SOURCE PDF [ Portable Document Format ] URL:
http://www.semioticon.com/virtuals/imitation/mblute_paper.pdf
-------
THE ROLE OF MEMES IN CULTURAL EVOLUTION: MEMES IF NECESSARY, BUT NOT
NECESSARILY MEMES
Abstract
The concept of cultural evolution is central to any discussion of “memes”. It was because
of the possible existence of Darwinian evolutionary processes beyond the gene-based biological
that Dawkins introduced the concept in the first place as a possible substrate. The meme concept
was generally not very well received in academic circles, albeit the reception among those
interested in Darwinian-style theories of cultural evolution was more mixed. Beyond the
sociology of its invention and reception, objections to the meme concept that there are discrete
units of symbolically-encoded biological information which evolve, - genes, but not of cultural
information - memes, are not persuasive as will be discussed. Both genetically and linguistically
encoded information include units of structure which are discrete and of function which are
symbolic, but because recombination can and does take place structurally within function units,
in neither case do the two necessarily coincide. Hence the relationship between structure and
function can be one to many and many to one (which is not to deny the capacity for evolution).
Despite the complexity, like the gene concept historically, the meme concept can be shown to be
capable of doing a useful job of scientific work. Although a great variety of other terms are used
in a variety of evolutionary social science disciplines, the meme term may be particularly useful
in interdisciplinary discourse. With respect to cultural inheritance, memetic talk may be most
appropriate when social learning is by linguistic instruction and least appropriate when it is by
individual learning mechanisms, but the situation is unclear when social learning is by
observation or true imitation.
1
THE ROLE OF MEMES IN CULTURAL EVOLUTION: MEMES IF NECESSARY, BUT NOT
NECESSARILY MEMES
The concept of cultural evolution is central to any discussion of “memes”. It was because
of the possible existence of evolutionary processes beyond the gene-based biological that
Dawkins (1976:191-201, 322-31) introduced the concept in the first place as a possible substrate.
Strangely enough he, of all people, did not initially clearly distinguish the gene and genome-like
from the phene and phenome-like aspects of cultural evolution, a confusion which he corrected
thereafter (e.g. 2003:119-127, albeit even that discussion somewhat confounded learning by
observation and by verbal instruction). This paper on the role of memes in cultural evolution is
divided into eight sections on the reception of memetics, the evolutionary gene, the related
puzzle of sex and recombination, the role of language in cultural evolution, the scientific
usefulness of the meme concept, alternative terms in evolutionary social sciences, memes and
social learning mechanisms and a conclusion.
The reception of memetics
The meme concept was generally not very well received in academic circles. Books on
memetics were interdisciplinary (which can itself be a problem); they often ignored many of the
conventions of academic discourse; they were sometimes written by non-professionals for a
popular audience; and they were commonly viewed by social scientists, when they paid any
attention at all, as yet another (post-sociobiology) incursion by biologists into their subject
matter. In addition, the fratricidal war between adherents of the gene-like biologically adaptive
view (which can most obviously be associated with vertical transmission) and adherents of the
virus-like biologically maladaptive view (which can most obviously be associated with
2
horizontal transmission) did not help when it should have been obvious that both are possible.
The overall result was predictable. Of the editor of the Journal of Memetics in its heyday and the
authors of all of the books on the topic, none currently hold an academic position. They include
an IT professional, a business person, a professional poker player, one dead of an accidental drug
overdose and a science journalist. On the other hand, the reception among those interested in
Darwinian-style theories of cultural evolution was more mixed. For example Aunger managed to
get a group of academics together including some well known ones including David Hull, Daniel
Dennett, Henry Plotkin, Robert Boyd etc. for the conference that led to the anthology on
“Darwinizing Culture” (Aunger, ed. 2000). That interest continues (e.g. see articles by Gil-
White, Greenberg, and Chater in Hurley and Chater EDS. Vol. II, 2005). Moreover, a lot of those
doing empirical and/or theoretical work on Darwinian-style cultural evolution in various social
science disciplines often at least casually refer to memes. They do so because it helps to
distinguish what they are doing from sociobiology/human behaviour ecology/evolutionary
psychology as well as from the developmental stage theories of history traditional in the social
sciences. A recent example that comes to mind are several of the essays in the anthology on
Mapping our Ancestors: Phylogenetic Approaches in Anthropology and Prehistory (Lipo et. al.
Eds. 2006). As a colleague put it to me, “the concept is out there” and it gets, and undoubtedly
will continue to get, picked up and used in interesting and surprising ways. For example Keith
Stanovich, an accomplished cognitive psychologist, argued in The Robot’s Rebellion: Finding
Meaning in the Age of Darwin (2004) that in pursuit of humanistic and democratic values, we
(the robots of the title) need to bootstrap our way to rebellion against both our genes and our
memes.
Given that in addition to Dawkins himself, some of the greatest evolutionary biologists of
3
our time including George Williams (1992:15-16, 18-19), John Maynard Smith (Maynard Smith
and Warren 1982; Maynard Smith and Szathmary 1995:309) and Paul Ehrlich (2002) have made
clear their awareness of the significance of a Darwinian cultural evolutionary process, and some,
including Luigi Cavalli-Sforza and Marcus Feldman (1981 and subsequently for Feldman) have
even made it a major part of their work, one might have expected biologists to display more
enthusiasm. There currently are some biologists working on cultural evolution (including Kevin
Laland and Alex Mesoudi). I believe however there is a reason why such work is not more
widespread among biologists. The meme concept was introduced just at a time when there were
rising “discontents” (Ruse 2006) within the biological community with neo-Darwinism (as it was
known in Britain), or the synthetic theory of evolution (as it was known in America), i.e. with
population genetics or the genetical theory of evolution. Those discontents included an implicitly
naive view of the origin of life; an extreme micro and gradualist emphasis; an overemphasis on
conflict as opposed to cooperation; a relative neglect of development and ecology; and overly
restrictive theories of speciation and macroevolution. Moreover, it was introduced by the very
person around whose work many of those discontents crystallized. Nevertheless, I think it is fair
to suggest that by its linkage in peoples’ minds, the wide diffusion of the meme concept gave
Darwinian-style cultural evolution a lift, helping move the latter some distance out of the small,
scattered academic niches in which it dwelt at the time.
