze scientific
discovery and analysis.
For example, it would not be impossible for the neurons to have, in
their own way, "discovered" how to solve problems using linear
algebra. Perhaps this takes place even among untutored Bushmen who
have next to no mathematical practice. It is something completely
different for a person as a whole to use linear algebra to solve
problems; that involves a molar capacity for operating with public
symbols in accordance with a technique; later it can be done silently
in thought. You might be able to teach the Bushmen to do linear algebra,
in which case the operations of their neurons might be prerequisite.
But only then do the Bushmen themselves learn to represent and solve
the problems.
I think this distinction is pretty uncontroversial, but it seems to
me to tell against the suggested idea that most of our important scientific
discoveries are really made by our neurons in their analyses.
>I interpret Neil as saying that we can deal with the world because our
>neurons are in certain structures, developed in reaction to inputs from
>the world. You respond that these "results" (the structures?) are not
>communicated to us, so we can't do anything with them. But my point
>(and I think Neil's) is that the structures *are* us (at least a very
>important part of us), so there's no need for any sort of
But you started with a vocabulary that suggests they are
emphatically *not* us but are rather parts of us.
>"communication" at all. We act the way we do because we have those
>structures. We *think* the way we do because of those structures.
It depends on what sort of "because" is at issue here. I agree that
we couldn't do what we do if the neurons didn't do what they do, so
that the neural operations are necessary prerequisites. But I also
think there is a description appropriate to the doings of the whole
person, in particular the person who can issue justifications for
conduct and give reasons or arguments or explanations in language.
And that that is the level where "the way we think" is to be found.
>>> But the computer as a whole plays the same role to its parts as we
>>>do to ours. And in Strong AI it is the *computer* that is said to have
>
>> Only if, for the purposes of this sort of explanation, we *are* computers.
>
>Maybe you read more into "plays the same role" than I meant. All I
>meant was that we can make the same distinction between computers and
>their parts as we can between humans and our parts. I said this because
OK.
>I get the impression that you reject this notion: that you would say
>that the computer communicates to the monitor, or that the keyboard
>communicates key presses to the computer. I say that the keyboard and
>monitor are parts of the computer, and so the only communication *to*
>(*from*) the computer in these cases is the communication to (from) the
>human using it.
That division is somewhat arbitrary, I agree. And in a complex system
there are many possible divisions that can be considered.
More fundamental as I see it is the division between what is
computationally explained -- treated as symbol manipulation, in the
central case of symbolic computation -- and what is treated as
peripheral not-computation in the explanation. As Pylyshyn illustrates
in his book _Computation and Cognition_ (or something like that), there
will be a "bridge between the physical and the symbolic" whenever (what
I call) computational explanation of a physical system is being done.
I would say that if you are explaining a device as achieving its
effects through symbolic computation, you pretty much have to have an
interface between the symbolic and the non-symbolic aspects of the
process.
Whereas I think at the level of ordinary person-level psychology, the
idea of an interface between the person and the world simply does not
apply. I think you can see this if you just get a little bit clearer
about the nature of intentional consciousness, (as, I would say, in the
phenomenology of Sartre. That's written in a metaphysical idiom, but
the good point can be extracted).
Subjectively speaking, visual consciousness is often a mode of direct
awareness *of* the public world, unmediated by representations, sensory
data, or anything of that sort at all. As John Searle says quite
rightly, visual experiences are not representational, they are
*presentational* -- in visual experience (sometimes) the world presents
itself to your awareness. As Heidegger put it, the objects "show up" for
you (but only if you have the right sort of understanding).
What lies behind these slogans, I would say, is appreciation of the
fact that one needs to mention the relevant objects in the world to
characterize the content of perceptual experience adequately at all.
For example, you would have to point into my office and say, "Weinstein
is conscious of that desk, that chair, etc, as seen from where he is
standing" in order to characterize my current state of awareness;
otherwise you would not be indicating which objects need to be
consulted to find out if what I say when I express it's content is
true. There are no mental representations involved *at this
phenomenological level*; the person's intentional directedness towards
the office furniture in public space is not mediated by representations
that are themselves intermediate objects of the person's awareness,
that go to explaining the person's epistemic standing with respect to
claims about the world.
To fully make out this line of thought requires some work, but it
includes the view that seeing by the person just *is* a process of
detection of semantic or conceptual information *about* the objects in
the environment *by* the conscious subject (the person who has the
understanding needed to deal with these things under descriptions). In
characterizing veridical perception a theorist needs access to the
environment to even describe or individuate the information content, if
he is to say what it is about, what it answers to. This can only be
done by pointing to the situation itself.
In this sense at the level of the whole person there need not be an
input interface, nor symbolic computation, nor even representation in a
"world model" or by any other means; rather capacities for information
detection or extraction. The information or semantic content itself is
of course a kind of non-physical abstraction. But we can understand it
to be immanent in the physical world, so it should not seem spookily
separated from the activities of embodied creatures.
I really don't see what the concept of computation has to do with any
of this. I take it the concept of computation might or might not be
fruitful for explanations at the sub-personal level. To me, that level
stands to intentional explanation of the person somewhat as chemistry of
paintings stands to art criticism of what is expressed by them.
>> defined, Strong AI is the thesis that a system can think (have
>> intentional states) *in virtue of* instantiating the right program,
>> i.e. in virtue of being the right sort of *computer*. I do not think we
>> can be said to understand a language in virtue of instantiating a
>> program
>
>That's not necessary for Strong AI to be true. "In virtue of" is a
>sufficiency relationship, not a necessity.
Maybe, I am not certain how to understand it myself.
It at least entails a sufficiency relationship, that is true.
Philip Jackson
H. M. Hubey wrote:
>
> Oliver Sparrow <oh...@chatham.demon.co.uk> writes:
>
> >Why don't pocket calculator do the same. Because they are dead,
> >simply chunks of junk when compared to the CNS of a chicken, is why.
>
> Sure, but there are unspoken assumptions, taken for granted
> by techno people. One of them is that motion is an understood
> problem. It's just that its space complexity is large. Since
> we don't need machines that walk, and it's easier to make
> machines that roll, fly or swim, nobody has attempted to
> make such machines.
Actually people have attempted, and are succeeding in making machines
that walk (and even learn to walk).
