Re: Symbol Grounding Problem (attn: Vend)


"HMSBeagle" <jsbach@xxxxxxxxxxxxx> wrote in message
news:dmg533dbffueailpi477154hklast59vpv@xxxxxxxxxx
On Fri, 27 Apr 2007 22:50:08 GMT, Neil W Rickert <phishing@xxxxxxxxxx>
wrote:

HMSBeagle <jsbach@xxxxxxxxxxxxx> writes:
On Wed, 25 Apr 2007 14:55:14 GMT, Neil W Rickert <phishing@xxxxxxxxxx>
wrote:

The real problem for humans is not grounding symbols. Rather it
is symbolizing the ground. We don't start with words and seek to
discover their meaning; we start with meanings and seek to usefully
attach words to that meaning.

Woah! I think you completely misread the article! The Symbol
Grounding Problem is a problem for AI researchers attempting to build
a robotic agent that can engage in symbol grounding. It's not a
problem that humans face, and the article never made any such claim.

Harnad is approaching the SGP from a psychology standpoint, which is
fine. From this perspective the question is not "how do we build it
or simulate it?" but rather "how do actual organisms perform it?"

And that is precisely what I was commenting on. Evidently I did
not misread the article after all. And my comment was that actual
organisms don't ground symbols.

I do not accept ungrounded symbol systems as the likes given by French
structuralists, or those given by semantic nets a la NLP.

I do not beleive that symbols refer to symbols which refer to symbols
in a disembodied run-around that never points back to an object in the
world. Even if you take the stance that "there are no objects" in
some sort of quantum-mechanical argument, I may then reply that there
are at least SHARED EXPERIENCES with the world between seperate
humans. And these "shared experiences" can be looked at rigourously
(even scientifically) as corpuscular entities; in short, as OBJECTS.

I believe that we do live in a universe that contains seperable
entities called OBJECTS, and do I beleive that human language is meant
to point to them as physical REFERENTS. The fact that our human
languages contain nouns is not an accident or a dream of false
neurological illusions.


And furthermore, you are wrong that we do not start with words and
seek to discover their meaning.

Strictly speaking, words don't have meanings. People have meanings,
and use words as they attempt to convey those meanings. Sure,
we are sometimes faced with spoken or written expressions that we
must try to understand. But how we do this -- or at least how I
do this -- isn't really discovery of meanings of words.

I agree. The author has an "intended meaning" that the reader is
meant to discover. Meanings are only within people.

What exactly did you think I was claiming here? Did you think that I
was claiming that the symbols contain the actual meaning WITHIN THEM,
in a physical sense? That's kind of absurd.


The article then gets into what the proper "meaning" entails, does it
also entail the referent itself existing in the real world? Is that
seperate?

I think you mean that it gets into what some philosophers of
language, particularly those of the Frege school, take "meaning"
to entail. You may safely assume I am not an adherent of that
point of view.

There's no reason to bring Frege up here. If an object in the world
makes a group of hunter gatherers slap a name on it so that they can
communicate, then how do we, as scientists, act as if we can remove
the referent from the "meaning" of their chosen name?

Maybe we can! If you want to make that argument then go ahead, but I
will not accept it if I must agree to an ungrounded symbol system. I
don't beleive in those anymore.


2nd Example: Have you ever read a novel? The book itself is just
ink-markings on pages. In effect, you START WITH SYMBOLS and have to
SEEK to discover their meaning.

Bad example. When I am reading a novel, most of the words used
are already in my vocabulary. So no meaning discovery is required for
reading that novel.

Every sentence has a new meaning. You are discovering its meaning.
(Or to be more scientific, you are discovering the intended meaning of
the author, since meanings are only "inside people", as above)


Any time you read a sentence that
you have never seen before in your life, you must seek to discover its
meaning by parsing it.

Oh, nonsense. We can normally read without having to parse the
sentences. We learn to read before we learn how to parse.

