Re: [pcp-discuss:] Fwd: "Intelligent Design"

From: Cliff Joslyn (
Date: Wed Jun 07 2000 - 18:57:39 BST

  • Next message: Cliff Joslyn: "[pcp-discuss:] New member: Graham Whyte"

    >Yes, we ascribe goals to events, but there is also a method to test if our
    >understanding of system's goal corresponds to its actual internal goal
    >(which is not observable directly). First, this goal should be beneficial for
    >the system, i.e., it should increase its chances to survive and reproduce.
    >Your ball example does not fit this criterion.

    In PCP we try to cast this in terms of a general "survival" requirement,
    based on a general concept of "stability". Thus survival is one kind of
    stability, and reproduction one way to achieve survival. If you insist
    strictly that goal-following requires an increase in REPRODUCTIVE fitness,
    then you cut off a lot of behaviors which enhance other kinds of stability,
    or in other ways contribute to survival of non-reproducing systems. You
    also limit yourself to the domain of living systems, but that's OK with me.

    >Second, a system should be
    >able to modify its behavior in order to reach the goal. If we put a ball in a
    >small hole on the slope, will it be able to jump out and continue its fall?

    The ability to deploy variable means to constant ends is a standard way to
    describe goal-following systems. Considering these variable means as
    trajectories, then if they are all not dynamically equivalent and following
    least-action, the implication is that there is some kind of control process
    which violates the dynamical laws (e.g. the ball "trying" to roll uphill).
    The problem is that determining whether this is true or not is
    model-dependent, in that we have to believe we know how the system WOULD or
    SHOULD behave if it were NOT goal-following, and then see that it violates
    this. This is what Bill Powers calls The Test for the presence of control.

    >The idea that enzymes have goals does not lead to a pan-semiotic view.
    >Each enzyme is an element of a self-reproducing system that includes the
    >gene that codes this enzyme. Enzyme functions (goals) increase the rate of
    >self-reproduction within a community of other genes in a cell. Many enzymes
    >are able to change their function in response to environmental changes, and
    >these changes are often adaptive. Enzymes have several discrete conformation
    >states that can be viewed as a primitive "code".

    As Luis pointed out, enzymatic action occurs within a whole system of a
    genetic organism, which has a symbolic component (the genome). Following my
    prior comment, enzymatic action necessarily shows all three forms of
    system: the dynamics of the rate-behavior of the catalysis, the
    combinatorial specificity of the chemical form, and its role within the
    whole organism to achieve control.

    The distinction between symbolic and non-symbolic semiotic systems is
    crucial, and roughly follows the pan-semiotic vs. "non-pan"-semiotic
    (whatever that is) debate, in that those who assert the presence of
    semiotic processes below the organismal level invariably mean non-symbolic
    sign systems. I am open to the idea that this distinction (symbolic vs.
    non-symbolic) might be more parsimonious than what I normally use (semiotic
    vs. non-semiotic), but it's a very deep issue I haven't though enough about.

    | Cliff Joslyn, Member of the Technical Staff (Cybernetician at Large)
    | Distributed Knowledge Systems Team, Computer Research Group (CIC-3)
    | Los Alamos National Laboratory, Mail Stop B265, Los Alamos NM 87545 USA
    |     (505) 667-9096
    V All the world is biscuit-shaped. . .

    ======================================== Posting to from Cliff Joslyn <>

    This archive was generated by hypermail 2b29 : Wed Jun 07 2000 - 20:06:50 BST