Luis Rocha:
> Self-organization refers to
>a higher order reduction of variety (as the system converges to smaller
>volumes of the state-space), often manifested by the (higher-order)
>observation of recurring patterns.
This definition of self-organization is too beautiful for me not to point
out that it sounds exactly like my definition of selection: selection is
what reduces the variety created by variation.
>You (and
>Ashby) are of course free to use the term selection to refer to
>attractor behavior in dynamical systems. What I strongly disagree with
>is using this term in one sense, and then hijacking a whole edifice of
>evolutionary biology thought, which relies on an entirely different
>definition of selection (statistical bias on rates of reproduction), as
>if the two notions, just because both are known by the same term
>(selection), were equivalent.
It depends what you call "hijacking a whole edifice of evolutionary biology
thought". Some phenomena from evolutionary biology have a counterpart in
attractor behavior, others do not. This is again similar to our discussion
about the differences between cultural and biological evolution. At least,
using the notion of selection may help you to find similarities between the
two types of phenomena which otherwise might have escaped your attention.
As to your "entirely different definition of selection (statistical bias on
rates of reproduction)", the only thing that is different from the one I
use is that I put the "re" between brackets:
selection is a statistical bias on the rates of (re)production of a
particular configuration.
If you consider that systems or states (what I call configurations) can as
well be spontaneously *produced* as *reproduced*, then the presence of an
attractor can be viewed as a bias toward the production of states inside
the attractor.
To show that this definition is equivalent with the previous one (reduction
of variety) it suffices to express variety by a statistical entropy
measure. In that case, both self-organization and selection are processes
that decrease statistical entropy, while variation is the process that
increases entropy.
To show that putting production and reproduction together is not merely a
play with words, I offer the following example. Imagine a number of salt
molecules in a saturated salt solution. These molecules will tend to
arrange themselves in a crystalline configuration. The crystal
configuration functions like an "attractor". The variety of movements and
positions of the molecules in the solution is reduced while they are
getting fixed in the crystal. If the saturated solution evaporates or cools
down, this will happen spontaneously, although it may take a while before
the process starts. This is a clear case of self-organization.
Suppose now that I throw a tiny crystal into the solution. This will
strongly accelerate the process: the crystal acts as an autocatalytic
system, that boosts its own growth. The reason is that the crystal provides
a template that helps other salt molecules find stable positions. However,
I might as well say that the tiny fragment of crystal has started to
*reproduce*, creating more and more pieces of crystal. If I want to mimic
Darwinian selection, I could throw different fragments of crystal, with
perhaps slightly different configurations, into the solution, and see which
ones grow faster. Or I might observe how slightly different configurations
appear because of copying errors, and see which one grows fastest. Apart
from the fact that DNA consists of complementary strands, the template
mechanism used by the crystal is not essentially different from the way
e.g. an RNA molecule reproduces, where one strand acts as a template to
assemble free-flowing molecules until they form another strand.
Of course, this is still far from Luis's "description-based reproduction",
but I hope you get the gist of my argument: as Bruce Edmonds suggested, you
cannot draw an absolute, sharp boundary between biological reproduction and
physical or chemical self-organization. Otherwise, you would never be able
to explain the origin of life. That is why I don't want to make a strict
separation between a statistical bias on production (what you call
self-organization) and a statistical bias on reproduction (what you call
selection).
>I like to find commonalties as much as the next guy (I am a systems
>scientist after all), but we must restrain from this vernacular
>hijacking so that we may be taken seriously by those in the particular
>fields we wish to theorize generalize about.
The question is who has hijacked whose terms. In evolutionary biology, the
term selection has indeed the strong meaning of bias on *re*production,
but for example in evolutionary epistemology selection is basically a bias
on the production of knowledge (while in memetics, it is again
*re*production of knowledge that is most emphasized). Maybe the
evolutionary biologists are in the majority, but does that mean that we
should not be allowed to use some of the same terms with a somewhat more
general meaning? After all, the term "selection" has changed meaning many
times, even in biology, where it after all originated with Darwin's
observation of artificial selection.
Of course, we can simply agree to restrict the term "selection" to the more
traditional biological definition, and use "self-organization" in all other
cases. The problem I have with that is that selection is a much simpler,
clearer concept to describe what you called "reduction of variety", than
self-organization. (what is "organization", what is "self", how do does the
one create the other...?).
In my experience, many people use the word "self-organization" when they
actually mean "now something weird happens that I don't really understand".
That kind of usage is much more difficult to justify with the more
straightforward term "internal selection".
Moreover, by making such a strict separation, you lose the motivation to
look at all the intermediate cases, combinations and interactions between
"selection" and "self-organization" that are implied by a systems theory of
evolution. Your concept of selected self-organization is of course a step
in the right direction, but there are many more interactions at many,
different levels than the mechanism you call "selection of attractors". In
my approach, you would start by analysing all the relevant systems,
subsystems and supersystems, and for each consider which processes of
selection and internal, which external, and then try to understand how they
all work together. (not an easy thing to do, obviously!).
Anyway, you should soon be able to read my detailed analysis of
self-organization from a selectionistic point of view, since I have been
invited to write a long article on "The science of self-organization and
adaptivity" for the Encyclopedia of Life Support Systems.
_________________________________________________________________________
Francis Heylighen <fheyligh@vub.ac.be> -- Center "Leo Apostel"
Free University of Brussels, Krijgskundestr. 33, 1160 Brussels, Belgium
tel +32-2-6442677; fax +32-2-6440744; http://pespmc1.vub.ac.be/HEYL.html