Re: natural intelligence

Alpha <omegazero2003@xxxxxxxxx> wrote:
On Feb 7, 11:10=A0pm, c...@xxxxxxxx (Curt Welch) wrote:


You did have some insightful comments above!!

No one has a clue how they play chess. =A0If they tell you they do,
they =
are
an idiot. =A0We look at the board, and "ideas" pop into our hand and
befo=
re
long we move a chess piece. =A0If we try to recall what happened, we
migh=
t be
able to roughly describe the sequence of idea that popped into our
head, but we would have no ability to explain why those ideas popped
into our head at that time, and other ideas didn't.

Oh no Curt - you are really off base here. I am an expert chess
player - been playing since I was 6 years old! Won championships in
high school. And I will tell you right now that I memorized hundreds
of game sequences and scenarios and thence strategies depending on
board positions which evolve over time. One thence enters a prediction
mode and runs throough scenarios likely to evolve and considers
offenses and defenses for anticipated movement scnarios And applying
heuristics all along the way - lots of them.

I'm a rank beginner at best at chess, so certainly what I think about while
playing a game will be a world of difference than someone with your skill
level. While I think things like, "can he take me if I move there?", you
instinctively know the answer to questions like that without really
thinking about it and your mind is spent thinking of what's likely to
evolve many moves down the road. Even though my thought process for chess
is very naive to anyone well trained in the game, I still do know basically
the _type_ of things I was thinking about as I was exploring different
moves.

But this is not the point I was trying to make. We spend so much time
thinking about the things we do understand, we never stop to realize how
much we don't understand.

Not so much about the game, but about why we are having the thoughts about
the game we are having (while playing), and why we aren't having different
thoughts.

Sure, the simple answer is that we have been trained by experience to think
about the things we think about. BUt what we have little understanding of,
us excactly how that works, and why the different things actually do show
up in our thoughts, when they do, and in the sequence they show up. We
pick a move (or path) in the game to explore and we spend some time
exploring options, and at some point stop, and start to explore another
path. By why did we stop at that point, and not spend 5 seconds longer
thinking about it? Why didn't we stop 5 seconds sooner? Whey didn't we
explore a given sub-path down that part of the game tree? These are the
things we have no clue about. Though we understand the types of thoughts
going though our head, and we understand why there are chess game thoughts
flowing through our head when we are playing a game (instead of thinking
about how to grow palm trees for example), but we have no real
understanding of all the fine details of how and why the thoughts happen as
they do. Why did we explore move X first instead of move Y for example?

Though it may feel like we have a good understanding of what and why is
happening in our head, we really don't much about it at all. What we know,
is historically what did happen, becuase we witnessed it happening.

Mostly, what we do, is just talk about what did happen (to our ability to
remember it - which isn't all that good), rather than having any real
understanding of why it happened.

There's an old psychology experiment done I think in something like the
20's or 30's that showed how the environment can trigger people's behavior
in ways they had no awareness of. I learned about this experiment in past
discussions in this group. Subjects were put into a lab and asked to solve
a problem of trying to tie to hanging strings together. They were too far
apart to make this easy. The only way to do it, with the stuff laying
around the lab, was to tie a weight (a Bunsen burner I believe was the only
weight that would work in the room), to one of the ropes, and start it
swinging. You could then grab the other string and then reach the swinging
weight, and tie the strings together.

This was a hard enough problem, that most people couldn't solve it on their
own - at least not very quickly. The experimenter would walk around the
room and observe. And after a while, he would give the subjects a
sub-conscious hint by staring one of the strings swinging. Almost without
fail, this "hint" would trigger the subjects to "see" the solution.

However, also almost without fail, when asked how they solved the problem,
they never realized they were given the hint. They said things like "It
just came to me", "or thought about how to reach the string and thought
about swinging it". No one said "I saw you swing it and that made me think
of the answer". Even more interesting, some said the string had not been
moved by the experimenter.

This result was repeated on many subjects. The bottom line is simple.
Ideas spring into our head, and we have no clue why most the time. Even
when there's something obvious happening around us that triggers the clue
to pop into our head, we still can, at times, not have a clue it happened,
or that it helped motivate our thoughts as it did.