Philip Jackson
I think you're just egging me on :-)
Neil Rickert
>You are right that I wasn't literally charging Neil Rickert with the
>mocked idea. I was trying to use its absurdity to suggest that we
>could not use "analyses" or "discoveries" made by our neurons as
>part of our account of scientific analysis of data and discoveries
>about the world at the personal level, the level of whole communicating
>organisms.
Of course. In a dualist's world the neurons are irrelevant.
>I meant: it is one thing for the neurons to "analyze data" or "make a
>discovery"; another for a person to analyze data or make discoveries.
>The former can only have a causal effect on the scientific analyses or
>discoveries of the whole person, it may be a necessary prerequisite,
>but it cannot constitute the content of the discoveries made by the
>whole person. I think the acts of the person exist on their own level,
>and that that is the level at which one ought to analyze scientific
>discovery and analysis.
Right. Science should not study what the neurons do, for such a
study would not be at the person level so could have no importance.
>For example, it would not be impossible for the neurons to have, in
>their own way, "discovered" how to solve problems using linear
>algebra. Perhaps this takes place even among untutored Bushmen who
>have next to no mathematical practice. It is something completely
>different for a person as a whole to use linear algebra to solve
>problems; that involves a molar capacity for operating with public
>symbols in accordance with a technique; later it can be done silently
>in thought. You might be able to teach the Bushmen to do linear algebra,
>in which case the operations of their neurons might be prerequisite.
>But only then do the Bushmen themselves learn to represent and solve
>the problems.
I really don't see why you think this sort of argument is useful. I
have never suggested that the neurons discover linear algebra. I
have never suggested that the exact abilities of the neurons
translate verbatim to exact abilities of the person. I suggest only
that when the neurons acquire new abilities, the person acquires new
abilities. That is communication enough. Your deliberate
misconstruals only demonstrate the extent to which you are driven by
dogma.
>More fundamental as I see it is the division between what is
>computationally explained -- treated as symbol manipulation, in the
>central case of symbolic computation -- and what is treated as
>peripheral not-computation in the explanation.
The division should be clear. Symbol manipulation by itself is
empty. Nothing is explained by symbol manipulation alone, except to
a solipsist. Symbol manipulation can only be part of an explanation
to the extent that the symbols can connect with reality, and that
requires what you, in your confusion, take as peripheral
not-computation.
> As Pylyshyn illustrates
>in his book _Computation and Cognition_ (or something like that), there
>will be a "bridge between the physical and the symbolic" whenever (what
>I call) computational explanation of a physical system is being done.
And as Weinstein repeatedly makes clear, that bridge is verboten. It
is taboo. It must not be intestigated. It must be accepted on faith
and on the basis of religious devotion to the almighy philosophy.
>Subjectively speaking, visual consciousness is often a mode of direct
>awareness *of* the public world, unmediated by representations, sensory
>data, or anything of that sort at all.
Well of course. Any solipsist can see that. After all, if there is
no world to represent, there could not be any representations.
> As John Searle says quite
>rightly, visual experiences are not representational, they are
>*presentational* -- in visual experience (sometimes) the world presents
>itself to your awareness. As Heidegger put it, the objects "show up" for
>you (but only if you have the right sort of understanding).
Right. There is a proper dualist language which philosophers must
use in polite society. They must also insist on the use of this
proper language by all scientists, for only by such insistence can
they interfere with the scientific work that they find so threatening
to their precious dualistic religion.
>What lies behind these slogans, I would say, is appreciation of the
>fact that one needs to mention the relevant objects in the world to
>characterize the content of perceptual experience adequately at all.
Fact -- hell no. Analytic truth adopted as a convention by dualistic
philosophers -- sure.
>For example, you would have to point into my office and say, "Weinstein
>is conscious of that desk, that chair, etc, as seen from where he is
>standing" in order to characterize my current state of awareness;
>otherwise you would not be indicating which objects need to be
>consulted to find out if what I say when I express it's content is
>true.
That is what we need to participate in the word games that support
the "Just So" story of epistemology. The concerns of science should
be very different.
> There are no mental representations involved *at this
>phenomenological level*;
From a scientific perspective, whether a representation is 'mental',
or whether 'mental representations' are merely ghosts in the "Just
So" story is irrelevant. It is the actual neural representations
that should concern us.
> the person's intentional directedness towards
>the office furniture in public space is not mediated by representations
Who cares? "Intentional directedness" is not a scientific term, and
it could never become a scientific term if we follow the Weinstein
way.
>To fully make out this line of thought requires some work, but it
>includes the view that seeing by the person just *is* a process of
>detection of semantic or conceptual information *about* the objects in
>the environment *by* the conscious subject (the person who has the
>understanding needed to deal with these things under descriptions).
Neither "person" nor "conscious subject" are scientific terms, and
under the Weinstein way neither could ever become scientific terms.
The aim, quite clearly, is to obfuscate and otherwise interfere with
any attempt at scientific investigation.
> In
>characterizing veridical perception a theorist needs access to the
>environment to even describe or individuate the information content, if
>he is to say what it is about, what it answers to.
In other words, the only possible theory of perception would have to
emanate from a perfectly perceiving God. So science must be
abandoned and we must all turn to theology.
>In this sense at the level of the whole person there need not be an
>input interface, nor symbolic computation, nor even representation in a
>"world model" or by any other means; rather capacities for information
>detection or extraction. The information or semantic content itself is
>of course a kind of non-physical abstraction. But we can understand it
>to be immanent in the physical world, so it should not seem spookily
>separated from the activities of embodied creatures.
>I really don't see what the concept of computation has to do with any
>of this.
Quite right. Theology can manage without computation.
H. M. Hubey
Philip Jackson <pjac...@mail.ic.net> writes:
>Actually people have attempted, and are succeeding in making machines
>that walk (and even learn to walk).
Yes, Iam aware of it. YOu can even see it in movies
where they bounce on a single leg :-).
That should have been taken in context.
Anders N Weinstein
In article <5jhhav$8...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5jguoe$7...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>
>>>since there is no known mathematically perfect answer. Given any two
>>>answers A and B, I can sometimes say that A is a better answer than
>>>B. But only someone confused by binary thinking would claim that,
>>>just because I cannot say whether an answer is right, therefore I can
>>>say whatever I please.
>
>>Fine, now how does this apply to the case at issue?
>
>I don't know. It is merely a counter example which completely
>refutes what you had said. I guess that means it does not apply.
