Re: Do we have proofs that human think in a fuzzy (logic) way?

From: Marco Mililotti (mmar@studenti.ing.uniroma1.it)
Date: Thu Jan 18 2001 - 07:55:50 MET

  • Next message: Ulrich Bodenhofer: "Re: Fuzzy logic and Goedel's Theorems"

    On Tue, 16 Jan 2001, Gordon D. Pusch wrote:
    Dear Gordon,
     I'm not an expert on the subject. I'm interesting in the relations
    between FL and human way of thinking, and the way we can emulate it. So:

    > It does not follow that because humans "deal with imprecision and uncertainty"
    > they must therefore be using the specific mathematical formalism called
    > ``fuzzy logic'' to do so, or that machines using this mathematical formalism
    > will ``think'' exactly the same way humans think.
    It's right for me. My vision is that FL is the only one that is closest
    enought to the way we think. But I'm not an expert in the others kind of
    logic, too (Bayesian in primis). First order logic is however too
    "simple".

    > Many different forms of
    > approximate, qualitative, and probabilistic reasoning methods have been
    > proposed in the literature; there is no evidence that the human reasoning
    > process uses _ANY_ of them.
    We can focus on which one emulate better than others the way we think, I
    think :)

    > Moreover, Cox's Theorem [Cox, R.T.: ``Probability, frequency, and
    > reasonable expectation,'' Am. J. Phys. v.14, pp.1--13 (1946)] proves
    > conclusively that =ALL= possible approximate reasoning methods other than
    > Bayesian Probability Theory fail to be logically self-consistent in some
    > way or other --- and that includes fuzzy logic.
    Again it's only my thought, but I don't think "human logic" is
    self-consistent. So why should we use self-consistent logic to reproduce
    it? Maybe, to be self-consistent is a "constrain" to hard if we want to do
    logics thougts the way human beings do.

    > Hence, if any method of
    > approximate reasoning is to be preferred, it would have to be Bayesian
    > reasoning. (This is =NOT= to say that _humans_ use Bayesian reasoning,
    > but only that _if_ we wish our machines to think in a way compatible with
    > logic, Bayesian reasoning has been mathematically proved to be the =ONLY=
    > approximate reasoning method that is capable of doing so. Any other method
    > --- including Fuzzy Logic --- will necessarily yield logical paradoxes
    > and inconsistencies under certain circumstances.)
    Exact thought is not a human quality. As a human being I can cope with
    paradoxes that I find through my life because of my kind of
    logic. Excluding paradoxes (logical paradoxes) is not a way to reproduce
    human kind of logic. But the /possibility/ of managing such paradoxes is
    important I think for human-like logic. Bart Kosko has written: "If human
    thought has a logic, in the best of hypothesis it's a fuzzy logic" (or
    something similar).

    And, just to know interested thoughts, what do you think about the Fuzzy
    Associative Memory to reproduce human-like patterns of thought? It seems,
    to me, very powerful as it enables us to "Compute With Words" and link a
    sequence of computation to reproduce a (fuzzy) logic sequence of
    thoughts. I remember that the way human brain processes information is
    serially ("... it processes information serially, even though the
    underlying neural hardware is operating in parallel..." -> University of
    Iowa professor Steven Luck). So I think that a Fuzzy Activation Pattern
    in a FAM could in a way or another represent a human thought. What do you
    think? Any reference available?

    Thank you for your attention,
    bye,

    -- 
    Strange game.
    The only winning move 
    is not to play.
    What about a nice play of chess ?
    -- 
    Marco Mililotti
    

    ############################################################################ This message was posted through the fuzzy mailing list. (1) To subscribe to this mailing list, send a message body of "SUB FUZZY-MAIL myFirstName mySurname" to listproc@dbai.tuwien.ac.at (2) To unsubscribe from this mailing list, send a message body of "UNSUB FUZZY-MAIL" or "UNSUB FUZZY-MAIL yoursubscription@email.address.com" to listproc@dbai.tuwien.ac.at (3) To reach the human who maintains the list, send mail to fuzzy-owner@dbai.tuwien.ac.at (4) WWW access and other information on Fuzzy Sets and Logic see http://www.dbai.tuwien.ac.at/ftp/mlowner/fuzzy-mail.info (5) WWW archive: http://www.dbai.tuwien.ac.at/marchives/fuzzy-mail/index.html



    This archive was generated by hypermail 2b30 : Thu Jan 18 2001 - 08:05:59 MET