The evolutionary gene
Beyond the sociology of its reception, objections to the meme concept that there are
discrete units of symbolically-encoded biological information which evolve - genes, but not of
cultural information - memes, are not persuasive. I am going to devote a fair amount of space
here to talking about genes. I think that may be useful because memes are intended to be
4
analogous to genes, and the gene concept itself is commonly taken for granted in the memetics
literature in a way that is problematic. If I can be so self-indulgent as to quote myself:
“Genetic units of structure, function, replication, mutation and recombination do not
coincide with one another. Units of structure are base pairs, nucleosomes, 30-nm fibres, loops,
and chromosomes. Units of function include codons, open reading frames (ORFs, between
initiator and terminator codons), cistrons (functional units defined operationally by cis-trans
complementation tests, once assumed to be coincident with the molecular gene concept of a
sequence coding for the polypeptide sequence of a single strand of a protein molecule), as well
as many, many potential others e.g. with introns counted in or out, adjacent and even distantly
acting regulatory sequences in or out, sequences coding for other strands of the same protein in
or out, sequences coding for other enzymes functioning in the same pathway in or out, and
ultimately even whole hierarchies and networks serving some particular ecological, sexual or
social function. Units of replication are replicons and chromosomes. Units of mutation can be
sequences of any length from a base pair to the entire genome and units of recombination can be
sequences of any length in crossing over and are chromosomes in independent assortment. This
lack of correspondence and consequent multiple ‘gene’ concepts has been the source of endless
angst in the history of biology” (modified from Blute, 2005).
Not only have concepts of what exactly “a gene” is changed historically (for a brief history see
Rheinberger & Muller-Willie 2004, for more in depth discussions see the essays in Beurton, Falk
and Rheinberger Eds. 2000), they also vary currently among biologists as some delightful survey
research has shown (Stotz, Griffiths, & Knight 2004). Some biologists even blog about it (Moran
2007)! This variation and change is of course exactly what a cultural evolutionist would expect,
including in science (Hull 1988).
One can sometimes get the impression from recent literature that this was a new problem
with the coming of the molecular biological discovery of “genes in pieces” and all that but that is
not the case. As early as 1957, Benzer was advocating, on the basis of results from his elegant
experiments on mapping within classical Mendelian genes and interpreting them in terms of the
new DNA theory and later of the Watson and Crick model, that the term gene be replaced by
cistron, muton, replicon, and recon for units of function, mutation, replication and recombination
respectively (Holmes 2000). And from then, at least through the 1960's and early 1970's, the
5
problem was a topic of discussion even in textbooks (for example in the several editions of
H.L.K. Whitehouse’s Towards an Understanding of the Mechanism of Heredity) where I first
read about them in the early 1970's.
Although it was not discussed in Adaptation and Natural Selection, this was the
background against which Williams (1966) classic book was written. Which, if any, of Benzer’s
genes is it, that evolutionists in general, and population geneticists in particular, are talking
about? It was apparent to Williams that it was none of them, that evolutionists needed their own
gene and hence his definitions: “that which segregates and recombines with appreciable
frequency” (p. 24) and “in evolutionary theory” “any hereditary information for which there is a
favorable or unfavorable selection bias equal to several or many times its rate of endogenous
change” (p. 25). Because the point of Williams’ definition was that the evolutionary gene needed
to be short enough to tend to remain intact through recombination as well as be long enough to
affect function and hence be subject to selection, his gene came to be known as that which is
“small enough to be different, and large enough to make a difference” (origin unknown to me).
These are impossible criteria to combine. If being a gene is both a negative and a positive
function of sequence length, which on the simplest assumption combine additively for example,
then no sequence - short, intermediate or long would be any more gene-like than any other. Later
in 1992, in distinguishing between the “domains” of information and matter (“codical” and
“material” domains respectively), emphasizing that a gene is “a package of information, not an
object”, he reiterated the theme that to evolve by natural selection, a “given package of
information (codex) must proliferate faster than it changes” (p. 11). He also noted that the same
thing is true of memes (p. 13).
In 1976, between these two books of Williams, Dawkins introduced the term “replicator”
6
in his classic, The Selfish Gene. Not only did it carry the unfortunate connotation that genes
replicate themselves instead of being replicated by enzymes, imply that the origin of life problem
was pretty much one of ‘once upon a time there came a replicator’, and lent itself to
misunderstanding by uninformed readers that the primordial replicator might possibly have been
DNA, but it also tended to suggest the structural (and hence the short) rather than the functional
(and hence the long) component of Williams definition. Later in The Extended Phenotype
subtitled The Gene as the Unit of Selection (1982) he got more specific by specifying the criteria
of copying fidelity, longevity, and fecundity. Note how the copying fidelity equates with the
‘short enough’ component of Williams definition and the longevity and fecundity (i.e. viability
and reproductive success, the two large components of fitness) equate with the ‘long enough’
component of Williams definition. (To further confuse the waters however he also called it the
“optimon” which suggests the latter i.e. that which functions, whereas replicator tended to
suggest the former i.e. that which maintains its structure!) It probably would have been better if
both Williams and Dawkins had gone to “evcon” which, like Mayr’s, “selectron”, rhymes with
cistron etc. but unlike it, does not unduly emphasize function over maintaining structure (see
discussion in Dawkins 1982: 81).
