[From Bill Powers (940809.1107 MDT)]
Martin Taylor (940808.1530) --
Remember that the "normally fit" population is far and away the
majority, whatever happens to the mutations in the relatively unfit
few. Almost all mutations in the next generation happen in those who
were unaffected by the previous bad luck.
Yes, I said that for low mutation rates, the relatively less fit
(unmutated) population would contribute the most to speeding up the
process.
"Fitness" is the relative probability that a gene is reproduced at some
later time. That's all.
Right. It really has nothing to do with the normal meanings of "fitness"
such as strength or health or intelligence, although there are plenty of
people who seem to assume such causal connections. If the organism
reproduces more successfully it is more fit; if it is more fit, it
reproduces more successfully.
I've wondered, by the way, how this definition of fitness applies to
species which overpopulate their niches. Couldn't an improvement in
fitness (as defined) prove to be a liability to the species?
Your main thesis seems to be an intuition that neither sexual crossover
nor mutation at a fixed rate are adequate to account for the rates of
change (in what? species? characteristic features of individuals?) that
are observed.
It's more like wondering whether the numbers used in explanations
haven't been cooked to make the calculated behavior match the observed
behavior. This is model-building, after all -- one adjusts the
parameters to get a reasonable result. To get these models to work
properly you have to assume success rates for mutations (or crossovers).
Since we don't know how either a mutation or a crossover actually
affects reproductive rates, we are forced to suppose that the effect is
great enough and specific enough to the problem to account for the
historical record. If a far more effective mode of selection were found,
then there could be a corresponding relaxation in our assumptions about
how effective purely random mutations or crossovers have to be.
The ECOLI phenomenon shows us a way in which random mutations can be
given far more direct and systematic effects than they would have if
there were no systematic adjustment in frequency. If that process can be
shown to work in evolution, our story becomes more plausible.
Part of the problem seems to be that you seem to think
that if one individual acquires a bad mutation, then the species has.
I was thinking more of the situation where progressive environmental
changes gradually reduce the fitness of the "normal" population, not
through mutation but simply by making life more difficult. I assume that
such selection pressures work on all individuals in the population. If
the normal population could speed up its mutation rate when under
pressure, it would be producing variants faster and would thus go
through fewer generations before creating descendants capable of
withstanding the pressure (assuming less than one mutation per
generation). Without the increased mutation rate, the population that is
now relatively unfit would just go on reproducing as is, producing more
unfit generations instead of "looking for" more fit variants.
It's not like the E-coli situation, in which if the single entity swims
the wrong way, all of it is further from the goal.
No, but if the species as a whole goes on reproducing in an ineffective
direction, there will be fewer changes toward an effective direction.
Fitness, in evolution, is a bit like a ratchet. If one good mutation
happens to occur, by which I mean that it raises the fitness of the
genome in which it occurs, then that gene is more likely to be
represented in genomes of the next generation. It may not be, but it
is more likely to be than is the unmutated version of the gene.
Right. So if you could watch the species and press the space bar for all
those individual lines that are reproducing relatively unfit offspring,
and do this whenever relative fitness declines, you would more quickly
eliminate the wrong lines of propagation and start more profitable
lines.
RE: simulations
Could be done, but is it worthwhile other than as a matter of
entertainment interest?
What a strange question. Are you saying that if the present model seems
to work, given a friendly reviewer, there's no point looking for a
better one?
If you remember, this thread started because I argued that PCT was
probably the ONLY way stable systems could have evolved, and I used the
metaphor of evolution as a "designer" of efficient systems. The intent
was to argue (as I have often done explicitly) that organisms that
waste energy on side-effects will be less fit than those that spend
more of their energy on controlling their perceptions, and THEREFORE
that we must have evolved so that "all behaviour is the control of
perception."