Certainly, we have awareness that things like this happen to us - that
things around us trigger us to have various thoughts and sometimes we are
very aware it has happened to us.

But more important, is the simple idea that what we do next, is not
triggered by just one or two big things in our environment. Far more
likely, is that the brain is weighing millions of factors in parallel to
make it's "choice" about what to think about next. The time of day, the
shoes we are wearing, how bright the room are, how many people are around
us, how high are the ceilings, what's pattern of tiles on the floor, all
these things can play some odd subtle roles in whether we chose the apple
pie, or cherry pie - or chose not to have any desert at all.

The brain is basically a huge and complex parallel evaluation function
which is picking what we do next - for everything we do, and everything we
think about.

If anyone things they "understand" "why" they had the thought they did,
then they don't understand the nature of the beast at work here.

What we do all the time, is to talk as if we do understand these things,
becuase believing we don't understand why we do things, makes us feel
uncomfortable. So we create as much understanding as we can, make up the
rest, and then talk ourselves into believing we "understand" when in fact,
we don't have a clue. What we do understand about the complexity at work
in the brain that went into making a decision, is only an insignificant
fraction of the complexity that actually was at work making it happen.

Deep Blue and such contain this sort of vast amount of finely tuned
info about *how* chess is played.

I don't know much about how these high end chess programs are written, but
I've been left with the impression that their main technique remains brute
force - they just search as much of the game tree as they can afford to and
pick the best option they find. The number one main power of the computer
is it's search speed. They improve that by adding dictionaries of opening,
and special heuristics for end games, as well as lots of fine tuning on the
heuristics it uses to prune the search tree.

But it's also well known that humans don't work anything like that. Humans
have a tree search systems that is many orders of magnitude slower than
even a single processor chess program. Where humans think about a few
hundred game positions in the tree, computers cover billions in the same
time. What they do have however, is a far more advanced board evaluation
function. A human can look at a board position and very quickly eliminate
most the moves as worthless and focus on the 2 or 3 that have merit.

These very advanced board evaluation functions is what we have never
duplicated in our chess programs. We just make up for weak evaluation
functions by adding what is simple to add - more game tree search speed.

We can build these advanced evaluation functions into the computers becuase
for 1) chess players have no idea how they do it so they are of no hope
explaining to the programmer why a given position was ignored (other than
saying "it was a bad move"), and 2) we haven't figured out how to build
learning systems that can create a board evaluation function as strong as
the human brain can.

And the odds of anyone
correctly recalling what they thought of for the past 60 seconds while
they were looking at the board I'm sure is near zero.

That is simply not true.

Well, you might be right there. Some people are better at memorizing facts
than others. I don't memorize facts very well. I tend to ignore the
details and focus on the abstractions. Some people tend to be blind to the
abstractions, but highly aware of the details. I would generally say you
are prime example of someone that is much more attune to details than I am.
That might in fact allow you, and others like you, to do a very good job of
remember details of your thought process for the past 60 seconds when I
could not.

However, no matter how good you are, we all have limits. If we were to
record you having a conversation with someone, and stopped the tape. How
much of what just was said, could you correctly recall word for word? I
sure the hell wouldn't expect that I could go back more than one or two
sentences and get them right word for word. I would actually be surprised
if I got even the right sentence correct word for word. I'd consider it a
miracle if I could even recall all the subjects we talked about for the
past 5 minutes, let alone what was said word for word.

Maybe you are much better at a test like that than I am, but there would be
a time limit where you brain would also fail to be able to recall what was
said.

If we can't pass a test like that, correctly telling someone what was said,
I don't believe we would be any more accurate trying to document what
thoughts went though our head in the same sort of time period.

Humans of course tend not to remembering what they have forgotten [:)], and
as such, can fool themselves into believing they have forgotten nothing
when in fact they had forgotten half of what happened. Unless you have
something like that video tape to show you what you got wrong, it's hard to
know how much we are forgetting. So even though someone may thing they
have "good memory" if you can't test them, then odds are, it's not as good
as they think it is.

Of course we are very
good at rationalizing our behavior, so when someone asks, "why didn't
you consider that move", we would just say, "oh, my gut told me it
wasn't important".

