A Metaphor for Evolution, and more

From: Chris Cogan (ccogan@telepath.com)
Date: Tue Sep 19 2000 - 14:24:44 EDT

  • Next message: Brian D Harper: "Re: Schutzenberger's Folly, part 1"

    Imagine that we have a grass species that lives on a plain. When one of
    these grass plants produces seeds, they fall within a quite small circle
    around the parent grass plant. Each individual grass plant lives its life
    at or near the spot at which it began life, and then, if it lives long
    enough, it reproduces, and it's offspring live either at the same spot or
    quite nearby. Imagine also that not all parts of the plain are habitable
    for this grass, but that there is no way beforehand for a grass plant to
    know which are and which are not habitable, so all seeds simply end up in
    some location and then are either allowed to reproduce or it is
    mysteriously killed off. Thus, there might be a large circular area where
    there are no living grass plants because all the ones that set up shop too
    close to it (in a band around it equal in width to the distance the seeds
    are spread from the parent grass plants) die without having any offspring
    that might try to live inside the circle.

    Now imagine that the plain has many areas where our grass cannot live, but
    that, from their starting point many generations ago, they have still been
    able to spread out widely over the plain, even though the areas where they
    can't live have sometimes shifted, killing off large populations in some
    cases, and sometimes opening new areas of safe living, into which
    successive generations of the grass landed and grew, spreading a little at
    a time over the now-habitable area. From an airplane, we'd see, in several
    directions around the point where the original grass plant started many
    generations ago, a sprawl of areas of grass and of areas where no grass
    grew. In a speeded up "movie" taken from above, over many generations, we'd
    see both expansion at the edges and some shifting where areas that had once
    been habitable became inhabitable and where new areas of habitability
    opened up.

    And just inside the edges of many of the *non*habitable areas that had been
    around for any time, there would typically be a band of dead grass or dead
    grass seeds, where grass plants just *outside* the area had dropped some of
    their seeds just inside the boundary.

    This whole setup is a kind of metaphor for my view of evolution, in which
    each location on the plain represents a different genome, and these genomes
    may find themselves inside an area where they can live, or inside (for a
    while) an area where they can't live. At the boundaries of the areas where
    they can't live, there are often genomes that will keep "trying" to have
    offspring that *can* live inside such areas. This will appear as a band of
    "fuzz" on a map, as genomes appear inside the area and then are removed.

    Of course, the genome, in this case, has only two genes, one for each
    dimension of the plain, but it can have many different alleles, as long as
    the new alleles are only slightly different from the parent's alleles of
    the corresponding genes.

    Thus, because of the two-dimensionality of the genetic specification, we
    cannot get what we would normally consider to be a high degree of
    complexity in the genomes.

    Or can we? In principle, we can still get some interesting results. For
    example, one allele might specify a location equal to 3.14159 (pi) inches
    south from the point where the first genome appeared on the plain, and the
    other might specify a location 2.71828 (e) inches west from the "founding"
    seed.

    Well, okay, but you get the point.

    One nice feature of this metaphor is that it can be translated into a
    simple computer program that draws dots on a plane (instead of a plain),
    each dot representing a particular pairing of different alleles for the two
    genes. By varying both the way the uninhabitable areas are determined, and
    their changes over time, I think we'd have a fairly good way of presenting
    the basic idea of naturalistic evolution to novices, and perhaps even a way
    to do some useful research.

    Obviously, the technique could be expanded to three dimensions, though
    displaying it on a computer screen would be more difficult. Additional
    "dimensions," such as color and intensity could also be added without
    making the result non-visualizable. We could also introduce some randomness
    so that *some* "seeds" dropped even in habitable areas would be killed off
    and occasionally a seed dropped inside a non-habitable area would survive
    for a while.

    Interestingly, depending on locations and shapes of habitable and
    nonhabitable spaces, we could get speciation, where the dots spread into an
    area and then became cut off from their ancestors as a nonhabitable area
    expanded between them and their ancestor dots. In fact, good sized
    populations of dots could become completely cut off from their ancestor
    dots, with no longer any way by which the ancestor dots could gradually
    shift "genetically" to become members of the new population because the
    nonhabitable area would be too wide for the allowed range of
    single-generation variations to cross.