I see no link between wasting energy on side-effects and being less
successful at reproduction. It is perfectly possible that the effect of
wasting energy on reproductive success, if any, is so slight that it
couldn't possible explain evolution. Your assumption is like too many of
those made by proponents of the current evolutionary model: anything
that even seems as though it might have some effect on relative
reproductive success is taken as a selection factor. This mode of
argument skips, of course, the step of showing that the supposed
selection factor would in fact affect reproductive success. All you have
to do is drop the assumption, and you will immediately think of other
ways of reasoning that could produce the opposite conclusion. The guy
who wastes energy on flashy behavior and profligate but useless displays
of expended energy might really excite the most fertile girls, whereas
the dull drab conservative energy-saver has trouble getting a mate at
all. I'm not saying that's true, either -- just pointing out that an
unbiased view of either explanation shows that it's just hot air,
generated by a foregone conclusion.
We then got into a discussion, and I did a trivial simulation that
illustrated the interesting fact that a steady mutation rate leads to
"punctate evolution." The reason for that is that when a good mutation
occurs, it may hang around for a few generations and then die off, or
it may get lucky and find itself expressed in several individuals quite
quickly.
Your simulation had nothing in it about good or bad mutations and their
effects on reproductive success. You went directly to reproductive
success. All your simulation showed was the properties of random
processes in a situation where the effects are cumulative and the total
numbers were limited. There was nothing in the simulation to explain
_reasons_ for reproductive success.
When that happens, the "good" genome usually "takes off" and very
quickly comes to be dominant in the population--always along with a
residue of the previous "normal" population, until yet another "good
mutation" occurs and the original normal population cannot stand the
added competition.
Those are your own embellishments on the results; the simulation showed
nothing about good and bad genomes, or actual competition (you just
limited the numbers). The outcome of your simulation doesn't depend on
why there was reproductive success, so it can't be used as proof that
any explanation for reproductive success is correct.
My thinking right now is that it doesn't affect the argument at all if
mutation rates are changeable in situations of biochemical stress.
Changes that make for more fitness are what will be seen in later
generations more probably than not.
Yes, precisely my point. We have no disagreement about what will happen
if there are differences in reproductive success. The whole question is
whether the causes of these differences, as usually assumed, are
actually sufficient to account for the observed differences. We can
agree that _something_ is sufficient to account for the differences. I
am simply raising the question as to whether the explanations offered
off the top of people's heads for no good reason are really sufficient,
or whether we need some more powerful (and demonstrable) mechanism to
account for the observed differences.
Organisms that are good at controlling their perceptions will be fitter
than similar organisms that control worse. And so we wind up with a
world full of organisms in which "all behaviour is the control of
perception."
That's an attractive hypothesis for anyone who wants to show that the
inevitable result of evolution is control systems. Your last statement
sounds, in fact, like a bribe. But simply to assert that control systems
will reproduce more rapidly than other kinds of systems (which is all
you are saying in saying they are "fitter") is far from a sufficient
proof of the thesis. If we aren't _forced_ to this conclusion, it's
useless. I of all people would be very pleased to find that PCT
describes an inevitable organization. But I can't accept proof of the
kind you offer; it's empty.
A great deal is said about the environment "selecting" organisms. I can
see nothing in the environment, other than organisms, capable of
"selecting" anything. To speak of natural selection as something the
environment does is to use a metaphor as a way of subtly asserting what
does not exist. It seems much more plausible to me to say that species
contain selection mechanisms which include the production of random
variations as a means of creating more possibilities among which to
select. Without those capabilities, organisms could not evolve any
faster than a rock does. Evolution is a continuing experiment conducted
by a species, not by the environment. That metaphor, I think, will stand
up to close scrutiny.
···
----------------------------------------------------------------------
Chuck Tucker (940909, direct) --
([Paul] is right - most of your posts are the same length).
So I must be controlling for posts of a certain length? Actually, what
happens is that I always run out of energy before having said everything
that has occurred to me. So you are observing how much I can stand to
write in one bunch.
----------------------------------------------------------------------
Best,
Bill P.
It's quite nice chat.