Sunday, 7 April 2013

Tim Lewens on memetics

Tim Lewens is one of the surviving critics of memetics. His attacks on memetics still seem to feature in encyclopaedia articles - and so perhaps are worth rebutting. Here he is in a 2013 article:

A second, and closely related, criticism of memetics draws on the fact that while genetic replication allows us to trace a token copy of a gene back to a single parent, ideas are rarely copied from a single source in a way that allows us to trace clear lineages. Perhaps you learned the tango from several teachers, and your style has been influenced by watching expert dancers. There is no clear single origin for your “tango” meme. Within the realm of biological evolution, an understanding of Mendel's laws has been important in explaining some aspects of evolutionary dynamics. Mendel's laws rely on an understanding of genes as discrete, transmitted units. But if token ideas can appear in an individual by virtue of that individual's exposure to several sources, then it is unlikely that anything close to Mendel's laws will be discovered within cultural evolution. Such an objection need not be fatal for theories of cultural evolution in general, as we shall see, but it does threaten the tight analogy memetics draws between ideas and genes.
There are a few points to be made here:

Tracing DNA genes backwards to a single parent is also often not possible - due to meiosis. Meiosis means that gene sections come from an exponentially-increasing number of ancestors as you go backwards in time - a situation that closely mirrors what happens if you try and trace the ancestry of memes. You can often trace memes back to a single ancestor in the previous generation. Similarly, you can often trace DNA genes back to a single ancestor in the previous generation. This hardly seems like a major difference.

A lineage is defined to be the set of descendants of some ancestor. This concept is equally applicable to both cultural and organic realms. The idea that you can't find "clear lineages" in cultural evolution is complete bunk.

Lewens invokes the idea of blending influences from multiple individuals. However, much inheritance in culture does not involve blending. Instead, it's simple copying from a single parent. Blending inheritance is not the rule. Tim's claim that: "ideas are rarely copied from a single source in a way that allows us to trace clear lineages" is wrong. Look at modern peer-to-peer networks, for instance. High fidelity copying with uni-parental inheritance everywhere. Such high-fidelity cultural copying is extremely common these days.

Also, much alleged blending inheritance turns out not to involve blending at all. My gut bacteria are a kind of "blend" of those of thousands of other people. However if you look closely at the genes involved, it's mostly simple uni-parental inheritance that's involved. Tango involves many cases that are like that. There are many behaviours copied when learning to tango - and often they are copied from specific individuals.

However, blending during cultural evolution does sometimes happen. The thing is, such blending takes place in the organic realm too. It is better thought of as blending selection rather than blending inheritance.

Many individuals per-generation may indeed influence tango performance. However, many individuals per generation may also influence a tasty moth's attractiveness to birds. Tasty moths come to resemble toxic moths because of their DNA - not because of cultural influences. The blending of influences from multiple models in a single generation is a ubiquitous phenomenon in organic evolution.

As for Mendel's laws:

  • Law of Segregation (The "First Law") - doesn't apply to most memes. Doesn't apply to most DNA genes either.
  • Law of Independent Assortment (The "Second Law") - technically wrong for DNA genes and memes - Mendel missed linkage.
Mendel's second law applies about equally to memes and genes. Linkage applies to genes and memes too. The first law doesn't apply to a lot of memes. However, it doesn't apply to a lot of DNA-based organisms either. Mendel's first law has narrow applicability. The laws have little to do with the link between genes and memes. The topic seems to be an irrelevant one.

After all of these points are considered, not too much remains of the original objection.

It's true that organic evolution features a phenomenon whereby single nucleotides are almost always represented in at least one parent in the previous generation - whereas that isn't true of cultural evolution. Indeed, cultural evolution doesn't really involve nucleotides at all...

It's true that cultural evolution is more likely to involve extrapolation, interpolation, medians - and other advanced types of recombination. Recombination in brains has some more possibilities than recombination inside cells does. Memetics doesn't deny this. Blending inheritance happens inside minds - and not inside cells - but... so what? Memes aren't like genes in every single respect. Nobody ever claimed that they were. Advanced types of recombination really have little impact on the link between memes and genes - which is based on both being small bits of inherited information that evolve.



  1. Such critisisms are strawman and tautological as Lewens is pointing out dysanalogies to something that is not the whole picture of genes, but a "eukaryotocentric" skewed judgement at best.

    "Tango" is a label (and a meme) that indexes two sequence of other memes as dance actions (and their variations). It would be more like a chromosome pair - in both cases it takes two to tango!

  2. Critics seem to think they get to choose which aspects of memes are supposedly like corresponding aspects of DNA genes.

    My perspective is that the people who are proposing that this relationship is a helpful one are those who get to explain exactly how it helps.

  3. Hmmm... been thinking about this one as it seems to be a fallacy that keeps cropping up: probably has some Latin name. Anyway, I'd like to call it the challenge to memetics via "anorexic analogy".