Sunday, 24 May 2020

Synopsis of my replicator critique

For most traditional meme enthusiasts, memes are "the new replicators" - or are practically defined in terms of the Dawkins replicator-vehicle framework. I'm reasonably convinced by the critics that this is an unsuitable foundation for a general purpose evolutionary framework.

I've gone into this in a video/essay Against Replicator Terminology and in my 2011 memetics book, but here's a handy synposis:

Part of the problem is terminology. Dawkins (1976) defined "replicator" as follows:

"A replicator is anything in the universe of which copies are made."

One issue (1) is the "replicator"-"replicatee" distinction. You would think the appropriate term would be "replicatee" - and "replicator" would be the thing doing the copying.

Another issue (2) is copying fidelity. The "replicate" term suggests high fidelity copying. The corresponding term without that implication is "reproduce". A replica is normally a type of high fidelity reproduction. The Dawkins definition steamrollers over this distinction.

These are terminology criticisms - but my last critique is scientific - rather than just being about what words refer to which meanings. The "replicator" terminology suggests a modeling framework in which two inheritied entities are either idenitical or different. It represents a kind of binary view of inherited information. That works quite well for advanced systems of herediy because those have evolved to be digital. Nucleic acid and language are both basically digital systems which involve and use redundancy and error correction to avoid contamination by noise. However, not all systems involving inherited information are like this. In particular some of the systems I am interested in are those which exhibit Darwinian family trees in a more-or-less analog media. For example, electrical discharges, propagating cracks and fractal drainage patterns. These exhibit copying (of position and other attribuutes) with variation filtered by the environment, and a good number of Darwinian models apply to these systems. However the replicator-vehicle framework seems pretty irrelevant to their study. This consideration has the effect of displacing replicators from the foundation of Darwinism. They are still applicable to more advanced systems with error detection and correction, (but see points 1 and 2). As students of cultural evolution sometimes point out, not all cultural inheritance involves error-corrected systems such as language.

My personal preference is to just grant this point to critics of memetics and move on. The "m" word is still pretty useful, replicator terminology aside.

Tuesday, 31 March 2020

Sylvain Magne: a new theoretical model

I've covered the work of Sylvain Magne here before, see:

Here are some updated thoughts from Sylvain about memes:

It is quite nice. I like what I would describe as the "information theory" perspective in these essays. However, I don't really agree with all of it. To go over some of the differences between our positions:

  • Sylvain likes and uses the "replicator" terminology, while I typically avoid it and think it is confusing.
  • Sylvain classifies varaints as identical or non-identical. IMO, that can work well for more digital systems, but isn't so useful for more analog ones.
  • Sylvain proposes that we divide evolving information systems into codes and readers. Readers classify and recognize codes. While readers are widespread for genes and memes I am not convinced that they are always present. It is often a useful idea - but "readers" seem non-fundamental to me.
  • Sylvain rejects memes inside brains. I like memes inside brains.
  • Sylvain proposes the term "transmemes" for memes that are routinely translated. For me that is practically all memes - so the terminology is not very useful.
Regarding brains: IMO, we ought to be able to agree that evolving systems include psychological ones - as well as organic and cultural ones. There is copying with variation and selection inside individual brains. That is where many ideas have sex. That is where many ideas are copied. There are lots of books and literature about within-brain Darwinism. Treating the brain as a black box, identifying it as a "reader" and then claiming that it doesn't contain memes is only one perspective. You could also open it up and consider how it works. I have a summary of the case for within-brain Darwinism here: Keeping Darwin in mind.

Regarding "replicator" terminology, I once explained my position in an essay: Against Replicator Terminology. The fight over the utility of the "replicator" term is now pretty well-trodden.


Saturday, 28 March 2020

Jason da Silva: When fear goes viral

Jason offers a Coronavirus spiel with a heavy side helping of memetics.

Wednesday, 11 December 2019

Developmental knobs

Development features a number of what are fairly widely called "developmental knobs". A "developmental knob" represents a scalar variable whiich affects development and which can be influenced, or "turned" by genes, the environment, or some combination of these factors. The knobs are implemented using a variety of means - including changing chemical concentrations, developmental delays or altered behaviour patterns. A few examples:

  • Reproduction/maintenance axis - controlled by the dietary energy restriction;
  • Muscle investment axis - controlled by exercise;
  • Hunger axis - controlled by hormones - e.g. grehlin;
  • Love axis - controlled by hormones - e.g. oxytocin;
  • r/K axis - controlled by resource availability;
Not all of these knobs are solely under genetic control. Many can be influenced by symbiotic visitors. Some gut microbes are known to promote ghrelin synthesis - and probably that helps stimulate their hosts to eat more food which in turn feeds the microbes. Plants attempt to manipulate the developmental systems of the animals that feed on them. An example is phytoestrogens - which are typically hormone-like compounds in pllants that bind to estrogen receptors in animals, blocking uptake of real estrogen and so sabotaging the sexual development of the animals involved.