And that is merely a synopsis of a boatload of heuristics that run
through the brain/mind for a given scenario. Most masters and
grandmasters can certainly delineate the various thought sequences
that went through their mind *in detail*!!!!!

Certainly, when you deal with a subject you are very familiar with, you can
remember details far better than when deal with subjects not so familiar to
us. Chess players are really good at looking at a chess board with only a
blink of an eye and being able to reproduce all the piece positions
accurately. Non chess players are lucky if they get 3 out of 20 pieces in
the right place after studying it many seconds. :) Likewise, you would
expect a grand master to have very detailed memories of the game, and what
was being considered while they played it. But like everything, there
would be limits to what they can remember correctly. Maybe the length of
one game is so short and so easy to them that it would be as trivial for
them to remember the entire game and everything they thought about as it
would be for me to remember a 3 digit number for 10 seconds. But we all
have limits and at some point, we won't be able to remember details of our
past, or our past thoughts with much accuracy at all. Whether in a chess
game that's 60 seconds (as it would be for me), or 60 hours of chess (as it
might be for a grand master), there is a limit where we simply can't
remember correctly what we did.

I know I could at that time (I am still very good but not as good as
I was in my teens and early 20's)

=A0When in fact, all we really know is that for some reason, the
idea to look closer at that possibility never entered our head so we
just assume it didn't enter our thoughts for some "good" reasons een
though we don't have a clue what that reason was.

The fact that a string of discrete behaviors emerge from us which we
can call "symbols" at times doesn't mean in the lest that we understand
why they emerge, at the time they did emerge.


The why is problematic until we get a mapping from level of
description to level of description. From QM (or some other ultimate
final theory of everything) through to the psychological/cognitive
levels of existence.

Well, I think the basics are obvious here. The brain receives a lot of
data from all it's sensors, produces a very complex mapping to changes in
it's internal state, and produces behaviors based on some highly complex
function implemented by millions of interconnected neurons. These mapping
functions that are responsible for selecting what we do next at each
instant in our life are way too complex for any human to directly
understand. A million neurons and 100 million synapse connections are
likely to have played a role in each little thing I do in my life - each
little finger move, and each word we produce next in our thoughts or each
part of the game tree we explore while playing chess. And though there are
some obvious high level cause and effect events at work, there are no
simple cause and effect events at work (as there are in our computers for
example). Each little neuron firing in the brain produces has a huge and
complex set of factors that were weighed by the brain in making so much as
a single neuron fire, or not fire. There are no simple answers to such
complexity - and if a chess player thinks he understands "simple answers"
to why they made the move they made, they don't understand the nature of
the great complexity at work in in a human brain making those decisions.
What we do understand about how we play chess, is a drop in the bucket
compared to what we don't understand.

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.

Relevant Pages

  • Re: natural intelligence
    ... into our head at that time, ... I'm a rank beginner at best at chess, so certainly what I think about while ... playing a game will be a world of difference than someone with your skill ... explore a given sub-path down that part of the game tree? ...
    (comp.ai.philosophy)
  • Re: Z22 vs. Gameboy DS
    ... I like the old 7-Up spot game. ... > You can also learn chess and get a chess program for the palm. ... > exercise for the brain; first time I played chess after a long hiatus I ...
    (comp.sys.palmtops.pilot)
  • Re: Games & Expectations
    ... might concede that it's really a puzzle, not a game per se. ... been comparing chess with "abstract Risk" (i.e., ... game as chess). ... much like a battle as Risk looks like a war). ...
    (rec.games.board)
  • Checkmate
    ... Life is a game of Chess ... The King stays safely behind the Castle gates ... Until and unless the Democratic Party makes a major course correction that includes renouncing the craven political chess game they play in which gays are pawns - they will never be a viabl... ... Amazon.com: Pawn Structure Chess: Books: Andrew Soltis ...
    (alt.politics)
  • Re: Is Disciples II really worthwhile?
    ... Databases didn't really benefit chess as an intellectual exercise, ... though my problem with that game is that it's very ... am still at a level where I improve simply by playing ... Anything that can bring a little more joy into one's life helps make ...
    (comp.sys.ibm.pc.games.strategic)
Loading