We're talking at cross purposes. I take it you are undertaking to
remind me that on at least some questions (possibly you mean to include
all) there can be a better or worse answer with no best. Therefore it is
a false dichotomy to suggest either there is a right answer or it is
just a matter of whatever seems right.
But that seems fine with me.
I asserted that there are social facts about whether an artifact is a
computer and that these are objective in my sense -- subject to a
seems-right is right dichotomy -- and that they are as objective
matters of fact as any question ever gets. I also asserted that they
are not just "matters of interpretation" *in the partcular sense* of
whatever the outside interpreter takes the objects as. In particular
the question about the standard use in a culture is *not* just a
question of whether the device is useable *as* a calculator. (The
anthropologist is actually acting as an interpreter of other
*interpreters* here, trying to determine how *they* standardly and
conventionally took the artifact in practice.)
Now I think you do not really disagree with anything I said there.
You do not assert that it is a "matter of interpretation" in the particular
sense that I was using. Rather, I take it you are trying to suggest it is
a "matter of interpretation" in your sense, the sense in which just
about every question we ask is a matter of interpretation, something on
which there are pragmatically better or worse answers but possibly no
unique best.
I think I should just grant the point. What you assert, if I have it
correct, does not seem to me to be what I was denying. It seems to me
you are agreeing that this question is as objective as anything is,
insofar as it is subject to standards that are not those of whatever
it seems right to the subject to say. I suspect you agree that the
question about the socio-functional fact -- how was it used? -- is
a different question than the question about the intrinsic functional
fact -- what could it be used for, given its structure?.
I am not really sure what it was in what I originally wrote that
moved you to disagree so strenuously. It is not as if you are pushing
what I see as the crude subjectivism of Pindor on putative social facts.
I can't resist a little more quibbling. I asked you to explain *how*
the particular case was *analogous* to the measurement case. That does
seem necessary if I am to take it as a "counter-example which
completely refutes what I had said". Unless you just meant it to refute
the false dichotomy above. But I don't think that false dichotomy was
behind what I originally claimed about the objectivity or factuality of
an inquiry into social facts. Of course this is a "matter of
interpretation" in a broader sense.
I think of the problems that can arise in such an inquiry rather as a
matter of borderline cases of the concepts, which have a substantial
range of clear applications. So that in a large number of cases, no
such continuum of more or less useful responses arises with such
questions as "was this a drinking vessel" -- they are asked and, if the
evidence is available, answered quite definitely. They are perhaps like
degenerate cases of the continuum of better and worse answers.
So I think I disagree if your idea is that *all* questions we ask
really turn out to be "matters of interpretation". In part this
recalls my disagreement with you on the comparison of all assertions with
measurement.
>>OK. What is the content for you of saying it is "a question of
>>interpretation"? When I said "it is not a matter of interpretation" I
>>meant that the inquiry is taken by all concerned to be objective, in
>>that there is a difference between what is right and what seems right
>>to the archaeologist.
>
>The need to kill Bosnians was probably taken by all Serbs concerned
>to be objective, in that there was a difference (in their view)
>between what is right and what seemed right to an individual Serb.
>Do we then conclude that it was objectively right? Perhaps you are
>suggesting that might makes right?
This example is *way* off the point. To say a question is an objective one
-- or even merely taken as objective -- in my sense is not to say
that anybody's particular view on it is the (or even a) right one.
It is rather a kind of meta-claim about the kind of question it is.
So people can agree that a question is objective and disagree on the
answer.
It is going to be problematic to what extent objectivity applies to
moral matters. I do in fact think we recognize them in practice as
objective in my sense -- not merely a matter of how things are taken to
be (nor merely a question of what is instrumentally "useful" to us to
do). But I don't know to what extent the Serbs in question were
critical and open to the idea that this was a mmatter for which moral
considerations in favor of or against the practice might be advanced
that were independent of its utility or desirability to them.
>>I might be conscious all the while. But still not be conscious of how I
>>am solving the problem
>
>By which you mean that you cannot give an explicit verbal
>description. But if the reasoning was non-verbal, we should not
>expect that you could give a verbal description.
OK, but in what sense is it reasoning? Can you tell me a little more
about its structure, does it proceed step-by-step? Or is its content
completely ineffable? Then in what sense can even I be said to be aware
of it, if I can't tell you how it is done?
>> -- my consciousness might be occupied entirely
>>with such thoughts as "I hope I catch it and don't make a fool of
>>myself" or something like that. More likely, it might be unpacked as "I
>>need to get *there* in time".
>
>In that case you would probably miss the catch most of the time.
>More likely you are concentrating on the ball and consciously
>positioning your hands so as to meet the ball. That requires
OK. Remember I take it that reasoning that can only be expressed using
context-depdendent (indexical) language -- I must move my hands *there*, or
move them *this way*, is still linguistically expressible.
>>> So Galileo could
>>>not have provided any justification. At best, he could have provided
>>>the opportunity for his critics to learn for themselves how to see
>>>through a telescope. Once they had so learned, they would be able to
>>>look toward Jupiter and see for themselves.
>
>>This does not follow. He could explain, *show*, or train them in how to
>>see through the telescope (my "conveying the understanding needed").
>
>Most of what you have to learn cannot be explained verbally.
Depends what you mean. First, verbal instructions and advice will have
a significant role in the training I expect. But second, you may be
right. In some sense it is correct to say that in in transmitting a
skill to someone I am in effect aiming to goad the person's neurons
into certain state changes, such that the person will wind up with the
skill. That does not at all mean that the real scientific discovery is
the "discovery" made by the neurons. The neurons change and thereby
enable the person to see, but the neurons themselves remain blind.
Thirdly, if the folks in question are to take what they see *as*
confirming a theory of the heavens, I think they damn well have to be
able to reason with the propositions of the theory. So that the
perceptual "input" they become empowered to obtain can not be the whole
story -- it must be the input side into a reasoning system, one that
can take it and use it as entries into assertions of those
propositional claims. Otherwise it is at worst just a dangling
perceptual experience, unconnected to any other cognition; at best
something that enables adaptive behavior with respect to the objects,
without the possibility of articulation and justification of any
claims.
So what Galileo might train them to do must be such as to enable
them to take it *as relevant* to the verbal claims he made that
the Earth moves.
>>>No amount of talk or cultural transmission of concepts could have
>>>conveyed that. They had to learn for themselves how to see through
>>>the telescope.
>
>>But why is training someone in how to see through telescopes not a
>>species of cultural transmission of a cultivated capacity?