Some memes are likely have access to some of these developmental knobs. They probably use them to divert host resources away from host reproduction - and into meme reproduction. Sterlizing their hosts is a common interest of many memes. That memes sterilize their hosts is very well documented. Getting a college degree appears to severely hamper reproductive output for many women. See The sterilization of females for more.

The r/K axis is a well-known developmental axis. One end of it favors rapid reporduction and the other favors lots of parental investment in each offspring. If memes could get hold of it, they could reduce host reproductive output, likely to their own benefit.

Another example is the "love" axis. Christian memes encourage their owners to love Jesus. That Jesus is a long-dead creature of myth and legend is apparently not a big deterrent. As far as I can tell, some Christians are genuinely in love with Jesus. That the associated memes benefit from creating this love seems likely.

Sunday, 13 October 2019


An article by Maarten Boudry proposes that the term "panmemetics" be used for the memetics visualized by Susan Blackmore - which is contrasted with Daniel Dennett's presentation in his recent book on cultural evolution. Over to the author:

in Dennett’s current account, in particular his use of Godfrey-Smith’s (2009) Darwinian spaces to make sense of the “de-Darwinizing” of culture and the conscious domestication of memes, moves away from panmemetics, perhaps more clearly than his earlier Darwin’s Dangerous Idea. Even though memes initially evolved by blind and unguided evolution, claims Dennett in his new book, many were gradually domesticated by their hosts, who became more reflective and self-conscious about them. As culture gradually moves in the direction of the “intelligent design” corner of the Darwinian space (top-down, foresighted, directed), the meme’s eye view loses traction and becomes less interesting. Let me try to spell out the differences between the two approaches, as I see them. First, unlike Dennett’s account in BBB, panmemetics leaves little or no room for human autonomy and creativity. Panmemetics suggests that we are not in control of our thoughts; if anything, the reverse is true. In the words of Robert Aunger: “Do we have thoughts, or do they have us?” (Aunger, 2002, loc. 120) A second and related problem with panmemetics is that it threatens to level all distinctions between harmful and useful cultural inventions. If the whole of human culture is seen as just a set of swarms of viruses bent on exploiting our brains as a breeding ground for their own reproduction, we no longer have any theoretical resources to distinguish between good and bad memes, between beautiful folk songs and annoying earworms, between science and superstition.
The distinction might be an interesting one, but I don't think I can stomach rechristening memetics, "panmemetics". I am not a big proponent of the "de-Darwinizing” of culture that Dennett proposes. My conception of Darwinism is big enough to include all cultural evolution. We can have a picture involving degrees of intelligent design without dragging the rather vague and subjective "Darwin"-based terminology into it. IMO, panmemetics is just memetics. Dennett's idea that the "Darwinism" fades out as the intelligent design fades in doesn't seem interesting enough to warrant a big terminological shift. We can just use the "Darwinian" term for the more general framework that includes ID.

Regarding the criticisms of panmemetics: I don't see much impact of memetics on the issue of human autonomy and creativity (as mentioned in the quote). It is true that memetics suggests that we are not in control of our thoughts - but surely that is hardly news to most psychologists. I don't agree that memetics "threatens to level all distinctions between harmful and useful cultural inventions". The most obvious distinction between "good" and "bad" memes is between meme interests and the interests of DNA genes of the nuclear DNA of the host. The host is likely to have more than one gene, and there could be more than one host, but these issues can be deal with by averaging - as in the "parliament of genes" concept.

The author raises the issue of personal host interests - and where they fit into the picture. My favored approach to this is to have three spheres involving Darwinian dynamics:

  • DNA gene evolution;
  • Meme evolution;
  • Psychological evolution;
The last one is the least well understood - but information is copied with variation and selection in the brain, and we can use Darwinian models to represent the resulting dynamics. Psychological evolution is more obviously multi-level - since copying takes place at the level of branching axon impulses as well as at the level of ideas, and probably at multiple other levels in between. Personal interests can then be discussed in terms of copied information patterns - using the same modeling framework as with memes and genes.

