Sunday, 4 September 2011

The claim that evolution doesn't require replicators

One of the most persistent criticisms of memetics by academic researchers in the "closely adjacent" field of gene-culture co-evolution is that memetics is based on the concept of replicators, which make a poor quality foundation for evolutionary theory.

Boyd and Richerson appear to have originated this critique. Here is my reply to them in my recent book on Memetics.

If you need to look up any of the references, they are here.

Evolution doesn't require replicators

One of the common complaints about memes is that memes are defined to be replicators, and replicators are not needed for evolution. For instance, Richerson, Boyd and Henrich, (2008) say - in a section titled: "Replicators are not necessary for cumulative adaptive cultural evolution ":
Much confusion about cultural evolution traces to Dawkins (1976, 1982) argument that discrete, accurately copied, long-lived “replicators” are necessary for cumulative, adaptive evolution. Dawkins argues that self-replicating entities are a requirement for cumulative evolution and must have the following characteristics:
  • Fidelity. The copying must be sufficiently accurate that even after a long chain of copies the replicator remains almost unchanged.
  • Fecundity. At least some varieties of the replicator must be capable of generating more than one copy of themselves.
  • Longevity. Replicators must survive long enough to affect their own rate of replication.
This argument has been repeated and elaborated by Dennett (1995), Blackmore (1999), Aunger (2002), among others, and has convinced many people that discrete, gene-like particles are a requirement for adaptive cultural evolution.

Dawkins made no such argument. Perhaps Richerson, Boyd and Henrich (2008) were unfamiliar with the terminology used by Dawkins. Dawkins (1982a) defined what he meant by "replicator" as follows:

I define a replicator as anything in the universe of which copies are made.

A very similar definition appears in Dawkins (1982c), and in Dawkins (2004b) he defined the term replicator by saying:

A replicator is anything of which copies are made.

Let's refer to this concept as a Dawkins-replicator. Now, Dawkins-replicators are definitely required for cumulative adaptive evolution - because without copying, nothing can possibly live. Now, of course you could argue (correctly) that Dawkins is misbehaving - by defining the ordinary English word "replicator" in a counter-intuitive manner. However, that seems to be a bit of a different complaint from the one being made here.

Dawkins did talk about fidelity, fecundity and longevity - but he didn't make anything like the argument that is being attributed to him here. Something Dawkins actually did say in The Extended Phenotype was:

This, then, is our candidate replicator. But a candidate should be regarded as an actual replicator only if it possesses some minimum degree of longevity/fecundity/fidelity (there may be trade-offs among the three).

I think this quote illustrates how far Dawkins was from giving the argument that Richerson, Boyd and Henrich apparently think that he made.

Others have made the same criticism as me - e.g. here is Sylvain Magne, writing in 2010:

Boyd and Richerson first introduce replicators as “material objects that are faithfully copied”. It certainly isn't Richard Dawkin's view who described replicators as “any entity in the universe of which copies are made”.

Sylvain does not seem too impressed with the resulting confusion:

It is very unfortunate that Boyd and Richerson misrepresent Richard's ideas in such a way. Because of this misunderstanding they then criticise memetics for the wrong reasons.

Additionally here is Susan Blackmore on the topic, from The Meme Machine, p.5:

A replicator is anything of which copies are made, including `active replicators' whose nature affects the chances of their being copied again.

There is no high fidelity copying here. Here is what she says on the topic in more detail:

This clarity is reflected in the new Oxford English Dictionary definition of “meme (mi:m), n. Biol. (shortened from mimeme ... that which is imitated, after GENE n.) An element of a culture that may be considered to be passed on by non-genetic means, esp. imitation”. Although many authors use widely differing definitions I suggest we stick to this simple one. Doing so avoids many problems. It also becomes clear why no existence proof is required. As long as we accept that people do, in fact, imitate each other, and that information of some kind is passed on when they do then, by definition, memes exist.

High fidelity copying is conspicuous by its absence. More from Susan:

This algorithm depends on something being copied, and Dawkins calls this the replicator. A replicator can therefore be defined as any unit of information which is copied with variations or errors, and whose nature influences its own probability of replication (Dawkins 1976).

Once again, no mention of high fidelity copying. One last one from Susan:

by definition, the information people copy is a replicator.
In the "Memes and Genes." video, Dawkins says:

Anything that is imitated is a form of replication.

I think this illustrates how Dawkins uses the word "replication" without intending any implication of high fidelity copying, supporting his definition of replication - which makes no mention of high-fidelity copying.

Here is Dawkins again, from The God Delusion (p.93):

Stitches in knitting, knots in ropes or fishing nets, origami folding patterns, useful tricks in carpentry or pottery: all can be reduced to discrete elements that really do have the opportunity to pass down an indefinite number of imitation generations without alteration. The details may wander idiosyncratically, but the essence passes down unmutated, and that is all that is needed for the analogy of memes with genes to work.
Dawkins (1983) denied that replicators were necessarily "particulate" and explicitly permitted 'blending' replicators:
A full science of Universal Darwinism might consider aspects of replicators transcending their detailed nature and the time-scale over which they are copied. For instance, the extent to which they are 'particulate' as opposed to 'blending' probably has a more important bearing on evolution than their detailed physical or molecular nature.
I think the idea of a version of memetics which depends on high-fidelity copying is essentially a daft one. As with genes, the mutation rate can be any value, and evolutionary theory just has to cope. I am not convinced that the memetic pioneers proposed such a thing - and in a number of places, they state their views pretty explicitly. Also, there actually is a valid point to be made about adaptive evolution requiring high fidelity information transmission. I think that the idea that memetics depends on high-fidelity copying is the result of an unsympathetic reading - or a misunderstanding. Of course, by using the term "replicator" - and giving it a counter-intuitive technical definition - it could be argued that Dawkins was setting himself up for this misunderstanding. However, there is a bit of a difference between a terminological misunderstanding and a scientific disagreement.



  1. Thanks for the post! I like to think about it in the following way: We may not be able to point to a physical structure that copies itself with high fidelity, but the fact that 7 billion people know the happy birthday song is evidence that some pattern of data is replicating itself across brains. I don't know what it looks like inside those brains, but I'm happy to call it a meme.

  2. I wasn't too impressed indeed :)
    Good points Tim !