unsubscribe
Arthur Jackson (ajackson@SSA.CO.SANTA-CLARA.CA.US)
Thu, 19 Nov 1998 07:51:47 -0800
Luis Rocha wrote:
>
> In my previous e-mail I responded to Alexei Sharov, Cliff Joslyn, and
> Francis Heylighen. Here I respond to Howard Pattee, Kalevi Kull, Stan
> Salthe, Jesper Hoffmeyer, and Alexei Sharov.
>
> H H Pattee wrote:
> >=20
> >=20
> > I just read Gould's Full House and I was not impressed with either his
> > objections to the analogy of memes and genes (all analogs are partial) =
> or
> > his statistics against increasing complexity. Your discussion "forced" =
> me to
> > write the followinf brief:
>
> Hi Howard! sorry to "force" you into this ;-)
>
> At this point I must say that I too believe this is not Gould's best
> book, in particular being European I hated all the boring baseball
> details! Also, Gould does not mention memes in his book but merely the
> difference between biological evolution through natural selection and
> cultural evolution. I am to blame for bringing memes into the picture. I
> want to make this clear in response to Kalevi Kull and others. Yes this
> book is incomplete in many aspects but I do think the comparison between
> the two types of evolutionary processes is most important, particularly
> given the nature of the meme metaphor so grounded as it is on the notion
> of gene and natural selection.
>
> My reading of Gould's distinction (as Stan Salthe also emphasized) is
> that his arguments apply to biological evolution qua neo-Darwinism. My
> own interest in separating the two different kinds of evolution is
> precisely to enhance the understanding of evolutionary phenomena both in
> biology and in human culture. I do believe (sorry Stan and Norman) that
> the prime mover in biological evolution is the neo-Darwinian process
> (formalized by the Von Neumann model), but it is very important to
> recognize other forms of evolutionary processes in biology such as
> self-organization, symbiosis, jumping genes, etc.=20
>
> In cultural evolution, on the other hand, I think the darwinian process
> is quite marginal even if it may indeed play a role (say with processes
> like Edelman's theory of neural group selection). It seems to me that a
> much more explicit form of selection has greater influence, call it,
> driven, active, explicit, amalgamation, adoption, or self-organization.
> I think it is a great disfavor to the understanding of evolutionary
> systems to consider all of these processes as some kind of grand
> evolution, especially, =E0 l=E1 natural selection, as the Campbelian blin=
> d
> variation and selection model implies. I favor a lot more the study of a
> taxonomy of evolutionary processes. This is why I like Artificial Life,
> where one can explore all possible evolutionary processes. The price of
> an over generalization of evolution so that in can encompass everything
> from the big bang to super-brains is absolute triviality. In other
> words, I think we should resist the temptation of a grand theory of
> evolution to explain the whole universe.
>
> >=20
> > Campbell's evolutionary epistemology argument is more robust and more
> > general. It shows the essential similarity: All evolution and learning
> > ultimately begin at a state of total ignorance where blind search and
> > heritable selection is the only strategy. How else? Evolved memes diffe=
> r
> > from genes primarily by having more intelligent (i.e., rule-based,
> > model-based, etc.) search strategies (although we are finding genomes t=
> o be
> > more intelligent than we thought).
>
> I disagree with this. Mainly because no one has explained what is a
> memetic search strategy. What is a meme space that is amenable for
> search? Such discourse seems serious but only works because we map it
> back to the genetic domain where it makes sense, but if one really looks
> closely none of these concepts are defined in memetics. Indeed, I do not
> think that memes (whatever they are) spread by blind variation. Take
> Christianity: if such a meme evolved by blind variation, then its
> appearance should have been observed by generation of a large population
> of memes, which by mutation or variation would produce Christianity
> among many others. But we know that it evolved from Judaism by
> incorporation (not mutation or variation) of other ideas coming from
> Rome, Egypt, India, Greece, etc. Also, such incorporation occurred
> because the roman empire happened to greatly facilitate the spread of
> ideas. So Christianity appears not my mutation from Judaism, but by
> adoption of foreign ideas, more a crossover than a blind variation.