Whatever term is chosen and whatever description of the criteria is used, the important
point is that the evolutionary gene combines a unit of recombination (which is the unit of genetic
transmission in sexual species) with a unit of function. However since these do not in fact always
or perhaps even often coincide - the evolutionary gene or replicator is as much a fiction, a
theoretical construct rather than a hypothetical entity, as is an ideal gas for example. Williams
(1966) understood this. He often spoke of “hereditary information” instead of genes, sometimes
put “the gene” in quotations marks, called it the “abstract” gene of population genetics (p.24),
7
and noted that such a gene “would produce or maintain adaptation as a matter of definition”
(p.25). Such fictions are useful one must hasten to add. One can write equations and prove
theorems about them just like one can about right angled triangles say, which is what theoretical
population geneticists do. To the extent that things in the real world approximate to the axioms
employed (and they always only approximate), then what can be proven deductively true about
the ideal entity will also tend to be true about the objects in the material as opposed to the
conceptual world. In the real, as opposed to the fictional world, because recombination
commonly takes place among structural rather than necessarily functional entities, the
relationship between structure and function is not always one-to-one but can be one-to-many or
many-to-one. This is the case not only in the traditional sense that one Mendelian gene can affect
many traits and many Mendelian genes can affect the same trait, but also on a finer scale so that
information from the same smaller genomic sequence can end up incorporated into more than
one polypeptide chain and one polypeptide chain can be woven together from the information in
scattered genomic sequences. Such cutting and pasting can and does take place in a variety of
ways at any stage from transcription to post-translation. However, none of this should be taken to
deny the fact that any particular sequence of whatever length has some mean effect on the fitness
of individuals in a population and hence is potentially subject to natural selection.
The puzzle of sex and recombination
The really interesting question is not just how to cope verbally with fact that the kinds of
units required for relatively accurate transmission on the one hand and selectability on the other
hand do not always coincide, but why they don’t. Why have sex and recombination which break
up favourable “gene combinations” (as it is said) in every generation evolved and been
maintained? That, after the origin of life, remains the mystery of mysteries. For example, it is
8
possible that it will turn out that on the level of chromosomes, they do coincide. Chromosomes
may turn out to be also units of function, the only list of the five above (structure, function,
replication, mutation and recombination) from which they are presently absent. If each controlled
some major component of life histories with some specialization by gender, they would probably
assort at random simply to reduce risk. For example, if because of sex-linked or limited
expression, males grow bigger and females live longer, as tends to be the case among mammals,
it would probably be advantageous for males to include females as well as males among their
offspring and vice-versa as a means of reducing risk. Sex is good portfolio management. On that
view chromosomes would be “near decomposable modules of a complex system” (Callebaut and
Rasskin-Gutman EDS. Foreword by Simon 2005), in this case of a genetical system.
Crossing over within chromosomes on the other hand may be as much about conflict (in
an overall cooperative context) or about innovation as anything else. Some years ago Hickey and
Rose (1988) hypothesized that sex had its origin in parasitic DNA adapted for horizontal
transmission. Non-Mendelian ratios as a result of gene conversion are not uncommon although
most easily detected in yeast and fungi where the products of a single meiosis can be observed.
Conventional reciprocal crossing over might well then be viewed as a balanced or dual form of
gene conversion, mutual exploitation if you like. On that view, it would not be surprising that
(ecologically) favourable gene combinations are broken up; crossing over, unlike random
assortment, may not be adaptive from the perspective not only of the diploid genotype, but also
from that of sequences on either side of those initiating homologous crossover events.
Alternatively, reciprocal crossing over which takes place without regard to units of function may
be viewed as a mechanism of innovation. From a functional point of view, such a recombinant is
actually a mutant. On the other hand, given that we do not know what most of the DNA in
9
eukaryotic organisms does, caution is needed in drawing such radical conclusions - we are only
in the very earliest phase of coming to understand the grammar of genomes. Whatever the merits
may or may not be of various views, the point is that the evolutionary version at least of the “lost
gene concept” from the title of Griffiths’ (2002) review of Beurton, Falk and Rheinberger (2000)
is lost in the territory of sex and recombination and his “reward to finder” will go to those who
search there.
In the meantime, there seems to be little choice other than to tolerate the variety of gene
concepts used in particular contexts in particular research traditions in biology which is what
many philosophers and biologists who have considered the question have concluded and what
the majority of biologists in fact do. As an example of how the saga continues, the website of the
HapMap project has taken to calling SNPs “alleles” (i.e. the historic term for different versions
of a gene). So a single nucleotide is now a gene! A second alternative would be to abandon the
term gene completely and use the various historically suggested “on” - recon, cistron etc.
substitutes. Not only would concepts of function need to be multiplied further given current
molecular knowledge, but the approach seems sociologically quite unlikely to be successful,
given that the transition did not take place in their heyday. As all evolutionists know, there is an
inertia to history. A third possibility is to do what Burt and Trivers (2006) largely do in their
monograph on intra-individual conflict, which is to talk about “genetic elements” or sequences,
or even better, genomic elements or sequences, all of which conservatively retain the historic “g”
word or its root while signalling more cognizance of the complexities involved.
Having disabused you, if you were not already, of the notion that the same “gene” is both
a unit of both recombination i.e. of transmission in sexual species and a unit of function - one
perhaps coincident with the molecular biological concept of that which codes for a single
10
polypeptide strand - where does all of this leave memes? In order to approach that question, it
will be useful to consider the role of language in cultural evolution.
The role of language in cultural evolution
Most cultural transmission in the human species takes place using language in whatever
medium it is embodied in - gesture, sound, print, electronic etc. Presumably it is because we can
tell our children (and others) what we know and what to do, rather than just show them that
partially accounts for the great success of our species. Eight points about genetic and linguistic
systems of transmission are basic here.
i) Both genetic and linguistic forms of heredity are digital at their base (base-pairs and
phonemes respectively) which facilitates stability in transmission (Dawkins 1995). So the
problem is not that genes are discrete and culture, at least in linguistic form, is not.
ii) In both genomes and language there are units of function as well as units of structure
(e.g. codons and morphemes respectively) and in both there are a number of more inclusive units
of each (which I will not try to specify further to avoid getting trapped among warring linguists -
linguistics and semiotics generally seem to be the only disciplines more contentious than
sociology!).
iii) Both are said to be symbolic because of the arbitrary or conventional (i.e. historically
evolved) nature of the link between symbol and what it stands for or represents (codons for
amino acids biologically; morphemes and lexical items for their reference linguistically).
