gerold firl  wrote:

> wilkins{at}wehi.edu.au (John Wilkins) wrote...
> > gerold firl  wrote:
> > 
> > > john.wilkins{at}bigpond.com (John Wilkins) wrote...
> > >  
> > > > I think that genes are only a part of what changes in
evolution, and
> > > > sometimes heredity is not genetic. It can be
environmental, it can be
> > > > somatic, it can be cultural. But, and this is my core
thesis, *all*
> > > > these things evolve in a *Darwinian* fashion. Not Lamarckian, not
> > > > progressively, and not because we are able to solve
problems in our
> > > > heads.
> > > 
> > > I'm not sure that the distinction of darwinian/lamarckian makes sense
> > > when speaking of memetic evolution. The meaning of words can change
> > > over very short timeframes, as a result of how they are used. That
> > > seems pretty 'lamarckian', if not downright lysenkoistic. That stuff
> > > doesn't work in biology, but I'm not convinced the terminology can
> > > crossover to these other areas.
> 
> > it can, but there are traps for young players...
> > 
> > The problem is finding the proper analogues to the
"benchmark entities"
> > of biology to use a phrase of Elliot Sober's. The reason why culture
> > looks "Lamarckian" is that we conflate the biological
entity (human
> > organisms) with the cultural interactor (agents, or profiles). But there
> > is no one-to-one mapping between these two domains; an organism (eg, me)
> > can be, with differing degrees of cultural fitness, many different
> > agents or social profiles. 
> 
> I don't understand the significance of your distinction.

What looks like Lamarckism from a biological perspective is perfectly
Darwinian from a cultural perspective. Much confusion has been exhibited
by not treating things at the appropriate level of abstraction. I might
add that this has also occurred in biology as well (e.g., immunology).
> 
> It seems to me the key issue here is the *mutability* of the
> replicator in question: genes store and transmit information in such a
> way as to minimize mutability: they 'try' very hard to not mutate.
> They are encoded as sequences of base pairs on dna molecules, and
> replicate very faithfully. Mutations are rare, and usually
> deleterious.

This is what philosophers call the fallacy of affirming the consequent
(unless they are the ones making it, when it becomes "just metaphor, you
know?"). Genes do not try to do anything of the kind. There is a complex
set of machinery in place that makes genes more stable than they would
otherwise be, in terms of methylation-directed error correction, and so
forth. One form of error correction is, in vivaparous animals,
spontaneous abortion. The stability we see in genes is an effect.
> 
> The lamarckian error resulted from a misunderstanding of the mechanism
> of genetic change; people had no clue regarding the structure and
> function of genes, and were unprepared to comprehend the vast time
> depth of evolution. They postulated that experience could alter
> inherited characteristics - dead wrong for biology, but not for
> memology.

It is dead wrong in both cases. It is wrong in culture because nobody
can contravene the strictures of Hume's problem of induction - if we can
predict how things will work out, it is only ever on the basis of past
experience. Changes to memes occur through a cultural equivalent of
mutation. And I argue that they are "acquired" only with respect to the
*biological* organism, not the cultural "organism". Experience alters
genes in exactly the same way that experience alters memes - through the
accrual of previously successful variant forms that bias the frequencies
of either in the relevant populations and lineages.
> 
> Memes *are* altered by experience. Memes transmit and store
> information using a much less durable substrate than genes. Memory is
> evanescent, and memes are constantly altered by usage. Of course, the
> timeframe over which memes evolve within a culture is significantly
> longer than it is for an individual, but still much faster than rates
> of genetic evolution.
> 
> That's why the distinction between darwinian and lamarckian evolution
> doesn't seem to matter much for memes.

Consider this: if scientific memes are transmitted, how do they
constitute a scientist the way genes constitute a Homo sapiens? There
has to be a developmental aspect to anything that is transmitted and
subjected to selection, otherwise there is no "phenotype" (or phemotype
- this is memetics, after all; we neologise like rabbits breed).