>
> In conversation with Cliff Joslyn today, he mentioned that indeed sexual
> reproduction seems to have sped up evolution based solely on mutation
> because of this more effective amalgamation of information. But genetic
> crossover happens vertically from generation to generation and with
> specific genetic (chromosomal) constraints, while cultural crossover
> happens also horizontally and does not seem to be constrained, at least
> between related species.
>
> I also disagree that genes do not have rule-based strategies, but maybe
> I do not understand you here. Many animals (your favorites, the spiders)
> seem to have purely genetic, extremely complicated, rule-based behavior.
>
> Finally, learning does not start at a state of total ignorance due to
> phylogenetic pre-dispositions and propensities.
>
> >=20
> > Complexity in all systems and levels increases open-endedly by virtue o=
> f
> > competing strategies in populations.=20
>
> So you do claim that higher complexity tends to have higher fitness? If
> this were the case, should not complexity tend to always increase?
> McShea's work shows that this is not the case. Minimum complexity
> remains largely unchanged, and species that evolve from other species
> can get either more or less complicated.
>
> > It should be obvious that by common definition,
> > "progress" is in no way related to growth of complexity. In fact progre=
> ss,
> > like success, is meaningless unless it is applied to a pre-established,
> > well-defined goal. Evolutionary progress is therefore meaningless (surv=
> ival
> > is not a goal, it's just whatever is not extinct). It is quite possible=
> that
> > technological progress may result in extinction.
>
> I agree, and Gould also acknowledges this in his book. The point is that
> progress HAS BEEN equated with complexity increase in evolutionary
> biology. It is only this sense of progress that is under discussion.
>
> >=20
> > What is the simplest semiotic system?
>
> Good question... A two-symbol system that can measure one observable
> with at least 2 states, and respond with one of two possible actions?
>
> ssalthe@binghamton.edu wrote:
> >=20
> >=20
> > (1) Note that Gould conflates phylogeny (evolution) with evolution
> > by natural selection. Phylogeny is the large scale process of
> > descent with modification, while evolution by selection is one
> > model (focused on small scale events) by which phylogeny might
> > come about. But for a Darwinian (even a fuzzy one like Gould)
> > evolution is quite simply synonymous with selection, and in that
> > way they can, as it were, outcompete all other models of evolution
> > by way of rhetoric alone.
>
> Hi Stan! Thanks for such a complete response.
>
> I agree. Please note my thoughts above. I don't know what are Gould's
> motives, but my desire to separate different kinds of evolutionary
> processes is based on the recognition that the neo-Darwinian model is
> incomplete and should be supplemented with more evolutionary processes.
> In particular, the Darwinian model is highly incomplete to model
> cultural evolution.
>
> >=20
> > (2) Gould maintains that complexity increases would not evolve by
> > selection, and that is quite true -- nothing can EVOLVE by way of
> > selection alone. Selection is a local generation to generation
> > process which merely picks out types of organisms that happen to
> > be more successful at reproduction in the local environment, while
> > evolution is (usually viewed as) a large scale (multigenerational)
> > phenomenon of organismic change, focusing on traits like eyes,
> > etc.. If we happen to not be scale reductionists, we would never
> > imagine that a slow increase in organismic, or trait, complexity
> > over many generations would be produced by selection. Why
> > should it when environments are constantly becoming altered? --
> > even, as Van Valen pointed out, as a result of the very success of
> > kinds of organisms in outcompeting others. Darwinians connect
> > selection to phylogeny by way of concatenating generations, but
> > that is not a fundamental part of the selection model, and is just
> > stuck on so as to deliver evolution. AND this works only with an
> > unchanging model of the environment, as in Fisher=D5s formulations.
