Read Darwin's Dangerous Idea Online
Authors: Daniel C. Dennett
[emphasis added] for which there is favorable or unfavorable selection bias
Times,
that it finds a crumb. To human beings, on the other hand, each meme equal to several or many times its rate of endogenous change." The vehicle is a potential friend or foe, bearing a gift that will enhance our powers importance of the separation between information and vehicle is even easier or a gift horse that will distract us, burden our memories, derange our to discern in the case of memes.5 The obvious problem noted by all is that it judgment.
is very unlikely—but not quite impossible—that there is a uniform "brain language" in which information is stored in different human brains, and this makes brains very different from chromosomes. Geneticists have recently 3. C
identified a chromosomal structure they call the
homeobox;
in spite of OULD THERE BE A SCIENCE OF MEMETICS?
differences, this structure is identifiable in widely separated species of animals—perhaps in them all—so it is very ancient, and it plays a central role
The scope of the undertaking strikes me as staggering. But more than
in embryological development. We may be startled at first to learn that a gene
this, if one accepts the evolutionary perspective, attempts to discuss
identified as playing a major role in eye development in the homeobox of
science (or any other sort of conceptual activity) become much more
mice has almost the same codon spelling as a gene dubbed (for its phenotypic
difficult, so difficult as to produce total paralysis.
effect)
eyeless
when it was identified in the homeobox of the fruitfly,
—DAVID HULL 1982, p. 299
Drosophila.
But we would be even more flabbergasted were we to discover that the brain-cell complex that stored the original meme for bifocals in
Memes are capable of instructing, not protein synthesis as genes do, but
Benjamin Franklin's brain was the same as, or very similar to, the brain-cell
behaviour. However, genes can do that too indirectly through protein
complex that is called upon today to store the meme for bifocals whenever
synthesis. On the other hand meme replication, by involving
any child in Asia, Africa, or Europe first learns about them—by reading
neurostructural modifications, is invariably associated with the induc-about them, seeing them on television, or noticing them on a parent's nose.
tion of protein synthesis.
What this reflection makes vivid is the fact that what is preserved and
—JUAN DELIUS 1991, p. 84
transmitted in cultural evolution is
informa-
This is all very enticing, but we have been glossing over a host of complications. I can hear a chorus of skepticism building in the wings. Remem-5 For a good discussion of the embattled relation between gene talk and molecule
talk,
see Waters 1990.
354 THE CRANES OF CULTURE
Could There Be a Science of Memetics?
355
Hon
—in a media-neutral, language-neutral sense. Thus the meme is primar-instead of just dutifully passing on their messages, correcting most of the ily a
semantic
classification, not a
syntactic
classification that might be typos as they go, brains seem to be designed to do just the opposite: to directly observable in "brain language" or natural language.
transform, invent, interpolate, censor, and generally mix up the "input" before In the case of genes, we are blessed by a gratifyingly strong alignment of yielding any "output." Isn't one of the hallmarks of cultural evolution and semantic and syntactic identity: there
is
a single genetic language, in which transmission the extraordinarily high rate of mutation and recombination? We meaning is (roughly) preserved across all species. Still, it is important to
seldom
pass on a meme unaltered, it seems, unless we are particularly literal-distinguish semantic types from syntactic types. In the Library of Babel we minded rote learners. (Are walking encyclopedias hidebound?) Moreover, as identify a set of syntactic text-variants as all falling into the
Moby Dick
Steven Pinker has stressed (personal communication), much of the mutation galaxy by virtue of what they tell us
about,
not their syntactic similarity.