Genetically-encoded information is legitimately said to be symbolic in that there is no physiochemically
necessary connection between a genetic codon (a triplet of base pairs in DNA) and
what it stands for or represents (an amino acid in a protein molecule). According to current
knowledge the connection is as arbitrary, is as much a product of history as is the connection
11
between red and stop rather than go, or between the word "father" and a male parent rather than a
female one. It could as easily have been the opposite (Crick, 1968). This is possible
biochemically because "adaptor" molecules are what associate a genetic codon (translated into
messenger RNA) on one side and the amino acid component of a protein it stands for, represents,
or encodes for on the other side. These adaptor molecules (tRNA's recognize the message with
their anti-codon and aminoacyl-tRNA synthetases recognize both the tRNA and the protein
component and link them) themselves have evolved - they are a product of history. To be sure the
meaning of "no necessary connection" in the two cases is substantively different. Biologists say
there is "no necessary connection" meaning no physio-chemical necessity. Linguists say there is
"no necessary connection" meaning in addition that there is no biological or psychological
necessity. Nevertheless the point is theoretically ultimately the same. The association between
symbol and what it symbolizes is a fact of history rather than a necessity.
‘Semiotic talk’ in biology has been subject to criticism in the last decade (stemming
initially from Sarkar 1996a, b), criticism which I believe is generally misplaced. Despite the fact
that the current functioning of the components of cells like nucleic acids and proteins could in
theory be given a complete description in terms of the chemistry of molecular recognition,
enzymatic action and so on, such a reductionist description would be utterly incapable of
providing a broader understanding of the historical evolutionary context that is the reason for
their existence. Nuclei may best be viewed as the brains of cells which perceive, calculate and
act (Blute 2005). Their genomes represent the cell’s encoded memory of events in the past
history of its lineage (including mutation, migration, drift, selection etc.) and, by the same token,
their expectations of and hence basis for action in the future. Moreover, it is sociologically
impossible to imagine a wholesale revision of the language of information, the code,
12
synonymous codons, transcription, translation, proof reading, editing and so on in molecular
biology taking place anytime soon. Indeed, ‘semiotic talk’ is currently expanding rather than
contracting. A couple of fairly recent introductions in genomics are “annotating” sequences (a
way of lumping all considerations of function together) and the search for sequence “motifs”,
variations on which are characteristic of some particular phenomenon.
iv) In neither case do units of structure and function necessarily coincide; in particular the
smallest unit of structure (base pair, phoneme) is smaller than the smallest unit of function
(codon, morpheme).
v) In both, the building of larger units of function from permutations of smaller units of
structure (e.g. of codons from base pairs, morphemes from phonemes) is what makes possible
“unlimited inheritance” (Maynard Smith and Szathmary 1995) and thereby facilitates adaptation
by means of the cumulative evolution of complex, diverse and unique entities in both cases. In
both however, there remains much to be learned about the relationship between structure and
function because in both many units serve purely internal organizing and controlling functions
(e.g. cis and trans acting regulatory sequences as well as probably a lot of the untranslated rna in
cells, many morphemes and words in language which serve purely grammatical functions. Note
that the term “functional” is used in the opposite sense in the two disciplines - e.g. for the purely
grammatical in linguistics.)
vi) Eventually in both one arrives at that which is capable of ‘standing alone’ (a genome,
and traditionally an utterance or a sentence, although many today would prefer a narrative or
text). This should not of course be taken to deny their “embodiment” in either case. Then of
course there are populations or species of such with variation among individuals. We all say
something a little, and sometimes a lot differently, even in similar situations, and we all have our
13
own versions of a story. (If we were talking about the evolution of languages rather than other
aspects of culture expressed in language such as hip hop, scientific or Serbian culture for
example - then these populations and species would be of the idiolects of individuals. A language
is a species with members able to exchange communications linguistically within its boundaries
but not beyond them analogous to a biological species with members able to exchange genes
within but not beyond its boundaries.)
vii) The famed recursiveness of language (clauses within clauses etc.) may or may not in
the long run turn out to be unique to human language among animal communication systems.
However, it is probably not unique in another sense in that it may represent one version of the
major means by which something new is generated in evolutionary processes including the
genetic - by over-duplication and divergence, inserted in this case internally rather than
adjacently.
viii) Grammaticization (in the narrow sense by which sense and meaning are emphasized
or expanded by additions but which then tend to become more economically expressed by the
use of purely grammatical elements both morphologically and syntactically) is a process
apparent in both realms. It is most obvious genetically in the large increase in the proportion of
regulatory “non-functional” DNA in the transition from prokaryotes to eukaryotes. What remains
unclear in language is how much of it took place in the biological evolution of the human
capacity for language (thus confirming the structural linguists’ dream of a universal grammar);
how much in the cultural evolution of languages plural (thus confirming the historical and
anthropological linguists’ view of the cultural evolution of grammar); and even how much takes
place in the acquisition of language by individuals (thus confirming the psychologists’ and
social-psychologists’ view of grammatical acquisition in development - no child begins to speak
14
with clauses).
When culture is transmitted lingustically, there is no systematic recombination process
every generation - which is not to say that recombination does not happen. Much of the famous
“productivity” of language occurs as a result of such ‘reslotting’. We can say the boy ran “up the
hill” or “across the road” or “over to her” etc. As with genomes, recombination in language can
take place structurally without regard to functional units which can destroy the result
semantically including the “sense” of a component and the “meaning” of a whole. The
potentially pathological effect is seen at its most extreme in the “word-salad” of classic
schizophrenic language in which syntax can remain eerily normal while meaning is destroyed.
“If we need soap when you can jump into a pool of water, and then when you go to buy
your gasoline, my folks always thought they should get pop, but the best thing is to get motor
oil” (quoted in Covington et. al. 2005).