    Further, in principle, we could add completely non-visible "genes" to the
    "genome" of each dot, so the dots could *do* things, such as eat other
    dots, or take resources away from nearby dots, or even from dots of other
    "species" (remember, the dots only represent location in an abstract
    "genetic space"; two dots *widely* separated in genetic space might live
    adjacent to each other as two completely different *species* in the
    corresponding real space). By adding, or by allowing the "genes" themselves
    to add, new "genetic dimensions," I suspect we could get some *very*
    interesting results. Selection criteria could be as complex as we like, or
    they could depend on the values of multiple "genes," so that, for example,
    the mathematical product of two genes might partially determine whether a
    dot lives or dies, etc. Since the real world is vastly more complex than we
    would be able to make the computer environment with most modern technology,
    it is unlikely that we could impose selection criteria any more complex
    than those of the real world, but by making them as complex as we wished,
    we could watch how the genomes evolve to take advantage of new open
    "spaces" and how existing ones died off as such complex criteria changed.

    Anyway, if we allowed enough ultimate flexibility in the "genes," we should
    see what we see in the real world, but in an abstract way: Each organism's
    genome reflects the evolutionary history of the organism and its past
    environments. Thus, if the environment does not allow a certain pairing of
    two alleles from two sites, but does allow others, we will see this
    reflected in the statistics of the existing genes (there will be only
    occasional active pairings of the "forbidden" genes from genetically nearby
    "permissible" gene pairings). Allowing for this statistical "fuzz" at the
    boundaries, each surviving genome will contain a significant amount of
    information about the "organism's" evolutionary history. If the selection
    criteria remain constant for a long enough time, each genome will be a very
    good, but accidental, reflection of its total current environment as well,
    because, in this case, the past environment and the current environment
    will be the same.

    If there are available niches of increasing complexity allowed by the
    selection criteria, then any genetically adjacent "organisms" will start
    spreading (genetically) into these niches (I'm so far assuming nonsexual
    reproduction). That is, as long as there is a "pathway" of ecological
    niches from simple to complex, some of the organisms will continue to
    evolve along this pathway into greater and greater complexity. As long as
    there is benefit in increasing complexity, complexity *will* increase.
    Further, at no matter *how* we set the specification of complexity, if
    there is a viable pathway of small genetic steps to get to it, some
    organisms *will* get to it in time. This includes, of course,
    Schutzenberger's "functional complexity," Dembski's "specified complexity,"
    and Behe's "irreducible complexity."

    They will get there *if* there is a good pathway and if the next small step
    along the pathway is open (i.e., habitable to the organism) at times when
    that small step occurs genetically. The pathway might not even exist
    initially; it only needs to open a little at a time, and even then, it
    might do so in fits and starts. Put another way, if an area of habitability
    on the hypothetical plane moves slowly enough that it can always have a
    genetic population within it, then no matter *where* on the plane it
    eventually ends up, descendants of that population will be *in* that area,
    no matter what genetic complexity it represents. Thus, if, in space and
    time, there has always been, for 3.8 billion years, at least one
    continuously-existing habitable area that has been on the way from the
    earliest life to the complexity of today's humans, and if it never moves so
    fast that the population within it is killed off, then that population will
    eventually reach a human level of complexity. "Irreducible" complexity
    might take longer because it might have to be reached indirectly, as the
    zone of habitability moved (or spread).

    But, *unless* there is something *prevent* this, such as extreme bias
    against complexity in the variation process or very severe restrictions on
    populations (and thus on the number of variations that can be tried), it
    *must* eventually reach a great many niches, including niches "for"
    extremely complex genomes. They won't, in general, be genetically like our
    genomes, but they can be ecologically very much like ours, with the same
    degree of "specified" complexity, and the same general levels of
    "irreducible" and functional complexity in both genome and resulting organism.



    This archive was generated by hypermail 2b29 : Tue Sep 19 2000 - 14:29:57 EDT