> > American Darwinians (Wright, Dobzhansky) explicitly state that
> > environments are continually changing, thereby formally (and
> > presumably inadvertently) hobbling evolution by selection.
>
> This is an excellent point. I tried to make this point when I mentioned
> that simplicity may be a strategy with better long-term survivability.
> As Alexei reminded us, Conrad proposed that increased adaptability is
> the best strategy. It seems to me intuitive, however, that it will be
> easier for added complexity to reduce adaptability than the other way
> around.
>
> >=20
> > (3) If phylogeny is a material process, then we can easily locate
> > the source of complexity increases in development. I believe it
> > has been demonstrated well enough, so that we can tentatively
> > hold it as a principle, that all dynamic material systems develop.
> > We discover developments by looking for similarities of change
> > across systems. In point of fact, one can always discover some
> > similarities -- making development a logical consequence of
> > inquiry, and therefore necessary for any material system. And, all
> > developing systems so far encountered increase in
> > complicatedness, and usually in complexity too, during
> > development. Hence, if phylogeny is a material process, I would
> > predict that it should show complexity increases -- if only we could
> > find the key to making the right measurements (which McShea
> > has yet to achieve).
>
> Maybe it's up to those that believe that complexity should increase to
> propose the right measurements... McShea did an extraordinarily thorough
> job, if you don't agree that his complexity measure is accurate, then
> propose another study. He makes that invitation himself. Though I mostly
> agree with your expectation, see my response to Sharov below.
>
> >=20
> > (5) The explicit Lamarckian selection of ideas in cultural evolution
> > will have no consequence for a Darwinian selection model like
> > Dawkins=D5. Only if (a) the mutations that organisms can will into
> > existence one way or another (or cultural inventions and
> > ideologies) are preternaturally always going to be just the right
> > solution for some adaptive problem [which seems highly unlikely -
> > - look at societies being led into failure by ideological
> > enthusiasms (Hitler, Soviet Russia, etc.)], and (b) if, by way of
> > horizontal transmission, all organisms in a population come to
> > have the same new trait, would natural selection be defeated by
> > Lamarckian process. Lamarckian inheritance is a red herring in
> > Darwinism, being irrelevant to its arguments (as Darwin himself
> > seems to have thought!). So, I don=D5t agree with Gould and Luis
> > and Heylighen that cultural evolution will be progressive BY=CAWAY=CA
> > OF=CASELECTION. By way of entrainment to structural attractors --
> > yes, OK, that can give you the progressive accession of structural
> > attractor.
>
> I agree with most of your points, but I disagree here. Sure not all
> cultural inventions or adoptions will have the right outcome for a given
> problem, you're right. But the odds are much better for success of
> explicit selection than implicit selection (or blind variation). When
> the Japanese adopted firearms they knew that they were going to work! It
> is a much more rapid transmission of information. In other words,
> adoption reduces largely the need for testing an idea/strategy. Of
> course that the more radical a new idea is, the larger the risk of
> success -- that's why I tend not to believe in revolution!
>
> Perhaps Hoffmeyer's vertical versus horizontal transmission is better
> that Darwinian versus Lamarckian, but the point is that
> adoption/crossover/amalgamation allows the transmission of previously
> tested (partially tested) strategies, whereas blind variation is just
> that, blind.=20
>
> Notice that John Holland often emphasizes his schema theory in the
> crossover operator of his genetic algorithms as precisely the
> transmission of portions of successful strategies. Because of this, I
> propose, Holland's genetic algorithm is much more a model of cultural
> evolution than biological evolution because every element of the
> population can crossover information (schema) with all other elements of
> the population. Crossover is not restricted by species-specific
> chromosomal and developmental constraints.
>
> I mostly agree with your subsequent points, and thank you very much for
> such a great exposition of the issues at stake.