Similar phenomena can take place in units smaller than sentences so that schizophrenics
commonly coin neologisms which are morphologically correct in that they could be a word in
the language - except that they are not, as well as in larger units - so that an entire narrative can
be “florid” as it is described, essentially not conveying intelligible meaning. On the other hand,
as in genomes, recombination can also be creative. Some neologisms and even fantastical
versions of a story for example are picked up and widely disseminated.
Van Driem (2001) in his symbiotic theory of language has argued that a linguistically
informed understanding leads to the conclusion that, “meaning” i.e. semantics, is or should be
central to the meme concept and I am inclined to agree. The concepts of function and adaptation
have always been central to the theory of evolution by natural selection and they roughly
correspond to the concepts of the “sense” of a component and the “meaning” of a whole
15
respectively in semantics. What they are properties of also correspond roughly to the concepts of
a meme and a memeplex respectively in memetics (see the discussion in Blackmore 1999:18 ff).
If a meme does not make “sense” (have a useful function) or a memeplex is not “meaningful”
(not adaptive in the setting in which it is found), they are unlikely to be transmitted further
(except in the former case as “uh” a parasite of the rest of the whole). Assuming a gene-meme
analogy, falling back on units of function and adaptation in memetics may seem to contradict
what was said above about genes but it really does not. To recap, genetic recombination may take
place for some or a mixture of all of the following reasons. First, from a larger perspective it may
be a pathological side effect of social conflict. Memes too do not always work well together in
memeplexes - it is not only in talking to schizophrenics that we have all struggled to understand
what someone else “means”. Secondly, recombination can be a mechanism of innovation. With
memes too, sometimes when we struggle to understand what someone else means, that may
because they are saying something new which may turn out turn out to be well worth hearing.
Thirdly, more recombination than we are currently aware of may align with units of function -
not only between chromosomes but also within them. There are many departures from
randomness in crossing over such as inversions and hot spots for example and more of it may
become intelligible as we come to understand the hierarchical structure of genomes as well as we
understand that of sentences which makes possible the productivity of language.
All of this says to me that to the extent that human culture exists in linguistic form which
it largely does, the “meme” concept is no more (but admittedly no less) problematic than is the
gene concept. Hence it cannot be banished on a priori grounds but perhaps like genomic
sequences for genes say, a better substitute should be sought. I do not know what would be best -
language string perhaps - or maybe just information for both as many have chosen would be
16
most useful in the final analysis. In any case, as with some particular genomic sequence, it
remains the case that any particular language string of whatever length has some mean
probability of being transmitted culturally, and hence is potentially subject to selection.
The scientific usefulness of the meme concept
Not surprisingly, as with the gene concept historically, despite the problems, “memes”
can be shown to do a useful job of scientific work. To cite just one good example, inspired by the
meme concept, Pocklington and Best (1997) used a text retrieval algorithm (latent semantic
indexing) to identify sets of rare words that co-occur (sets which they called “term-subspace
traits”) in posts to some news groups on the internet. Since many posts originate in response to
previous posts, they are threaded, i.e. the authors possessed genealogical data. They were able to
show statistically that in some cases the reproductive success of a post, i.e. its success in
generating in-reply-to posts, was a function of the degree to which certain of these traits were
expressed within it. There was some lack of clarity in terminology (they sometimes referred to a
term-subspace as a replicator, other times as a trait, and still other times as an indicator only of
the underlying “cognitive motif” which is the true “conceptual replicator” although later they
seemed to settle on the set of words that co-occur - Best and Pocklington 1999). Despite that, the
study clearly was a proof-of -principle illustration of cultural microevolution in texts.
But also note two things. A set of words that co-occur is not strictly compatible with the
evolutionary gene or replicator concept because the words are not necessarily adjacent to each
other or “linked” in genetical terms. For example in one case illustrated the words “James”,
“Smith” and “Nazi” were not necessarily found side-by-side. The relationship between structure
and adaptive function then was many to one. (If you are wondering how that is possible, an
evolutionist would explain that such co-occurrence without linkage is possible because in
17
addition to their ecological effects, separated sequences can have social effects on each other i.e.
have positively or negatively non-additive effects on fitness due to their co-occurrence.
Moreover, with positively non-additive fitness effects, an evolutionary geneticists would tend to
expect them to evolve towards becoming linked to make it less likely that they become
separated). Secondly, in some cases the adaptive effect of the terms on the post could have been
the result of someone saying “James Smith is a Nazi” but in other cases it could have been the
result of someone saying “How dare you call James Smith a Nazi” (an issue which the authors
subsequently addressed, Best and Pocklington 1999). The relationship then was-many-to-one. In
other work, by grouping posts similar in their term-subspace contents into quasi-species, Best
(1997) was also able to show that similar posts tend to be related by descent (come from the
same thread) and that competition among quasi-species is more intense the closer their
ecological niche.
Alternative terms in evolutionary social sciences
I used the Pocklington and Best work to make the point that despite the same structurefunction
disjunction problem as in genomes, the concept of a meme, like that of a gene, can and
has historically been used to do useful scientific work. It is not the only one to have done so
however. A lot of terms other than memes have been historically, and are currently in use in
different social scientific disciplines for the ‘iss and oughts’ normally expressed linguistically
and which evolve culturally. Mesoudi et. al. (2004, 2006), paraphrasing Boyd and Richerson
(1985) and Richerson and Boyd (2005) use information - “information such as knowledge,
beliefs, and values that is inherited through social learning and expressed in behavior and
artifacts” or “information capable of affecting individuals’ behaviour that they acquire from other
members of their species through teaching, imitation, and other forms of social transmission.”
18
Sociologists tend to refer to “norms and values”, linguists to “rules”, institutional economists and
organization theorists to “habits”, “routines”, and “competencies”. Anthropologists once liked to
talk about “folkways and mores” but today more often say “traditions” (albeit some students of
animal behavior like to call animal cultures “traditions” to distinguish them from human culture).
