DNAunion@aol.com writes
in message <c2.18ff1aa.271359f1@aol.com>:
> At another site, a person posted that there was a papered titled
> “Evol ving Circuits with Mutation and Selection” at
> http://www.cogs.susx.ac.uk/ users/adrianth/ices96/paper.html and
> asked for comments. These researchers res ults were mentioned at
> this site elsewhere. What follows are portions of my re sponses
> from the other site.
>
> DNAunion: The selection was performed by intelligently-created
> algorithms & #8211; not by the circuit or the environment. Each
> “generation” co nsisted of 50 different “genotypes”
> being tested for their respecti ve “phenotypic” expression
> (I love the way the authors throw around biological terms so
> loosely), and according to the results of a formula applie d to
> the outputs, the most “fit” was retained. The researchers
> cal led upon a separate mathematical formula (which they had a hand
> in creating and tuning) to determine which “genomes”
> to retain and which to discar d – the circuitry itself had
> no idea which were more fit – the sour ce of the selecting
> was external to it.
This appears to be a cut & paste from the ARN board of part of
a thread I was involved in. I guess the best response is to
cut & paste my response made at the time.
[From the thread "Evolving Circuits with Mutation and Selection" ]
To review, I'm trying to understand why an ID'er would claim that
many experiments that purport to demonstrate the power of natural
selection are actually invalid. I've heard several reasons from
DNAUnion but I find them to be unsatisfying.
The common theme of these reasons appears to be an objection to
selecting a goal and then developing a fitness function to reach
it. Since nature does not select a goal and then develop a fitness
function to reach it, it is argued, the conclusions of such an
experiment can not be extended to natural processes in any way,
shape or form.
While, superficially, such an argument has appeal, I observe that
there is some subtle confusion going on over terms and processes.
Let's look at the term "goal". We use the word to describe
practically any configuration that we find desirable or would expect
some intelligent agent to find desirable. However, our identifying
something as a goal does not logically affect in any way whether
or not it came about by intelligent or non-intelligent processes.
Even if we direct each step leading to the goal, this still
doesn't rule out natural processes imitating each step.
For example, let's suppose we wanted to model pothole formation
with a slab of limestone, sand and a water pump. We arrange the
items to achieve the steps that will lead to our goal -- the water,
sand, and we even go so far as to carve a slight initial channel
in the limestone until we observe the proper eddies in the current
best likely to produce pothole-like erosion. According to DNAUnion's
argument, the results of this experiment can't tell us anything
about nature because erosion in nature doesn't contain a directed
goal. Substituting some terms from my erosion experiment into his
argument from my erosion, we might hear the following:
[quote][i]
In the river erosion experiment, there was only a single functional
goal -- the pothole--, and it was not arrived at by pure chance.
The intermediate selection events -- the angle of inclination of
the limestone, the initial channel in the limestone -- all based
on a fixed, predetermined, future goal, by comparing the current
flow to desired final "pothole-forming" flow. The selection was
based on a future goal, not on any current advantage. The final
function was not arrived at by pure chance - the reason it occurred
is because the entire process was directed towards achieving that
single specified goal. This does not model natural erosion.
[/i][/quote]
Clearly, this criticism in invalid since such experiments in the
lab are quite useful and are quite predictive of the various stages
of pothole formation in nature.
However, suppose potholes had never been observed in nature. What
conclusions could we make from this experiment? Only that:
* [b]If[/b] natural process are capable of combining water and sand,
in a mixture flowing over limestone at a particular angle
to induce a certain form of current eddy, [b] potholes
will form in nature[/b] barring unknown processes that prohibit
their formation.
Now let's bring this back to the particular experiments at hand.
First, Dawkins' "ME THINKS IT IS LIKE A WEASEL" analogy. Obviously
Dawkins does not want us to conclude that nature is directly capable
of producing the sentence "ME THINKS IT IS LIKE A WEASEL" from an initial
jumble of letters, but rather that nature is capable of achieving
complexity through the process of mutation and natural selection.
I.e.
* [b]If[/b] natural processes are capable of natural selection
and mutation, and "MTIILAW" is analogous to a natural complexity
in nature having a superior degree of fitness, and intermediate
states from nonsensical sentences all the way up to the final
configuration are analogous to intermediate stages of natural
complexity having intermediate degrees of fitness, [b] then
natural complexity is expected to emerge from nature [/b] barring
unknown process that prohibit its formation.
And let's bring this back to the experiment that started the original
thread -- a genetic algorithm programming an FPGA to select between
two voltage values. Again, the authors don't want us to conclude
that nature is capable of evolving an FPGA and programming it to
select between voltage values, but rather that nature is capable
of creativity through the processes of mutation and natural selection.
* [b]If[/b] natural processes are capable of selection and mutation,
and the FPGA and associated program analogous to an environment
of lifeforms undergoing selection and mutation, then [b] analogous
creativity is expected to emerge from nature[/b], barring
unknown processes that prohibit its formation.
In conclusion, then, many experiments modeling nature or modeling
processes assumed to be present in nature use intelligently directed
steps and goals to imitate the undirected steps and "goals" of
nature (undirected, that is, to the limits of our perception).
Criticizing an experiment because it has a directed goal misses
the point. There are other, valid criteria to do that, such as:
1. Are the steps analogous to nature?
2. What unknowns (or "little-knowns") in nature may affect results?
3...
This archive was generated by hypermail 2b29 : Mon Oct 09 2000 - 18:25:42 EDT