that happens to memes—how much is not clear—is manifestly
directed
(Think of all the different translations of
Moby Dick
into other languages, and mutation: "Memes such as the theory of relativity are not the cumulative also the English abridgments, outlines, and study aids—to say nothing of the product of millions of
random
(undirected) mutations of some original idea, versions in film and other media!) Our interest in identifying and rebut each brain in the chain of production added huge dollops of value to the identifying genes over the evolutionary ages is similarly
primarily
because of product in a nonrandom way." Indeed, the whole power of minds as meme the uniformity of the phenotypic effects—what they are "about" (such as nests comes from what a biologist would call
lineage-crossing
or making hemoglobin, or eyes). Our ability to rely on their syntactic identi-anastomosis (the coming back together of separating gene-pools). As Gould fiability in DNA is a recent advance, and even when we cannot conceivably (1991a, p. 65) points out, "The basic topologies of biological and cultural avail ourselves of it (for instance, in deducing facts about genetic changes change are completely different. Biological evolution is a system of constant from what we can observe in the fossil record of species that have left no divergence without subsequent joining of branches. Lineages, once distinct, DNA for us to "read"), we can still confidently speak of the genes—the are separate forever. In human history, transmission across lineages is, information—that must have been preserved or transmitted.
perhaps, the major source of cultural change."
It is conceivable, but hardly likely and certainly not necessary, that we will Moreover, when memes come into contact with each other in a mind, they someday discover a striking identity between brain structures storing the have a marvelous capacity to become adjusted to each other, swiftly changing same information, allowing us to identify memes syntactically. Even if we their phenotypic effects to fit the circumstances—and it is the recipe for the encountered such an unlikely blessing, however, we should cling to the more new phenotype that then gets replicated when the mind broadcasts or abstract and fundamental concept of memes, since we already know that publishes the results of this mixing. For instance, my three-year-old meme transmission and storage can proceed indefinitely in noncerebral grandson, who loves construction machinery, recently blurted out a fine forms—in artifacts of every kind—that do not depend on a shared language mutation on a nursery rhyme: "Pop! goes the diesel." He didn't even notice of description. If ever there was "multimedia" transmission and transforma-what he had done, but I, to whom the phrase would never have occurred, tion of information, it is cultural transmission and transformation. So some of have now seen to it that this mutant meme gets replicated. As in the case of the varieties of reductionistic triumph we have come to expect in biology—
jokes discussed earlier, this modest moment of creativity is a mixture of discovering exactly how many different ways hemoglobin is "spelled" in all serendipity and appreciation, distributed over several minds, no one of which the species in the world, for instance—are almost certainly ruled out in any gets to claim the authorship of special creation. It is a sort of Lamarckian science of culture, notwithstanding the prophecies of a golden age of mind-replication of acquired characteristics, as Gould and others have suggested.6
reading one sometimes hears these days from the ideologues of neu-The very creativity and activity of human minds as temporary homes for roscience.
memes seems to guarantee that lines of descent are hopelessly muddled, and This would thwart only some kinds of memetic science, but isn't the that phenotypes (the "body designs" of memes) change so fast that there's no situation actually worse than that? Darwinian evolution, as we have seen, keeping track of the "natural kinds." Recall, from chapter depends on
very
high-fidelity copying—almost but not quite perfect copying, thanks to the exquisite proofreading and duplication machinery of the DNA-readers that accompany the DNA texts. Raise the mutation rate just a bit too 6. Usually, the "charge" that cultural evolution is Lamarckian is a deep confusion, as Hull high and evolution goes haywire; natural selection can no longer work to (1982) carefully points out, but in this version it is undeniable—though also not a guarantee fitness over the long run. Minds (or brains), on the other hand,
"charge." In particular, the entity that exhibits the Lamarckian talent of passing on an aren't much like photocopying machines at all. On the contrary, acquired characteristic is not the human agent, but the meme itself.
356 THE CRANES OF CULTURE
Could There Be a Science ofMemetics?