Maarten Boudry goes into some more details about what is wrong with "panmemetics" in another paper titled "Attack of the Memes". Overall, I'm not seeing it.

Another exhiibit from the latter paper:

Memes that are parasitic from the perspective of my genes may simply be the outcome of deliberate human choices. Talk of "selfish memes" and "rogue culture" can be misleading here, as if humans are the hapless victims of the ideas of modernity. These are not novel memetic purposes but distinctively human ones.
Well, maybe, but maybe not. Selfish memes / rogue memes are a possible and valid answer too. It depends on the circumstances. One must also ask where those humans got their preferences. Some come from genes, some come from memes and some come from psychological forces (e.g. alcohol addiction). Memes are on the list. They can influence host preferences via manipulation from within hosts minds. Because a human has some preference, that doesn't mean it didn't ultimately come from memes.


Wednesday, 9 October 2019

The Organism-Centered Approach to Cultural Evolution

I found a paper from 2015 that offers a reply to the criticism of mine, that cultural evolution in academia is "host centric". I summarize that criticism in my article "The host-centric approach to cultural evolution".

Ben Cullen also summarizes this criticism in his "Contagious Ideas" book. To quote from my review:

The third and fourth chapters mostly look at what Ben calls "American Cultural Selectionism". This is characterised by explaining culture in terms of human traits, and tracing their passage between humans in terms of "oblique" and "horizontal" transmission of those traits. Ben classified these models as being "inclusive phenotype" models - since they include traits encoded by genes and memes in the phenotypes of human individuals. Rather than featuring distinct cultural individuals, phenotypes and populations - like memetics does - the "inclusive phenotype" models combine cultural and genetic influences within one human phenotype - in what Ben patronisingly refers to as a "bio-cultural muddle".

Here is a paper that offers a reply. It classifies approaches to cultural evolution as "organism centric" and "meme centric" - and argues for the superiority of the former. Overall, it is as though Dawkins had never written The Selfish Gene - explaining why Hamilton, Trivers and others had shifted focus away from organisms and towards genes.

The article says:

Instead of offering a set of criteria, we will pose several questions that need to be answered in order for there to be a coherent MC. First, can there be multiple copies of a meme within an individual’s head? If so, how are neuronal states individuated into distinct memes?
The claim that these questions need to be answered for memetics to be coherent is a mistake. It is quite possible to treat the entire human brain as a black box and conly consider how memes are transmitted into and out of it, via, talking, listening, etc. Memetics doesn't depend in any way on implementation details of neuroscientific representation. It would be nice to know all the details of what went on inside brains, but it is perfectly possible to discuss meme frequencies without this. That is, by and large, what memetics has done, because all the neuroscientific details are not there yet.

Another question it is claimed that meme theorists need to answer is:

And for the spork example above, is this a case of fork and spoon memes reproducing and ļ¬nding themselves within the same utensil, instead of each occupying distinct utensils? Or is a spork a unique meme related to spoon and fork memes, but not an instance of either? These are just some of the sorts of questions that a MC needs to answer"
Those are alternative modelling scenarios within memetics. Researchers might choose between them depending on their research goals. However, memetics itself doesn't need to choose between them! It is the same situation as with genetics - the same section of DNA can be modelled as consisting of two or three genes, depending on what traits the researcher is interesting in studying. Genes can overlap, and that is just a fact of life, not a foundational challenge to genetics.

I've criticized host-centric cultural evolution as ignoring the dynamics of meme replication in machines, thereby missing the very possibility of a memetic takeover. Here is this paper's take on that. It claims that is actually an advantage to the organism-centric approach - because it makes things more palatable to social scientists:

If cultural evolutionists would embrace the MC andfocus on the number of memes (e.g. the number of pro-duced sporks), irrespective of whether these memes are actually adopted by humans, then this would decrease therelevance of cultural evolution for the social sciences andthe humanities. After all, these disciplines are mainly—if not solely—interested in culture in as much as culture affects the behavior and thinking of human beings

Cultural evolution within academia was developed by social scientists - and they went out of their way to make it more palatable to other social scientists, for example by stripping off biological references wherever possible, and exagerating the differences between cultural and traditional evolutionary theory. However, that is only a virtue if acceptance by existing social scientists is recognized as a valid aim. Existing social science failed to get to grips with evolutionary theory for 150 years. They typically think that "Social Darwinism" is a failed research program, and one which they do not wish to associate themselves with. It's not that pre-evolutionary social science is useless, it is that it lacked the firm foundation that evolutionary theory is now providing. Pandering to social scientists seems like a secondary goal to me. I favor going directly for correct and useful theories. If they seem revolutionary and foreign to the old guard, so be it.

