Date: Tue, 11 May 1999 11:43:50 -0400
From: Guy A Lukes <guy.a.lukes@frb.gov>
Subject: Re: Darwin and Lamarck
To: memetics@mmu.ac.uk
--0__=VNDrpRBk7aigGXlhOpazOWm3UhNhNfuHtwnnta64KZQNf5ktGZM08f9Y
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
Ton
Thanks for responding. Your comments have been very helpful. I have also
been taking a closer look at what Bateson sayed in "Mind and Nature".
>>> to emphasize the fact that Gregory Bateson has been "defending" Lamarck
>>> quite elegantly for many years.
I doesn
--0__=VNDrpRBk7aigGXlhOpazOWm3UhNhNfuHtwnnta64KZQNf5ktGZM08f9Y
Content-type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-transfer-encoding: quoted-printable
?t think Bateson actually defends Lamarck does he?
>>>His position was that Lamarck's error was one of logical typing:
>>The logical typing that Batson seems to be talking about is the
difference
>>between a structural bias (thermostat setting) and the adaptive
>>change to environmental perturbation (turning on the heater).
On further reflection, it seems the real issue is a mixing of inappropr=
iate
time scales.
> [what] Bateson has undertaken in his book "Mind &
> Nature; A Necessary Unity", is to draw a formal parallel between the
> process of natural evolution and the process of learning.
Therefore if the dynamic state of a homeostatic adjustment (learning) w=
as
transferred to the relatively static structure of the underlying genome=
based instructions (evolution), the homeostatic process would be loosin=
g a
degree of dynamic freedom, required to maintain its equilibrium in rela=
tion
to a dynamic environment.
>He does, however, emphasize at least one fundamental difference
>between the two: biological evolution is different from learning in
>that it contains rather stringent conserving mechanisms such as the
>Weissmannian barrier, preventing acquired characteristics from enterin=
g an
>individual's genes, thereby slowing down the process of adaptation
considerably.
Evolution is slower responding to long term statistical changes
>Learning, on the other hand, doesn't feature such
>inherent limitations on adaptation, and therefore is far more prone to=
>"unguided" creativity
Learning is faster responding to immediate environmental shocks, but is=
grounded on slower moving bias settings determined by statistical
evolutionary processes.
>One logical level up from the individual organism is the species, or
rather
>(when we speak of heredity), the gene pool of that species.
>The main limitation on the adaptability of a species is the amount and=
the
location
>of adaptive flexibility - or conversely, stress - in its variables.
>You can't maximize certain variables without "eating up" the flexibili=
ty
of the total system.
For example, suppose you create an artifical animal "memal" that has sk=
in
pigments that allow it to be white in the winter (in a snowy environmen=
t)
and brown in the summer (in a brown soil environment), so it is less
visible to predators. One Lamarkian design would transfer the state of=
the
parent to the genetically controlled bias of the child. Therefore
offspring in the winter would be white and have difficulty hiding in th=
e
summer and offspring in the summer would be predominately brown and hav=
e
difficulty hiding in the snow.
The bias setting for pigment needs a longer term memory for determining=
genetic changes. This memory is provided by the statistics of the gene=
pool.
If evolution is to inductively accumulate information over time, long t=
erm
changes in the environment must ultimately be incorporated into the
structure of the system (bias settings). In order to maintain stabili=
ty,
this process must take place at a time scale much slower than homeostat=
ic
changes. Therefore, evolution can not be directed by specific instance=
s of
phenotypic adaptation. It does seem possible, however, that an organis=
m
might be able to adapt the bias settings of its offspring, based on lon=
g
running statistics which have been accumulated over a period of time. =
With
symbolic communication, that period of time could, hypothetically, even=
be
longer than the life of an individual agent.
Memes could build in some kind of longer term source of information to
control genetic bias settings other than population statistics. We wou=
ld
however, have to change long term statistics about the dynamic environm=
ent
(state information) into structure (bias settings) crossing a logical t=
ype,
and translate between two different time scales.
Memetics may need to think more about the links between different level=
s of
autonomous control, instead of focusing on just gene like frequencies. =
For
example, the resent debates on the mathematics of a nonexistent gene fo=
r
homosexuality.
Guy
=
--0__=VNDrpRBk7aigGXlhOpazOWm3UhNhNfuHtwnnta64KZQNf5ktGZM08f9Y--
===============================================================
This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)
see: http://www.cpm.mmu.ac.uk/jom-emit