>
>If the telescope were left lying around, someone who found it might
>learn how to see through a telescope with no cultural transmission.
Sure, monkeys might well do this. But I doubt they could confirm a
theory about the Earth and the heavenly bodies without language,
certainly not a mathematical theory.
Neil Rickert
In <5jjjst$j...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5jhhav$8...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>Now I think you do not really disagree with anything I said there.
>You do not assert that it is a "matter of interpretation" in the particular
>sense that I was using. Rather, I take it you are trying to suggest it is
>a "matter of interpretation" in your sense, the sense in which just
>about every question we ask is a matter of interpretation, something on
>which there are pragmatically better or worse answers but possibly no
>unique best.
Well just a moment. I do not say that every question is a matter of
interpretation. Roughly speaking, the choice of a conceptual scheme
is a matter of interpretation. Some choices may have advantages over
others, but there are no criteria for determining that a particular
choice is correct (in any absolute sense). However, once a
conceptual scheme is chosen, it brings with it standards. So there
are ways of settling questions once there is agreement on a
conceptual scheme in which those questions are to be raised. Even
then, some questions may not have definite answers due to the
inherent uncertainty in measurement.
Now if an artifact from Mars shows up, there are no standards to
determine whether it is a calculator. Our commonly accepted
conceptual schemes do not account for such Martian artifacts.
Therefore we have to modify our conceptual schemes to accomodate it,
and that is what makes it a matter of interpretation.
>>The need to kill Bosnians was probably taken by all Serbs concerned
>>to be objective, in that there was a difference (in their view)
>>between what is right and what seemed right to an individual Serb.
>>Do we then conclude that it was objectively right? Perhaps you are
>>suggesting that might makes right?
>This example is *way* off the point. To say a question is an objective one
>-- or even merely taken as objective -- in my sense is not to say
>that anybody's particular view on it is the (or even a) right one.
>It is rather a kind of meta-claim about the kind of question it is.
There are some types of question that can be evaluated with accepted
standards. There are other types of questions which require the
establishment of standards. You throw around the term "objectively
right" without making such distinctions. I think it is an evasion.
>So people can agree that a question is objective and disagree on the
>answer.
Sure, that is easily possible given the way you misuse "objective."
>>>I might be conscious all the while. But still not be conscious of how I
>>>am solving the problem
>>By which you mean that you cannot give an explicit verbal
>>description. But if the reasoning was non-verbal, we should not
>>expect that you could give a verbal description.
>OK, but in what sense is it reasoning?
You are making assessments of the evidence, and using those
assessments in your choosing your actions.
> Can you tell me a little more
>about its structure, does it proceed step-by-step? Or is its content
>completely ineffable?
No, I cannot say more. This is non-linguistic reasoning, so that it
cannot easily be given a linguistic description.
> Then in what sense can even I be said to be aware
>of it, if I can't tell you how it is done?
Perhaps you are claiming that computers are completely aware and far
more conscious than humans, because all of their decision making can
be described symbolically as an explicit sequence of steps.
>>> -- my consciousness might be occupied entirely
>>>with such thoughts as "I hope I catch it and don't make a fool of
>>>myself" or something like that. More likely, it might be unpacked as "I
>>>need to get *there* in time".
>>In that case you would probably miss the catch most of the time.
>>More likely you are concentrating on the ball and consciously
>>positioning your hands so as to meet the ball. That requires
>OK. Remember I take it that reasoning that can only be expressed using
>context-depdendent (indexical) language -- I must move my hands *there*, or
>move them *this way*, is still linguistically expressible.
Then I think we should take you as claiming that reasoning is
something that the Chinese Room can very well do, but which sometimes
presents difficulties for mere humans.
>>>> So Galileo could
>>>>not have provided any justification. At best, he could have provided
>>>>the opportunity for his critics to learn for themselves how to see
>>>>through a telescope. Once they had so learned, they would be able to
>>>>look toward Jupiter and see for themselves.
>>>This does not follow. He could explain, *show*, or train them in how to
>>>see through the telescope (my "conveying the understanding needed").
>>Most of what you have to learn cannot be explained verbally.
>Depends what you mean. First, verbal instructions and advice will have
>a significant role in the training I expect. But second, you may be
>right. In some sense it is correct to say that in in transmitting a
>skill to someone I am in effect aiming to goad the person's neurons
>into certain state changes, such that the person will wind up with the
>skill.
Well, no, I wouldn't say so. I doubt that you can pick up
significant skills with neural state changes. You would require
structural changes in the organization of the neurons (the way they
are connected). Learning a skill is not a matter of making a
representation (setting neural states), but is a matter of changing
the representation system (the neural structure) so that it can
support representations that could never have existed without those
changes.
> That does not at all mean that the real scientific discovery is
>the "discovery" made by the neurons. The neurons change and thereby
>enable the person to see, but the neurons themselves remain blind.
Before the neural change, the discovery could not have been made.
After the neural change, the 'discovery' seems so obvious that it is
hard to call it a discovery. I can't see what else you would need so
as to agree that the neural change is the important step in the
discovery. Of course, I do see your difficulty. For if the neurons
make the discovery, then we have left out of the picture those
intellectualist stories which you deem so important.
>Thirdly, if the folks in question are to take what they see *as*
>confirming a theory of the heavens, I think they damn well have to be
>able to reason with the propositions of the theory. So that the
>perceptual "input" they become empowered to obtain can not be the whole
>story -- it must be the input side into a reasoning system, one that
>can take it and use it as entries into assertions of those
>propositional claims.
The neural structure creates the concepts which makes reasoning
possible. A lot of reasoning amounts to an effort to become
consciously aware of what is implied by the neural structure.
> Otherwise it is at worst just a dangling
>perceptual experience, unconnected to any other cognition; at best
>something that enables adaptive behavior with respect to the objects,
>without the possibility of articulation and justification of any
>claims.
That sounds like dualist talk. You seem to be saying that the
neurons only do so much. After that, it is up to you - presumably
your immaterial soul -- to reason out the rest. I suggest that is
nonsense.
Chris Malcolm
In article <5iqn05$5...@ux.cs.niu.edu> ric...@cs.niu.edu (Neil Rickert) writes:
>In <33507B...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>>I am confused by the Systems Reply if it is only a bald assertion. I
>>take
>>it as something more. The System in question is just following rules.
