Re: Behaviorism vs. evolutionary psychology

On Jul 10, 8:32 am, Don Geddis <d...@xxxxxxxxxx> wrote:
JGCASEY <jgkjca...@xxxxxxxxxxxx> wrote on Mon, 09 Jul 2007:

... Natural selection selects it does not create the things
it selects. These are the products of chance and necessity.


I'm not sure I quite understand your point. Of course evolution
requires both mutation and natural selection. It is mutation,
where the genotype becomes expressed phenotype, that new traits
appear in nature.

How does that fail to explain the tiger stripes or leopard spots?


Mutation and natural selection may not be the complete story.

A little book I read "Life's Other Secret" by Ian Stewart and
other books such as "The Web of Life" by Fritjof Capra gave me
a wider perspective of the subject.

In theory yes they stripes or dots may have evolved purely
by chance to then be selected for the reproductive advantage
they gave the owner. But did they occur by chance? Or are
they are common outcome due to the physical nature of things?


What part "simple mathematics" leads you to predict ibex
horn shape?


I'm really not understanding your comment...


It is not my comment but that of D'Arcy Wentworth Thompson,

http://findarticles.com/p/articles/mi_m1134/is_10_107/ai_53378962

Think about the problem of designing a set of "molecules" and a
set of laws that would allow these molecules to evolve. We know,
from observation, that it is the case with the current molecules
and whatever laws govern their behavior that they can evolve.

However they themselves are complex and may need to be complex.

Evolution doesn't explain why or how these molecules or laws
are primed to evolve. What basic structures can come into
existence, by chance, to be selected in stages and that these
stages in turn have the right stuff to produce more complex
structures, by chance, which in turn can be selected.

Shake a bag of coins. What are the chances of a column of
coins forming. How many millions nay billions of shakes might
be required to get a column of perfectly aligned coins 20 deep?

Now do the same with a bag of magnetic buttons. They are made
of the right stuff. But not the right stuff to evolve any
more complex structures.

Take reinforcement learning and the combinational explosion.

In theory trial and error should, given enough time, bring
about any pattern to be selected. But in fact the Universe
will not be around long enough for that to happen. Somehow
there are constraints in these systems that mean that they
can, within a reasonable time limit, produce the right
patterns to actually be selected.

The constraints make it possible. They deliver possible
things to be selected within a reasonable time limit.
That is not to be found in the mutation or selection process.

This is my objection to Curt's simple view that it is "nothing but"
reinforcement learning. But we know the combinational explosion
puts a damper on that. It is "nothing but" also of nature having the
right kind of constraints that can be taken advantage of to reduce
that RL trial and error component to a manageable size in the right
kind of learning system.

The idea of reinforcement is the easy part. Natural selection of
mutations is the easy part. That those mutations are ever worth
selecting is the hard part to explain.

--
JC


.

Quantcast