RE: implied or inferred memes

Raymond Recchia (
Sun, 26 Sep 1999 15:42:52 PDT

From: "Raymond Recchia" <>
Subject: RE: implied or inferred memes
Date: Sun, 26 Sep 1999 15:42:52 PDT

Reply to a John Wilkins post that started with a Derek Gatherer Post

> >Derek:
> >
> >Well, I think John can explain it better than I can, but > >basically
>it's a condition for evolution to be able to > >respond to selection. If
>the background level of mutation > > >is too high, there's too much of a
>chance that a selectively > >advantageous mutation will mutate into
>something else
> >before it has a chance to spread. One might argue (not
> > necessarily correctly but still one might argue...) that >modern
>multimedia cultures are too novelty-prone in most > > >areas. There's no
>time for a trend to take a grip before > >the next craze replaces it. This
>is (was) (perhaps) less > > >the case in
> >so-called less developed societies.

I wonder if reference to Manfred Eigen's error catastrophe theory might
useful here. John Maynard Smith and Eors Szathmary do a decent job of
explaining it in "The Major Transitions in Evolution" (1995) published by
W.H. Freeman pp 44-49. The equation is N < ln s/(1-q), where N is the
selectively maintainable amount of information, s is the selective
superiority of one state of that information and q is the copying fidelity
per digit. I first encountered this Eigen's
"Hypercycle" and I knew it was important but I did not quite understand it
until I re-encountered it "The Major Transitions in Evolution" . My
interpretation is that if the mutation rate in a population is too great,
more complicated units of evolution (be it memes or genes) will be unable to
sustain themselves unless they have a very high selective advantage.

John Wilkins wrote:

>I'm stepping into this thread late, so I may be missing the >point but here
>Fisher was dealing with the recently accepted particulate >theory of
>inheritance (Mendelian population genetics as he and
>others set it up).
>He was therefore discussing in the first chapter of Genetical Theory what
>this did to the arguments Darwin adduced in the > > third (IIRC) and
>subsequent editions of the Origin to deal > with Fleeming Jenkins'
>argument that selection would fail if inheritance were blending (the
>classic case of a "superior" > white man on an island of "inferior"
>Fisher showed that if inheritance were blending, any novelty would be
>swamped by the population norm within a thousand > > generations, and so
>blending inheritance required a much > higher rate of novelty for
>selection to be effective. But, > effective it could be if there was
>sufficient novelty and > high rates of selection. Of course, if heredity
>were > particulate (we would now say modular), then it could persist for
>very long periods, and even infinitesmal amounts of selection coefficients
>(in panmictic, that is, universally interbreeding, andeffectively
>infinite, populations) could > result in geologically rapid evolution.

I really do have to read Fisher at some point and follow that whole line
of reasoning. The modularity as opposed to blending fits in with some
thinking I did a few years ago and I think also points to a way to do an end
run around N in Eigen's equation.
A few years ago to explain some the advantages of evolutionary selection
as opposed to a purely random selection I came up with a little hypothetical
model evolution system that involved manipulating the string of symbols
OGIATWDEHS. In my system the string of symbols had to remain the same, but
it could be in any order. So IATODWESHG or TDEGDWAIHSO were possibilities
but changing the number of certain symbols or the length of the pattern were
out of the question. So therefore IAAAAIDDDD or WHS were not possible One
combination of the symbols was a best combination, that in less abstract
sense might be analogized to a string of amino acids that have a particular
formation that provides the best enzyme for a hypothetical biological
reaction. Randomly selecting among all the combinations it would take at
least few thousand selections to find the right one, but if you start with
one pattern, create variations that differed from their parent by one switch
in position of letters (like switching the O and the S in OGIATDEHS to get
SGIATDEHG) and then select one of the best of those to start a new
generation it would take only a few hundred combinations to get the correct
pattern of ISAWTHEDOG. (My numbers may be off here. The actual work is on a
floppy disk three hours drive away from where I presently live)
Now we would end up with a different set of numbers if we selected
either a more complex or more simpler method unit of selection. If for
example we started out with words instead of just letters, so that we have
something like DOGSAWITHE to begin with we have a lot fewer combinations to
work with. On the other hand selection could be at the level of the lines
and curves that make up the letters.
As I said this allows to do end run around N in the equation because the
total number of symbols is lower. Of course selecting at a more complicated
level of symbols leaves out interesting possibilities that could otherwise
arise like IWASTHEGOD because manipulation within the symbols wouldn't be
I think this is what Wilkins is referring to here. If evolution takes
place at a level of infinitely small mutations then large genes and large
leaps between genes are much more difficult. In biological life we see
evidence of mechanisms for recombination at many levels. At the most basic
level there is base pair alteration. At higher levels there is the
suggestion that introns and crossing over and the phenomenon of jumping
genes work to allow recombination at of higher level functional groups (I am
may be a bit dated on my information here. I have not really paid attention
to the molecular biology literature for a few years. I am a lawyer after
all.) At higher levels still there is the whole chromosome shuffle of
Recombination at higher levels is area where memetic evolution would
appear to have distinct advantage over genetic evolution. Combining memes
into more abstract levels that may then be manipulated and recombined
appears to be a much more prevalent process at the memetic level. We take
individual colors and form the higher level abstraction of 'colors'. We
take the process of making a hammer and the process of making a saw and
combine them together in 'toolmaking'. These higher level memes are then
manipulated together to form yet other memes.

I also liked the rest of Wilkins post because it mirrors some of my own
thoughts, but I do not really have anything to add to it.

Raymond O. Recchia

Get Your Private, Free Email at

This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)