Archaeological speak is quite varied - “techniques”, “design elements”, “traits” and “traditions”
for example are fairly common. In science studies they speak of concepts, theories and methods
although sometimes more inclusive entities such as “paradigms” and “research programmes” are
investigated.
It is not necessary, nor is it likely, that those doing cultural evolutionary research in such
a wide variety of social science disciplines should or will abandon the conventional terminology
used descriptively for their particular sphere of culture in favour of memes and memeplexes.
Whatever terminology is used, it is important that the sense not only of knowledge or
expectations but also of instructions be included. Both biology and culture include not only
knowledge or expectations about the world but also instructions about how to act in it, how to
deal with it. As noted previously however, talk of memes has been commonly found useful for
communicative purposes in this wild west of cultural evolution when the discourse is
interdisciplinary.
Memes and social learning mechanisms
Most social communication in animals, and perhaps in people as well, is not about social
learning in the sense relevant to culture i.e. about Boyd and Richerson’s (1985) “second
inheritance system” that creates similarity between individuals. More often it is about what is
“accomplished” rather than what is “conveyed” as ethologists and behavioural ecologists put it
(Owings and Morton 1998). When it is about what is “conveyed” however, it is not obvious that
19
it is appropriate to talk about memes irrespective of the social learning mechanism involved.
With appropriate caveats about units as discussed above, the meme concept can be useful when
the mechanism involved is linguistic because information is obviously being transmitted. But
what about other mechanisms? Psychologists have proposed a variety of mechanisms that could
be involved in social learning but we will consider only two other broad categories (for recent
reviews see Zentall 2006, and the articles in Hurley and Chater EDS. V. I 2005).
Before doing so it will be useful to note that we assume here that all learning is cognitive
- not in any sense related to conscious, but in the sense that it creates information
(knowledge/expectations and instructions) in the brains of individuals. Demonstrations of this go
back to experiments in place learning performed by the cognitive behaviourist Edward Chance
Tolman in the 1940's (e.g. Tolman, Ritchie and Kalish 1946). For example, in a four arm maze, if
you first teach rats starting from the south to get food in the west box, then test them starting
from the north box and they still get the food, then they obviously have not learned only a
behaviour (i.e. turning left which would have lead them to go east when coming from the north),
but instead they have learned to know, or more accurately to expect, where the food will be. This
can be most parsimoniously explained with the concept of a programme - what was increased in
frequency by the reward in learning was not a behaviour, turning left, but the execution of a
programme which minimally contained the instructions “If coming from the south, go left; if
coming from the north go right” (“south” and “north” here being simply markers for any
discriminable features of the environment such as a scratch on the floor or a light overhead). The
full content of the programme did not become obvious however until the testing phase. (Of
course no programme includes all conceivable ways of achieving a result which would imply a
naive form of pre-scientific teleology. For example, if we cemented up the west arm for testing,
20
the food would not be acquired. In that sense there is no such thing as learning to achieve a goal
without qualification.) However, most behaviour in animals, even those with quite simple
nervous systems, probably involves the execution of complex programmes including routines,
sub-routines and so on. So we assume that in social learning, the situation is similar, that social
learning is cognitive.
Now consider a case of social learning taking place by individual learning mechanisms
e.g. by instrumental or operant conditioning. An episode of social learning between ego and alter
with the direction ego to alter may actually be rather complicated. It may be in the
(reinforcement) interest of ego to shape alter’s behaviour to be similar to its own (in which case
the interaction could be neutral or predatory with respect to alter), or it may be in the
(reinforcement) interest of alter to have its behaviour shaped to be similar to ego’s (in which case
the interaction could be neutral or parasitic with respect to ego), or both (i.e. be in their mutual
interests). For example, the former was the implicit assumption in the first theory of social
learning by individual learning mechanisms - the theory of matched-dependent learning (Miller
and Dollard 1941, 1962) in which ego reinforces alter’s behaviour for matching its own. On the
other hand, the second is the assumption in Boyd and Richerson’s theory of the evolution of
cultural transmission according to which learning from others is favoured if learning individually
is error prone or expensive, and environments are neither too variable (which would favour
individual learning) nor too stable (which would favour genetic transmission) (Boyd and
Richerson 1985, Richerson and Boyd 2005: Chpt. 4) . Notice that in the first kind of case, ego
must learn individually how to emit “signals” (cues which induce the expression of some
particular behaviour in alter, warnings which suppress it, and/or rewards and punishments which
select among alter’s behaviours), in order to shape (as the psychologists call it) alter’s knowledge
21
and behaviour to be similar to its own. Because ego knows how and is able to do something does
not mean that it knows how to shape another to know and do similarly. In the second kind of case
on the other hand, alter must learn individually how to have its behaviour shaped by ego - i.e. to
receive and correctly interpret “signals” (cues, warnings, rewards and punishments) emitted by
ego in order to have its behaviour rendered similar to ego’s. In both cases, “signals” are in
quotation marks because they are signals in the ordinary language sense only to one, not both of
the parties. In the third kind of case in which behaving similarly is in their mutual interests, they
are signals to both and both are learning.
Given the assumption that all learning, including the social, is cognitive, in all three cases
information has been created in one or more brains. The important point however is that in the
first case information has been created in the brain of ego only; in the second in the brain of alter
only; and in the third, information has been created in both, but it is completely different
information. There would be no reason to believe that alter who has been shaped socially by such
a mechanism would thereby be capable of shaping another in turn. It would have to learn
individually to do so. In essence, these are social-psychological and not cultural processes. There
is no meaningful sense in which information has been transferred. Memetic language to describe
them would therefore be inappropriate.