>
> Jesper Hoffmeyer wrote:
> >=20
> > Thank you Luis for informing the list(s) on S.J.Gould's recent book. I =
> find
> > this discussion absolutely important considering the epidemic nature of=
> the
> > meme-"meme".
>
> Hello Jesper, your are quite welcome.=20
>
> >=20
> > If genes were just segments of chromosomes characterized by well-define=
> d
> > sequencces of nucleotide bases this would be true. But if genes are see=
> n -
> > as I would prefer - as semiotic resources for the egg's survival projec=
> t,
> > then genes and memes are perhaps not all that different.
>
> >=20
> > The problem with the meme idea is that the very term "meme" tends to re=
> ify
> > what it refers to.
>
> > What interests me here is, that perhaps just the same might be said abo=
> ut
> > genes after all. For genes as conceived in neodarwinian exegesis are no=
> t
> > "just segments of chromosomes characterized by well-defined sequences o=
> f
> > nucleotide bases". Only rarely are the concrete genes known in any deta=
> il,
> > more often they are imagined elements of genetic material thought to be
> > responsible for a given trait. So also in this case what we have is an =
> idea
> > of a postulated non-ambiguous correspondence between a portion of genet=
> ic
> > material and a trait. What if no such "non-ambigous crrespondence" exis=
> ts?
> > (just as we may presume that there is no non-ambigous individual meme
> > corresponding to THE meme-meme)
>
> I am not sure I understand what you mean by semiotic resources. But if
> your hypothesis that the "non-ambiguous correspondence" of genotype to
> phenotype is correct, then your argument is very sound and most
> challenging for evolutionary biology. However, I still think that the
> mechanism of information transmission would be quite different, and
> therefore lead to a different evolutionary process. As I said above,
> blind variation is very different from adoption of partially tested
> strategies.=20
>
> Your description of the reification problem with the meme idea is very
> very accurate -- I am going to use this in future discussions!
>
> Alexei Sharov wrote:
> >=20
> > If lineages can avoid prior "mistakes", then variation is not "blind".
> > What people call "natural selection" is actually PERCEPTION at the leve=
> l of
> > a lineage! Selective survival of organisms is the same for the lineage
> > as selective activation of visual receptors for animals. The lineage pr=
> obes
> > the environment by producing various kinds of organisms and then makes =
> copies
> > of survivals. This is a cybernetic feedback loop: the lineage first mea=
> sures
> > the environment, and then adjusts its activity according to the result =
> of
> > this measurement. Thus Darwinian evolution turns into something very
> > similar to the Lamarkian evolution if we consider a larger spatial and
> > temporal scale!
>
> The problem with this is that the neo-Darwinian model does not propose
> selection at the level of lineages but only at the level of individual
> rates of reproduction (see Stan Salthe's points).
>
> > I am not sure that "complexity" is a right word to characterize progres=
> s.
> > In many cases progress goes by finding SIMPLE solutions! What is more
> > simple than a wheel? But in many cases good solutions are also complex.
> > We need to distinguish short-term fitness and long-term fitness which i=
> s
> > ignored by neodarwinists. Michael Conrad suggested the term "adaptabili=
> ty"
> > for the long-term fitness. Progress is associated with adaptability
> > rather than with short-term fitness.
>
> I completely agree with this. As I said above, I (or Gould for that
> matter) never said that progress and complexity are the same thing. The
> issue is that traditionally in evolutionary biology this has been the
> assumption. The arguments developed refer to progress only in this very
> restricted sense, which I will be quite happy to change.
>
> >=20
> > >As for forced externally, I may not be understanding you well, but thi=
> s
> > >happens all the time in cultural evolution! Marco Polo went to China, =
> he
> > >saw that noodles were good, and he brought the idea to Italy with a
> > >precise, known, goal. It is not by accident that Italians eat Pasta.
> > >Cultural discoveries are not accidental.
