Re: James Mark Baldwin: "A New Factor in Evolution"

Bobby D. Bryant wrote:
> On Thu, 17 Nov 2005, r norman <NotMyRealEmail@xxxxxxxxxxxx> wrote:
>
>
>>On Thu, 17 Nov 2005 22:52:53 +0000 (UTC), bdbryant@xxxxxxxxxxxxxxx
>>(Bobby D. Bryant) wrote:
>>
>>
>>>What is the status of "the Baldwin effect" in the modern theory of
>>>evolution?
>>>
>>>Is the term "Baldwin effect" even used by biologists?
>>>
>>>Can you recommend a relatively recent paper that presents the modern
>>>view of the the topic?
>>
>>Check out the bibliography in
>> http://www2.create.human.nagoya-u.ac.jp/baldwin/
>>It is a year old, now.
>>
>>You will find no real biology papers. It is all theoretical and
>>computational, artificial life, and the like. A similar result is
>>found search PubMed. In fact, that site
>> http://www2.create.human.nagoya-u.ac.jp/baldwin/introduction.html
>>claims that
>>
>>The Baldwin effect concerns the trade-offs (the costs and benefits) of
>>learning, in the context of evolution. The Baldwin effect should be of
>>interest to:
>> * those who are doing research on Genetic Algorithm / Neural Net
>>(GANN) hybrids or other Genetic Algorithm / Learning Algorithm hybrids
>> * those with an interest in Evolutionary Psychology and the
>>relations among learning, instinct, and evolution
>> * researchers in machine learning, genetic algorithms, neural
>>networks, evolutionary theory, and cognitive science
>>
>>Noticeably absent from the list are evolutionary biologists.
>
>
> Yeah, that's why I was asking. I've got a big pile of papers for the
> machine learning use of the idea, but I was wondering what biologists
> think about it.
>
As Richard noted, they tend not to pay it much attention, and this bothers me.
It is clearly of interest that behaviour can deform or change the fitness
landscape, which is what the BE implies. It ought to be the subject of
discussion in biology. For example, Gavrilets doesn't mention it, Mayr gets it
wrong in his glossary to _What Evolution Is_, my aging edition of Futuyma
faisl to mention it, and the only mention of it in _Cycles of Contingency_, a
summary of DST and related topics, is made by Weber and Depew - at least they
get it right:

"... the Baldwin effect generally asserts that if an organism chances to
exhibit a behavior that permits more effective interaction with its
environment, and can pass that behavior along mimetically, then descendents of
such individuals and populations will on average do better than their
competitors, in turn making it more probable that germline factors that
promote the behavior, if they independently arise, will be more quickly and
effectively taken up and moved in an adaptive direction (Baldwin 1896). Even
though, properly interpreted, the Baldwin effects does not suggest, or at
least require, that multigenerational phenotypic continuity will induce the
kind of genetic change necessary to support it, this notion was downplayed by
the makers of the Modern Synthesis until it became very recessive indeed,
killed for the most part by the usual method of accepting it in theory but
denying it in practice." p250

*I* think the reason they downplayed it is because they simply didn't get it.
And the way it is expressed here by a sympathetic commentary shows why - it's
hard to wrap one's brain around it, so here's *my* formulation:

Any behaviour than can be individually acquired that allows the use of an
otherwise untapped resource will increase the fitness of the organisms that
exhibit it. Another way to express this is that the behaviour changes the
fitness landscape for that organism.

Once the behaviour is exhibited, it is clear there is a cost to fitness in
acquiring it (time taken to learn it by experience), balanced by the benefit
it brings. There will therefore be selective advantage to any method of
acquiring the behaviour more cheaply.

Two ways this can happen are: mimetic learning, or cultural transmission; and
selection for alleles that predispose the acquisition of the behaviour. In
other words, any mechanism or mutant that cheapens the "purchase price" of the
behaviour will increase the "profit" gained. Since genetic dispositions are
much cheaper than individual learning of culture or by trial and error, when
they arise, they will be favoured over the other modes, and so the behaviour
will become entrenched in the development of those organisms over time.

I think this is clearer than trying to be historically accurate about what
Baldwin, and the others who independently formulated it about the same time,
actually said.

--
John S. Wilkins, Postdoctoral Research Fellow, Biohumanities Project
University of Queensland - Blog: evolvethought.blogspot.com
"Darwin's theory has no more to do with philosophy than any other
hypothesis in natural science." Tractatus 4.1122

.

Relevant Pages

  • 1st CFP: Evolution and Dynamics of Learning Workshop (ECAL07)
    ... FIRST CALL FOR PAPERS ... Evolution and Dynamics of Learning ...
    (sci.bio.evolution)
  • The Baldwin Effect: What is it trying to say?
    ... Apparently the Baldwin Effect has some efficacy in computational models ... Baldwin Effect has had minimal effects on evolution. ... Learning is the most cited example of Baldwin's Effect. ...
    (sci.bio.evolution)
  • Re: James Mark Baldwin: "A New Factor in Evolution"
    ... > You will find no real biology papers. ... > learning, in the context of evolution. ... > * those who are doing research on Genetic Algorithm / Neural Net ...
    (talk.origins)
  • Re: The Baldwin Effect: What is it trying to say?
    ... Baldwin effect myself and would like to understand it better. ... Response: Baldwin's Effect is critiqued by some scientists as "social ... > Baldwin Effect has had minimal effects on evolution. ... plasticity by polymorphism of two phenotypes. ...
    (sci.bio.evolution)
  • Re: What did that thread indicate?
    ... I have to make up extremly simple low-level reinforcement learning ... > is a long way from understanding human intelligence. ... Evolution is only two things. ... DNA is not the mechanism of evolution. ...
    (comp.ai.philosophy)
Loading