While social learning by linguistic instruction clearly transmits information and social
learning by individual learning mechanisms clearly does not, the case of social learning by
observation (in any sensory modality), ‘learning to do by seeing it done’, variously called
observational learning, imitation or true imitation, the situation is unclear. The possibility or even
the probability of imitation arises particularly when the learning is latent (the learner just
observes, not performs in the learning situation) and when a long time lag exist between
22
observation and performance. According to Zentall’s review (2006), the evidence for imitation is
rather good for birds and primates. However most ethologists and behavioural ecologists would
at least extend the primates to mammals in general (including Cetaceans for example, Rendell
and Whitehead 2001) and even to fish and insects when the behaviour involved is female mate
choice (for an overview see Dugatkin 2000). The copying of mate choice among females is a
long established feature of lek mating systems for example. It whatever group, it would seem
that the only obvious mechanism to explain imitation must include an analogue of reverse
translation. One would have to “reverse translate” the actions of another into cognitive
descriptions of the behaviour, and then use the latter to guide one’s own actions. In humans of
course, this reverse translation would normally be into language, whether silent or overt. In
animals what form these descriptions would take is unclear (as is the nature of cognition in
animal learning in general for that matter). However, in either case it is reasonable to suggest, as
many have, that the existence of mirror neurons (discovered initially by Gallese et. al. 1996;
Rizzolatti et. al. 1996) - neurons which “mirror” the actions of another by being active both when
an action is observed and when it is performed may be relevant to understanding this form of
cognition. In either the human or the animal case, the issue of whether or not information has
been transferred and hence whether memetic language would be appropriate however, comes
down to the question of whether the resulting information in the two individuals’ heads is “the
same”. I do not know of any way of deciding that although ultimately neurophysiology may
provide the answer.
It is worth noting that the teaching of physical skills in humans (say a child to tie a
shoelace) is often accomplished by combining all three of these mechanisms - we demonstrate
the behaviour, shape their behaviour (perhaps guiding their fingers and offering rewards for a
23
step accomplished correctly declaring “that’s it”), all the while emitting verbal instructions such
as “now bring the top one down, back, then up again and pull on both ends”.
Conclusion
The subtitle of this article might better have been put as “memes if useful, but not
necessarily memes” (but it does not sound as good!) . Objections to the meme concept that there
are discrete units of symbolically-encoded biological information which evolve, - genes, but not
of cultural information - memes, are not persuasive as has been discussed because the same
problem with respect to units exists in both realms. Despite this complexity, like the gene
concept historically, the meme concept can be shown to have done useful jobs of scientific work.
Although it is unlikely that the descriptive terminology utilized in a variety of social sciences
including their evolutionary versions will be abandoned in favour of memes, memetic
terminology may be particularly useful in interdisciplinary communication. Because it implies
the transmission of information, memetic talk may be most appropriate when social learning is
by linguistic means, least appropriate when it is by individual learning mechanisms, but the
situation is unclear when social learning is by observation i.e. imitation, where a reverse
translation-like mechanism would seem to be required.
Finally, I note that memetics in particular, and cultural evolution in general, has done
little to address some of the most interesting problems in the general theory of selection
processes today - such as those involving the relationship between evolution, development,
ecology and heredity. That needs to change.
REFERENCES
Aunger, R. (ed.) 2000. Darwinizing Culture: The Status of Memetics as a Science. Oxford
University Press.
24
Benzer, S. 1957. “The elementary units of heredity.” Pp. 70-93 in W.D. McElroy and B. Glass
(eds.) The Chemical Basis of Heredity. Johns Hopkins Press.
Best, M.L. and R. Pocklington 1999. “Meaning as use: Transmission fidelity and evolution in
NetNews.” Journal of Theoretical Biology 196: 389-395.
Best, M.L. 1997. “Models for interacting populations of memes: competition and niche
behavior.” Journal of Memetics 1.
Beurton, P.J., R. Falk and H. Rheinberger. 2000. The Concept of the Gene in Development and
Evolution: Historical and Epistemological Perspectives. Cambridge University Press.
Blackmore, S. 1999. The Meme Machine. Oxford University Press.
Blute, M. 2005. “If the Genome Isn’t a God-like Ghost in the Machine, Then What Is It?”
Biology and Philosophy 20: 401-407.
Boyd, R. and P.J. Richerson. 1985. Culture and the Evolutionary Process. Chicago: University of
Chicago Press.
Burt, A. and R. Trivers. 2006. Genes in Conflict: The Biology of Selfish Genetic Elements. The
Belknap Press of Harvard University Press.
Callebaut, W. and D. Rasskin-Gutman (EDS.) 2006. Modularity: Understanding the
Development and Evolution of Natural Complex Systems. The MIT Press.
Cavalli-Sforza, L.L., and M.W. Feldman. 1981. Cultural Transmission and Evolution: A
Quantitative Approach. Princeton University Press.
Covington, M.A., H. Congzhou, C. Brown, L. Naci, J.T. McClain, B.S. Fjorbak, J. Semple and J.
Brown. 2005. “Schizophrenia and the structure of language: the linguists’s view.”
Schizophrenia Research 77: 85-98.
Crick, F.H.C. 1968. “The origin of the genetic code.” Journal of Molecular Biology 38(3): 367-
25
79.
Dawkins, R. 1976. The Selfish Gene. Oxford: Oxford University Press.
Dawkins, R. 1982. The Extended Phenotype: The Gene as the Unit of Selection. W. H. Freeman
and Company Limited.
Dawkins, R. 1995. River Out of Eden. Harper Collins.
Dawkins, R. 2003. A Devil’s Chaplain. Weidenfeld & Nicolson..
Dugatkin, L.A. 2000. The Imitation Factor: Evolution Beyond the Gene. The Free Press.
Ehrlich, P.R. 2002. Human Natures: Genes, Cultures, and the Human Prospect. Penguin
Gallese, V., L. Fadiga, L. Fogassi and G. Rizzolatti. 1996. “Action recognition in the premotor
cortex.” Brain 119:593-609.
Griffiths, P.E. 2002. “Lost: one gene concept. Reward to finder”. Biology and Philosophy 17:
271-283.
Hickey, D.A. and M.R. Rose 1988. “The role of gene transfer in the evolution of eukaryotic sex.”