>Is it? That is not obvious.
Searle asserted that it worked by following explicit rules. His Chinese
Room is a perfected version (supposing that to be possible) of Schjank's
script-based "understanding" systems, which are rule-based systems.
At a level beneath the AI concept of a "rule-following system" we also
have the fact that this is in turn implemented as a computer program;
and a computer program is an ordered set of rules about how to do
something.
So at (at least) two levels of description it is a rule following
system, and we furthermore have Searle's word for it that he *intended*
it to be rule following system.
>My automobile is a mechanism. It can be descibed as following
>rules. Those rule specify, for example, the relation of the opening
>of the distributer contacts to the position of the pistons in the
Yes, but this is a completely different sense of "rule following" from
the above. In this case the behaviour conforms to physical laws and
design expectations, but the behaviour is *not* generated by
interpreting explicit representations of those rules. The meaning of
"rule following system" in CS, AI, and Searle's Chinese Room argument,
is the latter. The former is a red herring.
--
Chris Malcolm c...@dai.ed.ac.uk +44 (0)131 650 3085
Department of Artificial Intelligence, Edinburgh University
5 Forrest Hill, Edinburgh, EH1 2QL, UK DoD #205
Chris Malcolm
In article <5iud7i$a...@ux.cs.niu.edu> ric...@cs.niu.edu (Neil Rickert) writes:
>The distinction you are making seems purely arbitrary. In the early
>days of computing, the programming was hard wired into the computer.
>The choice to place the program in memory, where it can be considered
>an inner instruction book, is purely one of convenience.
It is very far indeed from being one of pure convenience. It is the
absolutely crucial feature which enables the construction of levels of
virtual machinery. Theoretically this is what enabled the specific
purpose Turing Machine to be generalised to the general purpose Turing
Machine, i.e., it is what enabled the existence of the general purpose
computer both to be proved, and to be realised in practice. If theory is
not your bag, think of software tools.
Philip Jackson
H. M. Hubey wrote:
>
> Philip Jackson <pjac...@mail.ic.net> writes:
>
> >Actually people have attempted, and are succeeding in making machines
> >that walk (and even learn to walk).
>
> Yes, Iam aware of it. YOu can even see it in movies
> where they bounce on a single leg :-).
>
> That should have been taken in context.
Sorry, but I don't see how what you said was taken out of context.
Indeed, I provided more context in replying to your assertion, than you
have provided above. Here is the context again:
H.M. Hubey wrote:
>>>
>>> Oliver Sparrow <oh...@chatham.demon.co.uk> writes:
>>>
>>> >Why don't pocket calculator do the same. Because they are dead,
>>> >simply chunks of junk when compared to the CNS of a chicken, is why.
>>>
>>> Sure, but there are unspoken assumptions, taken for granted
>>> by techno people. One of them is that motion is an understood
>>> problem. It's just that its space complexity is large. Since
>>> we don't need machines that walk, and it's easier to make
>>> machines that roll, fly or swim, nobody has attempted to
>>> make such machines.
I quoted the above to provide context and replied:
>>Actually people have attempted, and are succeeding in making machines
>>that walk (and even learn to walk).
Please clarify how your assertion that nobody has attempted to make
machines that walk is valid in some context.
Mark Young
Mark Young:
>> I interpret Neil as saying that we can deal with the world because our
>> neurons are in certain structures, developed in reaction to inputs from
>> the world. You respond that these "results" (the structures?) are not
>> communicated to us, so we can't do anything with them. But my point
>> (and I think Neil's) is that the structures *are* us (at least a very
>> important part of us), so there's no need for any sort of
Anders N Weinstein:
> But you started with a vocabulary that suggests they are
> emphatically *not* us but are rather parts of us.
My apologies. Runaway rhetoric. The phrasing "the structures *are* us"
just sounded so pleasing that I was loth to give it up. I added the
parenthetical statement as clarification, but reading it back it looks
like equivocation.
>> "communication" at all. We act the way we do because we have those
>> structures. We *think* the way we do because of those structures.
> It depends on what sort of "because" is at issue here. I agree that
> we couldn't do what we do if the neurons didn't do what they do, so
> that the neural operations are necessary prerequisites. But I also
> think there is a description appropriate to the doings of the whole
> person, in particular the person who can issue justifications for
> conduct and give reasons or arguments or explanations in language.
That there is another level of description appropriate does not make any
difference to the "because" that I can see. We do not think the way we
do because of the kind of person we are; rather we are the kind of
person we are because of the way we think (among other things).
> And that that is the level where "the way we think" is to be found.
OK, so there are two levels of description for a human: personal and
sub-personal.
[... ... ...]
> I really don't see what the concept of computation has to do with any
> of this. I take it the concept of computation might or might not be
> fruitful for explanations at the sub-personal level. To me, that level
> stands to intentional explanation of the person somewhat as chemistry of
> paintings stands to art criticism of what is expressed by them.
The AI position is that the concept of computation *is* fruitful for
explanations of human activity. The Strong AI thesis is that if the
computer runs the right programs, *whatever they happen to be*, a person
will arise. And when the computer is running those programs (and is
thus a person, per hypothesis), any references to the programs
themselves are *necessarily* sub-personal references.
...mark young
Neil Rickert
In <5jkti3$8...@scotsman.ed.ac.uk> c...@dai.ed.ac.uk (Chris Malcolm) writes:
>In article <5iqn05$5...@ux.cs.niu.edu> ric...@cs.niu.edu (Neil Rickert) writes:
>>In <33507B...@accessone.com> Gary Forbis <for...@accessone.com> writes:
>>>I am confused by the Systems Reply if it is only a bald assertion. I
>>>take
>>>it as something more. The System in question is just following rules.
>>Is it? That is not obvious.
>Searle asserted that it worked by following explicit rules. His Chinese
>Room is a perfected version (supposing that to be possible) of Schjank's
>script-based "understanding" systems, which are rule-based systems.
Searle claimed that his conclusions applied to all ways of doing AI,
not just to those using Schank's scripts.
>At a level beneath the AI concept of a "rule-following system" we also
>have the fact that this is in turn implemented as a computer program;
>and a computer program is an ordered set of rules about how to do
>something.
I was not disagreeing with that. But that is rule following for
symbols that refer to internal operations. From that one cannot
conclude that it is following rules with regard to the public symbols
of the input.