357
10 (p. 293), that species are invisible without a modicum of stasis, but reinvented a wheel, rediscovered a cultural "universal" that will appear, on its remember, too, that this is an epistemological, not a metaphysical, point: if own, in almost any cultural evolution. The more purely semantic our species weren't rather static, we couldn't
find out
and organize the facts principles of identification are—or, in other words, the less bound they are to needed to do certain kinds of science; that wouldn't show, however, that the particular forms of expression—the harder it is to trace descent with phenomena weren't governed by natural selection. Similarly, the conclusion confidence. (Remember that it was peculiarities in the particular form of here would be a pessimistic
epistemological
conclusion: even if memes
do
expression that gave Otto Neugebauer his crucial clue in deciphering the originate by a process of "descent with modification," our chances of mystery of the Greek translation of the Babylonian ephemeris in chapter 6.) cranking out a science that charts that descent are slim.
This is the same epistemological problem, in the science of culture, that Once the worry is put in that form, it points to what may seem to be a taxonomists confront when they try to sort out homology from analogy, partial solution. One of the most striking features of cultural evolution is the ancestral from derived characters, in cladistic analysis (Mark Ridley 1985).
ease, reliability, and confidence with which we
can
identify commonalities in Ideally, in the imagined field of cultural cladistics, one would want to find spite of the vast differences in underlying media. What do
Romeo and Juliet
"characters"—literally, alphabetic characters—that are functionally optional and (the film, let's say, of)
West Side Story
have in common (Dennett 1987b)?
choices within a huge class of possible alternatives. If we found whole Not a string of English characters, not even a sequence of propositions (in speeches by Tony and Maria that suspiciously replicated the words and English or French or German... translation). What is in common, of course, is phrases of Romeo and Juliet, we wouldn't need autobiographical clues from not a
syntactic
property or system of properties but a
semantic
property or Laurents, Robbins, or Bernstein. We wouldn't hesitate to declare that the system of properties: the story, not the text; the characters and their coincidence of the words was no coincidence; Design Space is too Vast to personalities, not their names and speeches. What we so readily identify as the make that credible.
same thing in both cases is the predicament that both William Shakespeare In general, however, we can't count on such discoveries in our attempts at a and Arthur Laurents (who wrote the book for
West Side Story)
want us to science of cultural evolution. Suppose, for instance, we want to argue that think about. So it is only at the level of
intentional objects,
once we have such institutions as agriculture or monarchy, or even such particular practices adopted the intentional stance, that we can describe these common as tattooing or shaking hands, descend from a common cultural ancestor properties.7 When we do adopt the stance, the sought-for common features instead of having been independently reinvented. There is a tradeoff. To the often stick out like sore thumbs.
extent that we have to go to quite abstract functional (or semantic) levels to Does this help? Yes, but we must be careful about a problem we have find our common features, we lose the capacity to tell homology from already identified in several different guises: the problem of how to tell analog)', descent from convergent evolution. This has always been tacitly plagiarism (or respectful borrowing) from convergent evolution. As Hull appreciated by students of culture, of course, quite independently of (1982, p. 300) points out, we do not want to consider two
identical
cultural Darwinian thinking. Consider what you can deduce from potsherds, for items as instances of the same
meme
unless they are related by descent. (The instance. Anthropologists looking for evidence of shared culture are, quite genes for octopus eyes are not the same genes as those for dolphin eyes, properly, more impressed by common idiosyncrasies of decorative style than however similar the eyes may appear.) This is apt to create a host of illusions, by common functional shapes. Or consider the fact that two widely separated or just undecidability, for cultural evolutionists whenever they attempt to cultures both used
boats;
this is no evidence at all of a shared cultural trace the memes for Good Tricks. The more abstract the level at which we heritage. If both cultures were to paint
eyes
on the bows of their boats, it identify the memes, the harder it is to tell convergent evolution from descent.
would be much more interesting, but still a rather obvious move in the game We happen to know, because they told us, that the creators of
West Side Story
of design. If both cultures were to paint, say,
blue hexagons
on the bows of (Arthur Laurents, Jerome Robbins, and Leonard Bernstein) got the idea from their boats, this would be telling indeed.