> >=20
> > This is not a discovery, but an adoption! Adoptions are often
> > intentional. And I agree with you, that species are not capable of
> > adoption of ideas.
>
> Then you agree with me! Fine, the difference between biological and
> cultural evolution is that the latter possesses an horizontal mechanism
> of information transmission based on the adoption/amalgamation/crossover
> of partially tested strategies -- this IS a tremendous difference! I am
> quite happy with this discussion because it is helping me to choose the
> right terms. Cultural evolution cannot be described by blind variation
> alone, it requires a model of evolution by
> adoption/amalgamation/crossover of ideas/concepts/strategies.
>
> >=20
> > >One of the tests that McShea proposes for distinguishing
> > >active (driven) from passive evolution is the "test of
> > >ancestor-descendant pairings". Basically, one tries to find a species =
> in
> > >the midlle of a lineage (thus not close to a complexity minimum) and
> > >then eveluate if its descendants become more or less complex.
> >=20
> > Can you define what do you mean by "active"? Why do you think that
> > diffusion is always passive? Following your logic we can conclude that
> > chemical reactions are also passive. But then any organism including
> > you and me is passive because our body functions via chemical reactions
> > and diffusions. I think that any process (including diffusion) can be
> > active if it is used for achieving some goal or value. If you consider
> > only material and efficient cause, then all processes will appear
> > passive. Activity requires a final cause.
>
> The notion of active evolution is the same as McShea's driven evolution.
> Such evolution is obtained from a more explicit form of selection,
> namely, the selection by adoption I discussed above. I would encourage
> reading McShea's work for a more explicit definition. He proposes 3
> tests to decide if we have a passive or a driven (active) kind of
> evolution. Here it suffices to say that in driven evolution, all
> elements (say species) will tend to increase in complexity (or reach
> some goal faster). If you observe the history of such evolutionary
> process, you will observe that not only does the maximum value of
> complexity (or being close to some goal) increase, but also the minimum.
> In a passive type of evolution (diffusion), only the maximum value
> increases, the minimum remains essentially unchanged, and so does the
> mode. The average increases in both cases. This way we can decide by
> looking at the history of some process whether it is passive or more or
> less driven/active.
>
> >=20
> > Nobody argues that the evolution of genes and memes is different.
> > The problem is that you don't want to see any similarity.
>
> You and others have argued that they are essentially the same. I do see
> a lot of similarities and have mentioned those from the start: both
> refer to an evolutionary process of semiotic selected self-organization.
> Furthermore, I do think that cultural evolution has aspects of blind
> variation, and biological evolution has aspects of evolution by
> adoption/amalgamation/crossover, but blind variation is the prime mover
> of biological evolution, and evolution by adoption the prime mover of
> cultural evolution. What I insist is that we should aim at understanding
> the different evolutionary processes that exist, and then give
> appropriate credit to them in each context. We should also be aware of
> the risks of over-generalization.
>
> If I am right about the predominance of different evolutionary processes
> in biological and cultural evolution, more empirical studies such as
> that of McShea should confirm this. In particular, I concede, one will
> probably find (as Stan Salthe suggested) that biological evolution,
> though mostly passive, will also show some increase in minimum
> complexity due to non-Darwinian evolutionary processes. Likewise,
> studies of cultural evolution should show that progress and complexity
> (however those may be defined in social systems) will not increase as
> expected of purely driven evolution, due to the failure of some radical,
> revolutionary, (blind variation) strategies -- though, of course, some
> revolutionary events will also be successful, and needed, just like in
> natural (blind) evolution.
>
> Cheers,
> Luis
> ______________________________________________________
>
> Luis Mateus Rocha (Postdoctoral Associate)
> Los Alamos National Laboratory
> Computer Research and Applications Group (CIC-3)
> Los Alamos, NM 87545, USA
> e-mail: rocha@lanl.gov or rocha@santafe.edu
> www: http://www.c3.lanl.gov/~rocha