>>My automobile is a mechanism. It can be descibed as following
>>rules. Those rule specify, for example, the relation of the opening
>>of the distributer contacts to the position of the pistons in the
>Yes, but this is a completely different sense of "rule following" from
>the above. In this case the behaviour conforms to physical laws and
>design expectations, but the behaviour is *not* generated by
>interpreting explicit representations of those rules.
I have not seen a single example of a digital computer, where the
rule following specified in the program is not being carried out by
virtue of the conformance with physical laws.
Neil Rickert
In <5jku35$8...@scotsman.ed.ac.uk> c...@dai.ed.ac.uk (Chris Malcolm) writes:
>In article <5iud7i$a...@ux.cs.niu.edu> ric...@cs.niu.edu (Neil Rickert) writes:
>>The distinction you are making seems purely arbitrary. In the early
>>days of computing, the programming was hard wired into the computer.
>>The choice to place the program in memory, where it can be considered
>>an inner instruction book, is purely one of convenience.
>It is very far indeed from being one of pure convenience.
We will have to disagree on that.
> It is the
>absolutely crucial feature which enables the construction of levels of
>virtual machinery.
That in no way contradicts what I said.
> Theoretically this is what enabled the specific
>purpose Turing Machine to be generalised to the general purpose Turing
>Machine,
And it is that generalization of the universal TM which demonstrates
the ease with which one can interchange hardware with software. And
that is what make the distinction one of convenience.
> If theory is
>not your bag, think of software tools.
But theory is my bag, or part of it. The point I was making, is that
given a computational process, there is no way of specifying what is
the software and what is the hardware. Given a particular physical
computer, then we can make a distinction for that device. But the
distinction has to be made on the basis of the hardware used, rather
than on the nature of the computation treated as an abstraction.
In the case of designing an AI system, what is put into hardware, and
what is done in software, is a matter of convenience (which includes
cost, engineering practicality, and other such considerations.)
Anders N Weinstein
In article <5jkdil$e...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>In <5jjjst$j...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>>In article <5jhhav$8...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>
>>Now I think you do not really disagree with anything I said there.
>>You do not assert that it is a "matter of interpretation" in the particular
>>sense that I was using. Rather, I take it you are trying to suggest it is
>>a "matter of interpretation" in your sense, the sense in which just
>>about every question we ask is a matter of interpretation, something on
>>which there are pragmatically better or worse answers but possibly no
>>unique best.
>
>Well just a moment. I do not say that every question is a matter of
>interpretation. Roughly speaking, the choice of a conceptual scheme
>is a matter of interpretation. Some choices may have advantages over
OK. In my terms you want to make something like Carnap's distinction
between internal questions -- those made within some accepted
conceptual framework -- and external questions -- questions about the
choice of a framework. Like Carnap, you want to say that questions of
the second sort are subject only to pragmatic evaluations (a better or
worse, for a purpose, in a context, without necessarily a unique best).
And so questions of interpretation.
I tend myself to see both sorts of questions as on a par.
>choice is correct (in any absolute sense). However, once a
>conceptual scheme is chosen, it brings with it standards. So there
>are ways of settling questions once there is agreement on a
>conceptual scheme in which those questions are to be raised.
Right.
>Now if an artifact from Mars shows up, there are no standards to
>determine whether it is a calculator. Our commonly accepted
>conceptual schemes do not account for such Martian artifacts.
>Therefore we have to modify our conceptual schemes to accomodate it,
>and that is what makes it a matter of interpretation.
Good, now I can see better what we disagree on. I took it we would be
applying the more or less generic ontological category (= conceptual
framework) of social functional artifacts to the thing. We would also
be applying the somewhat more specific concept of device for automating
calculation procedures. But both of those are concepts that we already
have in place when we ask the anthropologist's question I alluded to.
So in the above terms I took it to be an "internal" question, one
within an accepted framework, not an external question which required
development of a new framework.
>>>The need to kill Bosnians was probably taken by all Serbs concerned
>>>to be objective, in that there was a difference (in their view)
>>>between what is right and what seemed right to an individual Serb.
>>>Do we then conclude that it was objectively right? Perhaps you are
>>>suggesting that might makes right?
>
>>This example is *way* off the point. To say a question is an objective one
>>-- or even merely taken as objective -- in my sense is not to say
>>that anybody's particular view on it is the (or even a) right one.
>>It is rather a kind of meta-claim about the kind of question it is.
>
>There are some types of question that can be evaluated with accepted
>standards. There are other types of questions which require the
>establishment of standards. You throw around the term "objectively
>right" without making such distinctions. I think it is an evasion.
I think our conception of objectivity involves susceptibility to *both*
sorts of evaluation -- you don't have a grasp on objectivity unless you
are critical, unless you appreciate that accepted standards can themselves
come in for criticism. That is in fact exactly why I reject the distinction
between internal and external questions *if* it is supposed to say that
objectivity only applies to internal ones (as it seemed to Carnap).
For example, conventionally accepted standards had it at one time
that Negro slaves were not to be treated as full persons. Those standards
were the wrong ones. Recognizing that they were the wrong ones might be both
a paradigm shift and a both discovery of a moral reality. In this way
it might be on a par with recognizing that the standards constituting
phlogiston chemistry were the wrong ones, which is also (in my terms)
both a paradigm or conceptual shift and a discovery
about the world (that there is no phlogiston).
At any rate it is important to recognizing a question as objective
in my sense to recognize a certain open-endedness, that it is not just a
question limited to evaluation by presently accepted standards.
>> Then in what sense can even I be said to be aware
>>of it, if I can't tell you how it is done?
>
>Perhaps you are claiming that computers are completely aware and far
>more conscious than humans, because all of their decision making can
>be described symbolically as an explicit sequence of steps.
Well I do not believe that they are conscious at all because the computer
as such cannot perceive the world, doesn't care about the world,
doesn't show itself responsible to standards, and in most cases
cannot join in critical reflection on the standards.
But I think it is right that what they do do is *analogous* to a high
level of conscious awareness in that all the steps are very explicit
and verbally articulable. I think there is a good usage according to which
greater explicitness corresponds to a higher degree of conscious
awareness.
>>OK. Remember I take it that reasoning that can only be expressed using
>>context-depdendent (indexical) language -- I must move my hands *there*, or
>>move them *this way*, is still linguistically expressible.
>
>Then I think we should take you as claiming that reasoning is
>something that the Chinese Room can very well do, but which sometimes
>presents difficulties for mere humans.