The paper at least recognizes the issue of there being two rival approaches here. However, it offers a series of mischaracterisations of memetics and gives the wrong answer to the issue. It would be better to just say that organism-centric cultural evolution is holistic and meme-centric cultural evolution is reductionistic. For me, even that is too flattering. Part of the problem is the theory is not just organism-centric it is host-centric - and the only hosts considered are humans. Machine intelligence as a host of memes mostly isn't on the radar, and the possibility that "the Bible" might act like a cultural organism with its own largely independent genotype and phenotype is also largely neglected.


The Organism-Centered Approach to Cultural Evolution by Grant Ramsey

Wednesday, 18 September 2019

Geoffrey Miller on who benefits

Another passage from Geoffrey Miller's meme critique (see also the previous post) reads:

genes and memes are not the only alternatives as beneficiaries – there are institutions that can be treated as self-interested agents for purposes of economic, political, sociological, and cultural analysis
That seems like a mixture of levels of explanation. "Institutions" are a mix of genes, memes, gene products, meme products - and maybe some joint meme-gene products. The "products" are phenotypes. In biology, genotypes are what is inherited and phenotypes are everything that is derived from them. From this perspective, only the genes and memes form important lineages and can qualify as beneficiaries. The "products" are more transitory - they are not inherited from they affect gene and meme frequencies via selection, but don't "benefit" because they die without leaving any offspring - and "benefit" in evolutionary theory is measured in terms of fitness. Only genes, memes and other forms of inherited information really qualify.

Of course you can also talk coherently about larger-scale entities benefitting in evolution. An animal might be said to benefit by having offspring. However it is just a different level of explanation. It is quite compatible with the meme's / gene's eye view. It's a lot like saying that their genes benefitted - on average. A focus on low level beneficiaries doesn't somehow exclude higher levels. I think this critique of Miller's doesn't really go anywhere. Meme enthusiasts aren't somehow ignoring institutions. Institutions are composed of memes, genes and their products. You don't need to do fitness accounting on memes, genes and institutions. That would be counting things twice.

Geoffrey Miller on memes

Geoffrey Miller comes across as yet another evolutionary psychologist who is clueless about memes and cultural evolution. However, I recently found and read Geoffrey Miller's review of "The Meme Machine". It has its moments. However, here I'd like to quote a section from it and then explain the problem with it.

To make a strong case for memes evolving contrary to our genetic interests, Blackmore would have to show that most of our memes lower our sexual attractiveness. This seems unlikely to be the case, given that the classic examples of memes – songs, fashions, moral ideals, religious convictions – are adopted and advertised by young adults precisely for their sexual appeal. Also, the fact that young males invent and propagate many more memes than other demographic groups suggests that meme-spreading remains genetically adaptive: males try to attract multiple sexual partners through various artistic, musical, and ideological displays, while most females still invest much less in this sort of courtship effort.
This is, I think a straw man attack. The evidence that memes have historically been adaptive on average from the POV of human genes is pretty strong. Humans have genetic adaptations for language - such as a "babbling" stage of development which other great apes lack. We have also been spectacularly successful in a wide range of ecosystems. I can't think of any meme enthusiast that denies this.

The basic idea in memetics is not that memes are maladaptive on average, but that some memes are sometimes maladaptive. Copying machinery might print out helpful manuals, but it also prints out fake news and other deleterious nonsense. Yes, we have cars and bridges, but we also have the obesity epidemic, the smoking epidemic, the heroin epidemic, etc. It is not: "all memes are bad for your genes" or even "memes are bad for your genes on average" - instead, it is "some memes are bad for your genes".

Having argued that meme and gene interests have been somewhat aligned historically, this does not necessarily mean that that situation will continue indefinitely. If population density increases, and horizontal meme transfer becomes the dominant force, memes could switch their strategy to one more like the ebola virus - i.e. treat the host as an expendable bag of resources and burn through it as quickly as possible. If humans are not the only hosts of memes - and many memes are also propagated by intelligent machines - such dynamics could drive the human population down dramatically. No law of nature demands that meme interests will remain somewhat aligned with the interests of all of their hosts. There are examples in nature of pathogens that wipe out one of their host species. It happens more when there are multiple host species. Then each individual species becomes expenddable. Clearly, we should bear this possibility in mind. However, you really need to have a basic understanding of cultural evolution to even be aware that this sort of possibility exists.