Re: Universal grammar

In article <1162808452.699396.36940@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>, "Franz
Gnaedinger" <frgn@xxxxxxxxxxx> wrote:

I try to rely at first hand sources. :-) Such stuff is sometimes discussed
in newsgroups like sci.astro.research and sci.physics.research, and
sometimes it is not possible even for dedicated experts to figure what is
meant in those massmedia popular science reports. :-)

Original publications in physics are too demanding for me,
and so I must rely on journals such as Science, Nature,
The American Scientist, The Scientific American, New
Scientist, and on scientific sections of our newspapers.
Can I trust an article?

I can give one example : in the newsgroup sci.astro.research, there was
recently a thread "The universe changes shape", where some respondents
fell that the idea of giving the universe a shape and other do not. Giving
the universe itself a shape suits well in a popular science magazine as it
is highly intuitive, but it is unclear whether the idea really makes sense
in a hard-core scientific context.

But one could also take for example, intuitive reasoning used by say
Albert Einstein when constructing GR. Even though such reasoning may not
conform with what the actual theory says. One example is the use of an
"inertial frame", one acting as though there is no gravity
present. Strictly speaking, no such inertial frame exists in typical
gravity environments, only good approximations of it.

I developed criteria of my own,
especially the one of matching complexities. I was very
good at mathematics in school, and solved many a problem
without calculating: I just "saw" the complexity of a task,
and plucked the solution in the corresponding (equally
complex) area of potential solutions.

So the problem arises between what the actual physical theory says, and
the approximations that humans use, attempting to understand it.

Also, I look at how
a problem is embedded. Let me explain this by means
of an example from my military service. Once we had to
pepare oruselves for a test in airplane recognition, we had
to study a booklet with pictures and details of airplanes
from the western hemisphere (Nato, USA) and of the
eastern hemisphere (USSR, Warsaw pact). I was still tired
from the exercize the night before, so I didn't learn, instead
I took a nap in the warm midday October sun in the Swiss
Alps (I can't tell you where, military secret, otherwise you
come and conquer Switzerland  ;-)  One hour later we were
called in, given large sheets covered with tiny photographs
of airplanes, and were asked to mark whether a plane was
from the western or eastern hemisphere. Easy peasy.
Western planes were sharp, eastern ones blurred. I solved
the test within five minutes, and was released for another
nap in the pleasant sun at a secret place in the Swiss Alps.
My comrades needed up to an hour for the test, and got
a lot wrong, while I got everything right. So I was marveled
at as a prodigy in plane recognition, and given the honor
of participating in an advanced course. Oh no! less spare
time! So I told our instructor how I did it, and he was baffled:
yes, it's true, the western planes are clear and sharp, the
eastern ones blurred ... So I helped them improve their
test (and I always said that Switzerland must above all be
defended with intelligence, gray matter is more important
than weapons).

So relating to some other discussions here: you had found a shortcut, a
rule, that worked in place of the examples. A good airplane recognizer
should of course be able to recognize not only the model of the plane,
but whether it is a new one or not, and if it is, be able to draw it. That
takes a bit more of training. The rules, like the ones you indicated, may
serve as shortcuts to pin down general structures. The structure becomes
similar to that of other learning, like a natural language grammar.

Now for non-locality. I find this a very complex and deeply
rooted phenomenon we have not yet really taken seriously.
The ideas brought forth in the article Black Hole Paradox,
New Scientist, 28 October 2006, are the first ones that
give me a feeling of matching complexity, so I keep them
in mind, even if I can't read the original papers and check
them out myself.

I do not know what you mean by non-locality in GR. I know some physicist
named Hawking or so had an idea of black holes communicating, but that
seemed to be in the fantasy sector of physical theories.

I don't know whether my babbling helps you further with
your math prover, ...

Not really, but if you want some feedback on your scientific thoughts, I
am willing to help. :-)

...but I can show you how I handle also
this question: I try to learn more about the complexity of
a problem, and how it is embedded. My personal view
is that parts of this procedure may perhaps one day be
implemented into computers. Top mathematicians got
the same and much more developed abilities: instead
of calculating they wander around in a landscape of
numbers, a marsh here, a hill over there, and just pluck
the solution - they know where to go in their landscape
and find the answer. They must have the same sense
of complexity, ...

When I was younger, I liked problems that took a week to solve. Provides a
good training.

...plus an amazing ability of visualization.

Though visualizing is important, I am somewhat sceptic at the idea that it
must be used, because I do not use it when understanding abstractions.

Von Neumann computers are very good at visualizing.

I am not sure what you mean here: The von Neumann model is just where
program and data shares memory. And the latest historical research claims
he was only the guy who write the paper - it was already invented and more
or less known when he did it.

Now combine them with neurons for pattern recognizing
and you get a hybrid computer that may achieve much
the same task. Yes, I see a future in hybrid computers:
Von Neumann computers for reliability and number
crunching, neurons for intuition and pattern recognizing.

This is a popular theme in SF, with various cyber-humans and androids.
Check out the "Borg" in "Star Trek".

This is in fact starting to happen, with humans getting brain implants to
control a computer or artificial limb or something.

But the idea of a standalone hybrid computer in your office, I think is
very far away.

If I remember correctly, Paul J. Kriha said here in this
thread that we can tell him how we solve a problem
in our mind, and he will write a software that does
the same. Well, here I tell how I solve problems,
but I doubt very much that it can be translated into
software for a Von Neumann machine, we must wait
for neuronal computing. My personal view, as I said
above.

I think this is the very point: we do not know very exactly how we, the
humans, solve problems. And because of that, it is not possible to put it
into computers.

Be also aware of that the idea of computer is to eliminate QM effects, and
the idea of a human brains is to cope with it. There are research on QM
computers, that would serve as an intermediate of todays binary computers
and the biologically neural computers you hope see coming into being.

--
Hans Aberg
.

Loading