I don't think you are appreciating the difference for me between using
public symbols in real contexts in the world (including, in part, their
use to express silent reasonings) and the supposed manipulaton of inner
symbols inside the brain. The former can be meaningful and can be used
to show us what some conscious and explicit reasoning is. The latter
is mostly a myth, I think.
Consider: if the Chinese room were a robot that could point at things
and say "this" and "there", and manifest other abilities to make use of
these in its life and interactions with things, then sure it could show
itself capable of doing some explicit reasoning sometimes.
For example the following sentences express a bit of reasoning that might
constitute a good argument if uttered in the right context:
that's is a house [points to house1]
if I go over there [point to location facing side of house1]
I shall see the side of it
of course we don't *have* to consciously go through such step by step
rigamaroles. But we can articulate them nonetheless in giving reasons.
Now I just don't see that the supposed symbols inside the brain can do
this. They can't use indexicals like "this" or "there" or "that" in
accordance with purely formal rules. Standard formal logic simply never
deals with these devices, it uses only *context-independent* devices.
That is why standard formal logic is doomed to empirical
meaninglessness and is appropriate only to a Platonic intellect
contemplating pure abstractions like numbers and sets. Because in order
for its material (extra-logical) symbols to have meaning they must be
*connected* to things in the world.
I take it this connection comes in *perceptual consciousness*, in the
ability to apply concepts to things in the world. In doing so one can
enjoy propositional attitudes whose intentional content is *only*
expressible via such context-dependent devices as the ones above.
So I think I agree with you that traditional formal symbolic models
are doomed to meaninglessness. I think meaning comes in through
the ability to have "demonstrative thoughts" of the sort expressed above.
But only a creature embodied in the world could have such a thing,
not a disconnected symbol manipulator.
I am not in fact clear on how proponents of the symbol-manipulation
paradigm hope to deal with the issue of demonstrative thoughts; mostly
they either ignore the problem or introduce non-conceptual mental content
like sense-data to handle it. But I want to do epistemology without
non-conceptual content or sense-data, instead resting on a special
sort of intentional state, namely demonstrative thoughts whose content
derives in part from bodily situation in the world.
>Well, no, I wouldn't say so. I doubt that you can pick up
>significant skills with neural state changes. You would require
>structural changes in the organization of the neurons (the way they
>are connected). Learning a skill is not a matter of making a
>representation (setting neural states), but is a matter of changing
>the representation system (the neural structure) so that it can
>support representations that could never have existed without those
>changes.
OK, you are making a distinction between changes in "state" and
changes in "structure". I was not employing any such distinction (surely
in a broad enough sense, changes in "structure" count as changes in
state).
>>Thirdly, if the folks in question are to take what they see *as*
>>confirming a theory of the heavens, I think they damn well have to be
>>able to reason with the propositions of the theory. So that the
>>perceptual "input" they become empowered to obtain can not be the whole
>>story -- it must be the input side into a reasoning system, one that
>>can take it and use it as entries into assertions of those
>>propositional claims.
>
>The neural structure creates the concepts which makes reasoning
>possible. A lot of reasoning amounts to an effort to become
In my view this is a bad use of "concept". Neurons don't have concepts
and can't apply or use concepts or conceptual structures. For they can
only respond to proximal stimulation. But concepts must apply to things
in the public world. This is where I think you doom your view to
solipsism or idealism.
I think concept application involves skills for dealing with objects in
the world. Conceptual structures are better thought of as consisting
of the rules or standards applied by whole people to these things in
the world in their interactions with them.
Anders N Weinstein
In article <335E51...@ace.acadiau.ca>,
Mark Young <MYo...@ace.acadiau.ca> wrote:
>> And that that is the level where "the way we think" is to be found.
>
>sub-personal.
In a way you could say that is my whole point. There is a parallel
distinction even in the case of animals.
I would recommend especially John McDowell's nice paper "The
Content of Perceptual Experience" in _Phil Quarterly_ 44 (1994).
Note that on that view the two levels come with their own distinct
notions of representational content -- at the person level,
vehicle-less intentional states individuated semantically somewhat a la
Searle; at the sub-personal level, transformations of representational
vehicles whose content is individuated functionally.
>The AI position is that the concept of computation *is* fruitful for
>explanations of human activity. The Strong AI thesis is that if the
>computer runs the right programs, *whatever they happen to be*, a person
>will arise. And when the computer is running those programs (and is
>thus a person, per hypothesis), any references to the programs
>themselves are *necessarily* sub-personal references.
I take it that what is intended by Strong AI is mainly AI in the sense
of Haugeland's GOFAI. I think this involves some further commitments
about the nature of the programs, in particular it involves a view
about the *semantics* of the representations being manipulated.
For example, I think it requires that the computer program that is
inside a person playing chess has representations about the chess
board, and indeed that the contents of the person's beliefs are pretty
much the same as those semantic contents represented in the program.
So that there is a kind of identity between thought as commonly
conceived and the inner symbolic computation.
Anders N Weinstein
In article <5jkti3$8...@scotsman.ed.ac.uk>,
Chris Malcolm <c...@dai.ed.ac.uk> wrote:
>>[Rickert:]My automobile is a mechanism. It can be descibed as following
>>rules. Those rule specify, for example, the relation of the opening
>>of the distributer contacts to the position of the pistons in the
>
>Yes, but this is a completely different sense of "rule following" from
>the above. In this case the behaviour conforms to physical laws and
>design expectations, but the behaviour is *not* generated by
>"rule following system" in CS, AI, and Searle's Chinese Room argument,
>is the latter. The former is a red herring.
Actually as Neil mentioned in another post, computers can be (and
originally mostly were) "hard-wired" devices that do not interpret
explicit representations of their programs. For example, a digital
logic circuit such as an and-gate or a complex of such elements like a
binary adder.
In the beginning _Psychosemantics_, I believe, Jerry Fodor asserted
that it was *not* crucial to his Representational Theory of the Mind
that the program instructions -- the rules for transforming symbolic
representations -- be themselves explicitly represented in they system,
although they might be. What he said *is* crucial is that the *data
structures*, the symbolic representations being successively
transformed one into another, be represented.
In the adder, they would be the voltages constituting binary digits at
the input and output in the computational interpretation. In a parser,
they might be representations of the syntactic structure or meanings
of the input sentence. And so on.