In R. E. Michod and B. R. Levin (Eds.) The Evolution of Sex: An Examination of Current
Ideas. Sinauer Associates Inc.
Holmes, F. L. 2000. “Seymour Benzer and the definition of the gene.” Pp. 115-155 in P.J.
Beurton, R. Falk and H. Rheinberger. The Concept of the Gene in Development and
Evolution: Historical and Epistemological Perspectives. Cambridge University Press.
Hull, D.L. 1988. Science as a Process. An Evolutionary Account of the Social and Conceptual
Development of Science. Chicago: University of Chicago Press.
Hurley, S. And Chater, N. 2005. Perspectives on Imitation: From Neuroscience to Social
Science. V I Mechanisms of Imitation and Imitation in Animals, VII Imitation, Human
Development, and Culture. MIT Press.
26
Lipo, C.P., M.J. O’Brien, M. Collard and S.J. Shennan. 2006. Mapping Our Ancestors:
Phylogenetic Approaches in Anthropology and Prehistory. AldineTransaction.
Maynard Smith, J. and N. Warren. 1982. “Models of Cultural and Genetic Change.” Evolution
36:620-27.
Maynard Smith, J., and E. Szathmary. 1995. The Major Transitions in Evolution. W.H. Freeman
and Company Limited.
Mesoudi, A., A. Whiten and K.N. Laland. 2004. “Is human cultural evolution Darwinian?
Evidence reviewed from the perspective of The Origin of Species.” Evolution 58:1-11.
Mesoudi, A., A. Whiten and K.N. Laland. 2006. “Towards a unified science of cultural
evolution.” Behavioral and Brain Sciences 29:329-383.
Miller, N.E. and J. Dollard 1941, 1962. Social Learning and Imitation. Yale University
Press.
Moran, L. A. 2007. “What is a gene?”
http://sandwalk.blogspot.com/2007/01/what-isgene.
html
Owings, D.H. and E.S. Morton. 1998. Animal Vocal Communication: A New Approach.
Cambridge University Press.
Pocklington, R. And M.L. Best. 1997. “Cultural evolution and the units of selection in replicating
texts.” Journal of Theoretical Biology 188:79-87.
Rendell, L. and H. Whitehead. 2001. “Culture in whales and dolphins.” Behavioral and Brain
Sciences 24:309-382.
Rheinberger, H.J. and S. Muller-Willie. 2004. “Gene.” Stanford Encyclopedia of Philosophy.
Richerson, P.J. and R. Boyd. 2005. Not by Genes Alone: How Culture Transformed Human
Evolution. The University of Chicago Press.
27
Rizzolatti, G., L. Fadiga, L. Fogassi and V. Gallese. 1996. “Premotor cortex and the recognition
of motor actions.” Cognitive Brain Research 3:131-141.
Ruse, M. Darwinism and its Discontents. 2006. Cambridge Univesity Press.
Sarkar, S. 1996a. “Biological information: A skeptical look at some central dogmas of molecular
biology.” Pp. 187-231 in S. Sarkar (ed.) The Philosophy and History of Molecular
Biology: New Perspectives. V 183 in Boston Studies in the Philosophy of Science.
Kluwer Academic Publishers.
Sarkar, S. 1996b. “Decoding ‘coding’ - Information and DNA.” BioScience 46:857-864.
Stanovich, K.E. 2004. The Robot’s Rebellion: Finding Meaning in the Age of Darwin. The
University of Chicago Press.
Stotz, K., P.E. Griffiths, & R.D. Knight. 2004. “How biologists conceptualize genes: an
empirical study” Studies in the History and Philosophy of Biological and Biomedical
Sciences 35:647-673.
Tolman, E.C., B.F. Ritchie and D. Kalish. 1946. “Studies in spatial learning II Place learning
versus response learning.” Journal of Experimental Psychology 36:221-226.
Van Driem, G. 2001. Languages of the Himalayas: An Ethnolinguistic Handbook of the Greater
Himalayan Region Containing an Introduction to the Symbiotic Theory of Language.V I.
Brill.
Whitehouse, H.L.K. 1965, 1969, 1973. Towards an Understanding of the Mechanism of
Heredity. Edward Arnold.
Williams, G.C. 1966. Adaptation and Natural Selection: A Critique of Some Current
Evolutionary Thought. Princeton University Press.
Williams, G.C. 1992. Natural Selection: Domains, Levels, and Challenges. Oxford University
28
Press.
Zentall, T.R. 2006. “Imitation: definitions, evidence and mechanisms.” Animal Cognition 9:335-
353.
29
![HIGHLY RECOMMEND VERSION TO LISTEN TO : [ Fast ] : The Quickening : Are You An Innie Or An Outie [mp3]](Gmail%20-%20feb%2017%2020%2017%20--%20you%20know%20is%20a%20very%20special%20day%20%202%20+%201%20+%207%20+%202%20+%200%20+%201%20+%207%20=%2020%20which%20is%202+0%20=%20[%202%20]_files/p2qUrx0egyeKkxSHMb3dc3rD2yXD7XqDq91aJ4a_xA9iPpvVsYIgR7Q1w6WO.jpg)
![[DIR]](Gmail%20-%20feb%2017%2020%2017%20--%20you%20know%20is%20a%20very%20special%20day%20%202%20+%201%20+%207%20+%202%20+%200%20+%201%20+%207%20=%2020%20which%20is%202+0%20=%20[%202%20]_files/U2iVpRJDQFe5x32RnIm6OVAsWdB-GU-jVen5LvftY6FRvcMdE1rM8i434ZFo.gif)
Print
Email
![[DIR]](Gmail%20-%20feb%2017%2020%2017%20--%20you%20know%20is%20a%20very%20special%20day%20%202%20+%201%20+%207%20+%202%20+%200%20+%201%20+%207%20=%2020%20which%20is%202+0%20=%20[%202%20]_files/3xQo2DvCX1N3BA6WePdYtQbk309iSwNWuxc78ho7TzEEzsWQiPnhmGtOJZzG.gif)
