Re: Is "(Random Mutation) + (Natural Selection) => Complexity" Scientific?


"shlimazel" <aashkinazy@xxxxxxxxxxx> wrote in message news:1182359330.038800.212300@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
On Jun 20, 11:33 am, "Perplexed in Peoria" <jimmene...@xxxxxxxxxxxxx>
wrote:
"shlimazel" <aashkin...@xxxxxxxxxxx> wrote in messagenews:1182344100.335628.113630@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
On Jun 19, 6:00 pm, shlimazel <aashkin...@xxxxxxxxxxx> wrote:
The combination of natural selection and random gene mutation
is claimed to be capable of creating the incredible complexity
of life.

Is this statement scientific? If so, it would be necessary to
be able to make verifiable and/or falsifiable predictions that
follow from it.

But what specific prediction can be made? Since the underlying
process of variation is random, the only possible positive statement
is: "some kind of organism, fit enough to survive and reproduce,
will evolve from a previously-existing organism".

But this is an obvious, trivial statement which, by definition of
natural selection, is a tautology that can't be falsified. So
therefore, the claim does not seem to be scientific.

So, what am I missing here? What predictions/tests/observations
could possibly be made to falsify the theory that complexity arises
from random mutations plus natural selection?

[snip]
PIP interpreted my questions fairly accurately, but the tests he
proposed don't quite address the fundamental issue: how does one
distinguish between random and nonrandom mutations after the fact?

It is difficult, but perhaps not completely impossible. A
'smoking gun' falsification of a hypothesis of 'randomness' is
probably not possible. But there are some statistical tests that
can be applied. The hypothesis of randomness is a 'null hypothesis'
in the terminology of statistics, and that branch of applied
mathematics teaches us how and when to reject null hypotheses.

All we see are the end results of evolution. ANY DNA differences
between species can always be attributed to random mutations (of
varying probabilities, assuming we could calculate them).

The trick is to realize that ALL of the DNA differences between
species come from (proximate cause) mutations, but that only some
of them have been filtered by selection. For example, changes in
the third base of a codon are mostly neutral from the standpoint
of selection. And statistical tests can be made which fail to
reject the null hypothesis of randomness for this subset of the
DNA bases. But changes in the first and second base are usually
not selectively neutral. So, your point seems to be that these
mutations could have been directed - how would we know otherwise?

The answer comes down ultimately to Occam's razor. We have good
statistical evidence that random mutation explains what happens
in the third-base positions. There is no reason to suppose that
random mutation rates would have been any different in the first
and second position than they are in the third position. So, we
can assume that random mutation was also operating in the first
two positions. There may have also been some directed, teleologically
guided mutations in the first and second positions - we can't
exclude that by purely statistical tests because of the confounding
filtering force of natural selection. But we can ask whether
random selection + NS would be sufficient to explain the results
that we see. I understand that we find that they are.

For example, if we see evidence that some combination of three
specific mutations to a protein coding sequence must have occurred
almost simultaneously (because we know that having only one or two
of those mutations would be strongly detrimental) then we can
calculate the probability that random mutation would have produced
all three at once. If the odds against this happening are
astronomical, yet we know that it must have happened, then we have
strong evidence that those particular mutations must have been
guided, rather than being random. As far as I know, we have no
such evidence for non-random, guided mutation.

The thing that you have to realize here is that the hypothesis
of random mutation is a 'null hypothesis' in the terminology of
statistics. It occupies a privileged position against the host
of competing non-random hypotheses. The burden of proof lies
on the hypothesis of non-randomness. In a sense, it is the
hypothesis of non-randomness which must be 'scientific' in Popper's
sense.

Basically, the problem I see with random mutations being a driving
force in speciation is the same as Popper's problem with
psychoanalysis
and Marxism: since it can explain anything and everything, it's not
falsifiable, and therefore not scientific.

The writings of Karl Popper are not Scripture. Popper's ideas on
'the demarcation problem' have some heuristic value, but they
don't define what is or is not 'scientific'. Popper never really
understood statistics.

Actually, I agree with most of your post. My main difficulties with
random mutation + NS is EXACTLY that the probabilities seem so low
(but don't ask me to compute them) that any reasonable statistical
test would reject them.

The scientific method would be to make this the null-hypothesis, and
then compute the probability of a known mutation. As you probably
know, accepted the statistical uses probability values (P-values)
around .05, and certainly no lower than .001 to reject the null-
hypothesis if the observed outcome has a lower probability. But per-
codon mutation probabilities are of the form (1/20)^n, which obviously
falls below .001. So I would think that any accepted statistical
testing approach would reject random mutation +NS.

But it is easy to demonstrate that this line of thinking is wrong.
I assume you are familiar with Dawkins' "methinks it is a weasel".
In this case, the letter mutation probabilities are of the form
(1/26)^n, but we still deterministically reach the selected string
over a period of just a few thousand generations.

Now admittedly, Dawkins' selection model is naive and unrealistic.
But he does demonstrate the error in your naive (1/20)^n thinking. So,
if you are going to continue to spout your skepticism, you are going
to have to accept the fact that people here are going to ask you
to "compute them" (the probabilities, that is). Compute them
realistically, taking both chance and selection into account.

The writings of Karl Popper are not Scripture. Popper's ideas on
'the demarcation problem' have some heuristic value, but they
don't define what is or is not 'scientific'. Popper never really
understood statistics.

Obviously though, I believe in Popper's definition of science more
than you do.

I'm not sure the word 'belief' is appropriate here. Instead, I
would say that you 'approve' of Popper's definition more than I
do. You want the word 'science' to apply to a narrower set of
activities than I would prefer. But neither one of us really gets
more than a single vote on how the word will ultimately come to
be used.

.

Relevant Pages
Loading