Author: Robin Smith
Date: 21:56:37 03/20/01
Go up one level in this thread
On March 20, 2001 at 23:33:10, Christophe Theron wrote:
>On March 20, 2001 at 19:16:20, Robin Smith wrote:
>
>>On March 20, 2001 at 01:50:10, Christophe Theron wrote:
>>
>>>That's it.
>>>
>>>Let me give another example, for which I have fought here for quite some time.
>>>
>>>If a player A is better than player B at fast time controls, I just begin to
>>>assume that player A will be better than player B at longer time controls.
>>>
>>>From that point, if I notice that it might not always be the case, I might work
>>>to provide evidence that it is not the case.
>>>
>>>But my first idea is to keep things simple and see if it works. So I would first
>>>assume that time controls do not matter. I change my mind and add complexity to
>>>my model only if I can prove that my "simpler" model does not work.
>>>
>>>I do that because I noticed a long time ago that general concepts are much more
>>>powerful. A general concept (or "idea" or "principle") is one you can apply in a
>>>large number of cases. So I try to keep my ideas as general as possible. It
>>>hurts me when I have to add special cases to an otherwise "clean" (simple)
>>>model.
>>
>>Christophe, I believe your assumption here is generally correct. The principle
>>even has a name, "Occams razor" named after a philosopher and theologian,
>>William of Occam, who lived ~700 hunderd years ago.
>>
>>But I do have good evidence that in at least one case, Fritz5.32, it is not. I
>>am convinced Fritz5 is much weaker than comparable programs when playing at
>>super long time controls. My guess is this is probably due to heavy use of root
>>processing.
>
>
>It might be, but you understand that my point is not to discuss if it is or if
>it is not.
OK
>> But starting first with the simplest assumptions makes excellent
>>sense.
>>
>>>I do not know what to do with the concept of "conscience". I don't need it to
>>>cover a hole in the big picture of "intelligence", and it explains nothing
>>>anyway. Worse: those who talk about it find it mysterious and impossible to
>>>explain.
>>
>>Correct. It is similar to trying to explain vision to a person blind since
>>birth, you can explain the mechanics, but not what it is like, on a subjective
>>level, to be able to see a beautiful flower. Unless you share the experience,
>>just explaining the mechanics doesn't explain everything.
>
>
>Let me try something about "conscience". I'm trying hard to understand what you
>mean, I don't want to look too stubborn. :)
>
>Could it be that we call "conscience" the part of the information processing
>entity (notice how I avoid the term intelligent entity) which is monitoring the
>rest of the entity?
>
>If I understand correctly the word "conscience", being conscient is to be able
>to "realize" that I am thinking about something, or that I am saying this or
>that. Or that I am currently looking at a flower. True?
>
>In this case, "I" is the part of the entity which is not conscient (or less
>conscient?), and the "conscience" is the part that collects only the higher
>level informations. The conscience continually monitors the state of "I".
>
>All the process of converting the signal received by a light sensor (eye,
>camera...) and coming to the conclusion that a flower is in front of the light
>sensor is out of the reach of "conscience". "I" do it, but conscience is not
>involved at that stage.
>
>When the signal is finally converted into the symbol "flower", the "flower"
>connection is activated, which in turn is going to activate a number of related
>concepts. Like a control panel where suddenly some lights turn on. The
>conscience only sees this controls panel, and the entity then *thinks* "I see a
>flower".
>
>BTW this means there would also be two kinds of "memory". What happens can be
>memorized in an "unconscient" memory (maybe a kind of memory for low level
>events), and also in the conscience, as a sequence of high level events.
>
>Notice that by this "definition", an entity can "lose consciousness" without
>being totally unable to respond to external signals. What is going to happen is
>that the entity will respond only with its lower level processing abilities, and
>will not "record" the sequence of events ("I" responding without being
>conscient).
>
>Wow, sounds terribly close of what can happen to a human being.
>
>I guess my reasonning sounds very naive to people who have studied psychology. I
>feel like I am re-inventing the wheel. :)
>
>Anyway, if this model of "conscience" makes some sense, then I don't see any
>reason why it could not be implemented in transistors rather than in neurons...
>It is possible, because this definition of conscience does not mention the level
>of complexity of the symbolic processing unit which is the conscience.
>
>Notice BTW that by this definition, a high level processing unit (or a symbolic
>processing unit) can be called a "conscience" only when it is connected to the
>"real world" thru a set of interfaces directly connected to this world. Here I
>want to emphasize the fact that -maybe- a conscience cannot exist without a
>world to which it is connected.
>
>OK, now I think I have gone too far. :)
The hypothesis you are describing is very close to one I read in a Scientific
American article. I think it was from 1995 or thereabouts. And I think the
hypothesis has a very good chance of being essentially correct, though like many
good hypotheses it not only answers a question but creates new questions as
well. One nice aspect is that it addresses the issue of why a sleeping person
(and sleeping brains are still VERY active) is in what is called an unconcious
state ..... the sleeping brain is probably *not* engaged in trying to relate "I"
to "the world". Now if only someone could devise an experiment to test this
hypothesis I would be happy.
>>> So in short it raises questions you cannot answer, or that you do not
>>>need to answer! I fail to see how it helps me to understand anything about
>>>"intelligence" (a concept supposed to be closely related).
>>>
>>>So I just drop it for now...
>>
>>Let me know if you want to take it up again later.
>
>
>I hope I have not been too long. :)
Not at all. But now we should probably let CCC get back to comp chess. Feel
free to e-mail me if you wish to discuss it further.
Robin
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.