You can't avoid parsing it. Reading involves parsing. When you were
listening to your mommy talk as a toddler, your brain was PARSING her
sentences, wether you were aware of this or not. You may have been
using her inflection as parsing cues. In fact, I'm certain you were.


The SGP, as the article states, seems to be the set of "rules" with
which an organism goes from symbol to referent.

And I am disagreeing with this whole "set of rules" idea. There is
no such set of rules.

Okay... read this very slowly:

Consider three phrases:
1) "Tony Blair"
(2) "the UK's current prime minister"
(3) "Cheri Blair's husband"

They all point to the same referent, but none of them have the same
meaning. It is insufficient to say that human beings simply
associate these phrases with the referent via conditioned coupling in
which both stimulus and phrase appear at the same time.

Who is it that said this? But still, are you arguing that the presence of,
say, a dog, is irrelevant to "acquisition of the meaning" of "dog"? Whatever
happened to your bold stance concerning the interaction of animal and
environment? You know, the "modern" view that you like to tout?

That is
wrong. What human beings do, however, is create this thing called
MEANING in their brains which then points them to the correct
referent, even if they have never seen the phrase before in their
lives.

So, is their history with respect to "the phrase" irrelevant? And, if
relevant, what exactly is the role? "BZZT Niccckkkllllrrwm HGFGFGF." What's
the "referent"? Did your brain point the way?


It is through the unique and creative meanings of phrases that human
beings are able to "solve the puzzle" of connecting creative phrases
to actual referents in the world. They way in which we do this is
mostly unconscious, but it can be done deliberately as well. In
either case, there is a set of "rules" which we use in order to "solve
the puzzle" of a unique phrase (we have never seen before ) and
identify the proper referent.

What, precisely, is the role of the animal's "interaction with the
environment" in producing the, ummm, "phenomenon" that you describe above?
What creates the "rules"? How does the animal "know how to follow the
rules"? How does it "know how to follow 'unconscious rules'"? How does the
person know when to use the rules? How do they know which rule to employ? Is
the employment of rules itself a matter of following rules? If so, where do
those rules come from? Why are you an idiot?


Furthermore, it is wrong and insufficient to say that what humans have
done is couple the MEANING with the phrase via conditioning in which
the stimulus and MEANING appear at the same time.

Who is it that said this? But still, are you arguing that the presence of,
say, a dog, is irrelevant to "acquisition of the meaning" of "dog"? Whatever
happened to your bold stance concerning the interaction of animal and
environment? You know, the "modern" view that you like to tout? Why are you
an idiot?

How do I know this
is wrong? Because the meaning is different in each of the Tony Blaire
phrases above.

There is no coupling, or repetive appearances of these stimuli which
can lead a human being to accidently associate them because their
neurons are associating them.

Are you arguing that the presence of, say, a dog, is irrelevant to
"acquisition of the meaning" of "dog"?

Humans create unique and creative
sentences ON THE FLY,

Right! And that's exactly why you need to understand the history that
produced this. And I thought you were stupid!

and relay them to other human beings ON THE FLY,
which transmits a unique meaning which neither human nor speaker has
considered before.

There is no training phase to the above phrases.

So, the animal's interaction with the environment is irrelevant?

There is no
conditioning cycle, there is no reward and punishment motivating
synaptic chemical changes that are uncontrollable by the listener.
The meaning is produced SUDDENLY and uniquely.

So because of all of the above, there must be a methodology by which
our brains are solving these meaning puzzles , in the sense of
connecting them to proper referent. Because there is a method, there
must be a SET OF RULES to this method.

To say that you are a pinhead is just so, well, inadequate. Oh, BTW, I
noticed that you didn't mention much about the animal's interaction with the
environment which, at one point, was the be-all-and-end-all of your view.
Where do the "rules" come from? How do they know how to follow them.
"NHHHYTYRT nslkujtf kkkkflfop!" Follow that rule, dumb ***!


.

Quantcast