The real point, I take it, is that the automobile engine is not
involved in transforming or manipulating representations of any kind.
So it is certainly not following symbolic computational rules.
Neil Rickert
In <5jmb45$1...@usenet.srv.cis.pitt.edu> ande...@pitt.edu (Anders N Weinstein) writes:
>In article <5jm4h2$h...@ux.cs.niu.edu>, Neil Rickert <ric...@cs.niu.edu> wrote:
>>>For example the following sentences express a bit of reasoning that might
>>>constitute a good argument if uttered in the right context:
>>> that's a house [points to house1]
>>> if I go over there [point to location facing side of house1]
>>> I shall see the side of it
>>>of course we don't *have* to consciously go through such step by step
>>>rigamaroles. But we can articulate them nonetheless in giving reasons.
>>For myself, I see no reasoning at all there. All I see is a verbal
>>report of reasoning that might have taken place. For example, if I
>What the verbal report expresses is a certain bit of reasoning.
I still do not detect any reasoning.
> Note
>also that reason-giving explanations can often be true even if the
>reasoning did not consciously occur
A random character generator might happen to generate some true
statements. It would not follow that reasoning occurred.
>>But if it was are report of reasoning, then the reasoning was in the
>>evaluation of the evidence, and the making of decisions on the basis
>Which evidence? As I understand it "that's a house" *is* the evidence,
>at least some of it.
It is only some of it -- a very small part.
>>By contrast, the computers available to us today have no way of doing
>>a corresponding evaluation of the evidence. They are limited to a
>They have no evidence at all, I would say, since they are incapable
>of perceptual consciousness.
Given that you have no experience of being a computer, you cannot
really know that. More to the point, you really cannot justify such
claims except on the basis of an effective theory of perception. But
you object to any attempt to discuss potential theories.
>I think I avoid the problem in the case of people because I do not
>think of publicly expressible reasoning as "syntactic" or
>formalizable. My view is that mental states and such cognitive
>transitions are *expressible* in public language, which may have a
>syntax. But the states themselves are not purely syntactic, they are
>semantic. The public symbols embody the semantics in words.
Well then there is no problem. If the symbols embody the semantics
in the words, the the Chinese Room just had to have had semantics,
contra Searle.
>>Well, I don't quite know what point you are making. I did not
>I am saying the *real* problem with "logicism" is that classical
>logical approaches simply cannot deal with the context within which
>alone symbols and mental states get their meaning.
Not so. There is no magic in context, although there might be an
increase in complexity. The real problem is that it based on the
type of intellectualism supported by people such as Weinstein. But
such intellectualism provides a seriously deficient account.
>Yes, of course. But by my lights your criticism of "formalism" or
>"syntactic" or "logical" conceptions is misapplied, it throws out
>the baby with the bathwater. Of course context-independent symbols in
>the head cannot get meaning. But that does not prevent us from saying
>that context-dependent symbols used by living human beings in the world
>can express meaningful intentional states.
You can say it, but that does not make it right.
>You might put it this way: you say: an abstract formal system is empty
>and therefore requires connection to non-symbolic "information". I
>think I agree in some sense with the idea that abstract formal systems
>are empty. But what it requires I think is grounding in demonstratively
>expressible thoughts (i.e. perceptual consciousness), something left
>out of classical formal systems which are free of demonstratives.
But this is an empty claim as long as "perceptual consciousness" is
itself undefined.
>That in turn is only possible on top of a background of embodied
>skills for interacting with the world and much else.
Empty words, as you use them. They are no more than part of the
excuse which allows you to remain in self denial over your substance
dualism.
>>One of my criticisms of logic, as used by philosophy, is that too
>>often it is like the example you have given above. The real
>>reasoning is not by logic, and is non-linguistic. But the nature of
>Why can't the "real reasoning" be whatever those symbols
>in context *express*?
I will agree to that when you agree that the Chinese Room has
semantics and is capable of real reasoning. It was part of Searle's
hypothesis that the CR be behaviorally correct, and that would
require that it got the context dependence right.
>In general, I think our only handle on mental states comes via their
>linguistic (and other) expressions. But the states are one thing, the
>expressions another. I never said the reasoning was done *in* language.
>Only that its content was expressible in such language.
We don't need a handle on mental states.
>>the real reasoning is ignored, and instead we are presented with a
>>description which has been given so as to have the form of logic.
>>But it is only a pretense that logic was used in the reasoning.
>You have given no reason to think that the articulation is a pretense.
Actually, I have given many over the months of debating. But you
always reject them as incompatible with your dogma.
>I don't think the reasoning must be done with symbols. I don't think
>it must be done with anything. (Who was it who answered: "Do you think in
>language or pictures?" with "I think in thoughts" ?).
Empty rhetoric.
>Again, it's content must be expressible in language. If you like, I
>take intentional states to be potentialities for being *expressed*.
That is part of your commitment to an unsupportable intellectualism.
>If I am conscious as a linguistic creature of a house as a house, it is
>in part because I have acquired the verbally infected concept of a
>house and can express it using the words "that's a house". That is
>hardly very magical.
More intellectualist dogma.
>I do not need non-conceptual sense-data if I have a perceptual state
>aptly expressed by "I see that house".
In any case, the term "sense data" is just as unclear as the term
"mental state."
> I don't see what problem you are finding. In part
>it requires an ability to respond behaviorally to houses in the world.
The wind responds behaviorially to houses in the world.
>In another part it requires an understanding of what a house is, what
>it's for. In another part it requires an ability to perform inferences
>about it.
But you explain nothing if you use "understanding" to explain
"intentionality," and then use "intentionality" to explain
"understanding."
>>What you mean by that, I take it, is that your purely subjective way
>>of looking at things implies dualism and solipsism are the only
>>possible alternatives, and you vote for dualism.
>That is a very strange way of looking at it. It is rather simply
>that if your view commits you to solipsism then you lose.
Well my view does not commit me to solipsism, and your claims to the
contrary have not demonstrated otherwise.
>I don't understand why you keep thinking there is some problem with my
>"dualism" Is there something wrong with distinguishing different
>classes of predicate ("property dualism") that I haven't heard? Isn't
>anyone who offers computational explanations already a property
>dualist, distinguishing software from hardware properties? Isn't it as
>innocuous as distinguishing the economic attributes of a dollar bill
>from its physical structure? Don't predicates sort into classes in
>that way?
You deceive yourself. You are a substance dualist in denial.