Now, what is it that memes *do* developmentally regulate? It certainly
isn't the maturation of human beings per se - if you express a gene for
male pattern baldness then you will at most slightly affect that
expression through culture. Memes regulate the maturation of cultural
agents. Hence, I argue, that the biological entity - the organism - is
not directly related to the cultural entity (the professional, the
expert, the agent; in short, the cultural profile).
> 
> > I am unlikely to memetically take the pop
> > music world by storm with my shower renditions of modern and past hits.
> > I do slightly better with my attempts at philosophy (at least I get
> > published occasionally). Neither of these greatly affect my *biological*
> > fitness, say, with respect to malarial resistance or lactose tolerance.
> 
> I don't think that is germane to this issue. Memes evolve according to
> their own evolutionary logic, without reference to physiological
> fitness enhancement or degradation. Considering the mechanism of
> memetic selection, variation, and transmission, lamarckian evolution
> seems like a reasonable description.

This is going to depend on what you mean exactly by "Lamarckian". This
term is an equivocation on several meanings, and it would be best to be
clear about what is meant. I distinguish three senses of "Lamarckism":

1. Directed variation, or anticipation of future needs

2. Soft inheritance, of the instruction of the hereditary substrate with
information gained through the lifecycle of the relevant individuals

and

3. Progressivism, or the idea that change must be increasing in some
significant property.

Now, cultural evolution is not directed (Adam Smith and David Hume
demostrate this), although it is constrained (as biological mutation can
also be). The inheritance of memes is not modified by the experience of
the cultural agent - it is a direct effect of the success had by the
agent in teaching those memes. Consider a scientist who, through trial
and error, happens on a good technique or idea. That will succeed
memetically to the extent that it can be passed on to others, not on the
basis that it was acquired through experience. And finally, and perhaps
more controversially, culture is not a progressive process (not even in
science - all advances in a field are local adaptive trends).

So I reject the idea that culture, and in particular science, is
"Lamarckian" in any of the three senses.
> 
> > > That is a forlorn hope - there is no way human biologic can compete
> > > with AI when it comes to computational complexity. Obviously there is
> > > still a lot of room for improvement in the human genotype - why
> > > shouldn't old-fashioned 'genius' -type intelligence (ala newton,
> > > einstein, mozart etc) be average rather than exceptional? That seems
> > > like an obvious target for genetic engineering. But trying to keep up
> > > with inorganic intelligence using wetware is a loser.
> 
> > I'm unsure that this is entirely coherent. What are the measures that
> > are in common? The computational complexity of something depends largely
> > on what sort of computational power would be required to model all
> > cogent facets of the system. Human cognition does not end at the
> > interface of brain and sensory system in an individual. If there are X
> > million scientists, then there are the number of neuronal
> > interconnections in each brain to the power of X possible connections in
> > human cognition, degraded by the amount of information that can be
> > passed between them (the channel capacity, in Shannon terms). I'd like
> > to see th AI that can even get into that ballpark, let alone exceed it.
> > 
> > Come to think of it, I probably wouldn't.
> 
> You're being overly pessimistic - but lets stick to the computational
> issue.
> 
> There is no reason to believe that AIs will not be able to surpass
> human intelligence - possibly fairly soon (within a few decades).
> Looking at the organization of the human brain, it appears eminantly
> plausible that we will be able to build a more intelligent AI simply
> by piling-up raw connectivity.

I agree it is possible in-principle. I just think that we underestimate
the computational complexity involved in human intelligence - it is the
sum of all the individuals in a culture; and not just the capacities of
an isolated individual CNS. And I cannot see us getting to that level of
complexity through simple engineering any time soon (all bets are off if
we can grow AIs, but then I would suggest that the claim changes).
> 
> Are you familiar with the idea of the 'technological singularity'
> pioneered by vernor vinge? Seems pretty ironclad.
> 
Argumentum ad scientifiction? :-)

-- 
John Wilkins
[I]magine a puddle waking up one morning and thinking, "...interesting
 hole I find myself in - fits me rather neatly, doesn't it? ...
must have been made to have me in it." Douglas Adams, Salmon of Doubt
---
þ RIMEGate(tm)/RGXPost V1.14 at BBSWORLD * Info{at}bbsworld.com

---
 * RIMEGate(tm)V10.2áÿ* RelayNet(tm) NNTP Gateway * MoonDog BBS
 * RgateImp.MoonDog.BBS at 10/26/02 3:09:11 PM