Tuesday, 31 May 2011

Intracranial memetics and intercranial memetics

Intracranial memetics is the subfield of memetics that deals with the dynamics of ideas and memes inside a single mind. Memes and ideas compete with other memes and ideas inside minds. Intracranial memetics studies these dynamics - and the other forces on memes while they are inside minds.

Intercranial memetics is the subfield of memetics that deals with the dynamics of memes between minds. It covers interactions between memes and their environment while the memes are not inside minds.

The dynamics of intercranial memetics and intracranial memetics have some differences - because the environments involved are very different.

Intracranial memetics

Intracranial memetics deals with memes inside a single mind. The memes compete with other memes for space and attention. They also compete with protomemes and other ideas for resources. Intracranial memetics deals with these dynamics.

Intercranial memetics

Intercranial memetics ideals with memes when they are between minds. Many of the selective forces that act on memes do so while they are not inside human minds. Many memes can be copied without being inside minds. They can also be destroyed, damaged, put into storage and transported around. Intercranial memetics deals with these dynamics.


There's a nice Linus Pauling quote that refers to intracranial memetics:

The way to get good ideas, is to get lots of ideas and throw the bad ones away.

Martine Rothblatt's idea of "bemes" also relates to intracranial memetics.



Protomeme is an abbreviation of prototype meme. It refers to a meme which is under construction.

Before memes become fully-fledged memes they are protomemes. The definition of a meme involves social sharing - but before memes can be shared they must first be created - and often the creation takes place in a single mind. When a meme is under construction it is called a protomeme.

Protomemes are usually part of what I have called intracranial memetics - the area of memetics that deals with memes when they are inside human brains.

Not all memes make it into the annals of history. Some protomemes go on to become failed memes.

As far as I know, the term "protomeme" was first used by Liane Gabora in 1997.


Blooming marketing

Memetics has access to a plethora of negative words to describe the spread of ideas.

"Viral", "contagion" and "epidemic" are some of the most commonly-used terms. These draw on the language of epidemiology. There are also words to describe rapid growth sometimes exhibited by these systems: "explosion", "boom", "ignition" and "wildfire". These terms are associated with fires and explosions.

Unfortunately, a lot of these terms are pretty negative. This is unfortunate - since we know that ideas were - on average - positive among our ancestors - since humans have idea-collecting and spreading adaptations.

Marketers would probably prefer not to use such negative terms. After all, they are typically trying to hook consumers up with producers in win-win deals - and not trying to infect them with some kind of deadly plague.

So: what positive terms are there out there? Not so many, alas. After surveying the positive terms for growth: "branching", "budding", "sprouting", the most appropriate positive term I managed to find was "bloom" - as in "algal bloom".

So, perhaps in the future, positive marketing campaigns will bloom - and then bear fruit.

Thursday, 26 May 2011

Tuesday, 17 May 2011

Tower of optimisation

In Daniel Dennett's 1995 book "Darwin's Dangerous Idea" he describes a concept he calls the Tower of Generate-and-Test. This is a kind of model of the evolution of intelligent agents. A summary of the tower, starting from the bottom:
  • Darwinian creatures - use natural selection as the generate and test mechanism.
  • Skinnerian creatures - can learn by simple reinforcement learning.
  • Popperian creatures - have a world model, can virtualize sense data and test actions under simulation.
  • Gregorian creatures - tool makers including language and culture.
  • Scientific creatures - Dennett proposes that the scientific method warrants a further floor of the tower.

The model has been extended by Alan Winfield to include a final category:

  • Walterian creatures - artificial, engineered creatures, named after W. Grey Walter, robot inventor.

I think this addition is fairly reasonable. I do think this tower needs to get back to its roots a bit more, though!

If you consider the class of all optimisation processes, you get a rather different picture. Describing each stage in terms of what it adds to the previous level of the tower - and starting from the bottom:

  • Stateless search - optimization with no memory - e.g.: random search.
  • Serial search - uses one agent - e.g.: Newton-Raphson.
  • Parallel search - uses multiple agents - e.g.: simulated-annealing.
  • Splitting - uses agents that can divide and reproduce - e.g.: a simple asexual genetic algorithm.
  • Merging - uses agents that can merge together - e.g. sexual reproduction and parasitism.
  • Learning - uses evolving agents that can additionally learn.
  • Virtualisation - uses agents that can perform most evaluations under simulation.
  • Culture - uses agents that have developed cultural transmission.
  • Artefact symbiosis - the memes start to build tools, minds and bodies for themselves.
  • Genetic engineering - the agents apply their tools to their own germ-line.
  • All-engineered - uses entirely engineered agents - e.g.: machine intelligence and robots.
At the lowest level, the tower is neatly defined - but higher up, it is more like identifying what the major transitions in evolution are. Various steps which seemed fairly important in our world could be interspersed - for example communal living - and writing. For our ancestors, communal living allowed specialization to get going, and writing also provided a major boost. However, these stages do not really seem critical or universal enough to put in this list.

Monday, 16 May 2011


Phylomemetics is the name for the study of the historical evolutionary relationships between groups of cultural entities. The results of phylomemetic studies are often expressed partly in the form of phylomemetic trees.

Phylomemetics is named by analogy with phylogenetics - which it closely resembles.

Phylomemetics typically suffers from more problems derived from horizontal gene transfer than phylogenetics does - but neat trees can nontheless frequently be constructed.

The image to the right shows the relationships between Indonesian folk songs.

Phylogenetic relationships are often promoted as being some of the best evidence for organic evolution. Similarly, phylomemetic analysis provides excellent evidence for cultural evolution. One area where evidence is abundant comes from the evidence supporting the so-called "tree model" of languages.

Phylomemetic relationships between languages were known back in the 1850s - before Darwin wrote The Origin. Darwin actually used the descent with modification of languages to explain his theory of organic evolution. Back then cultural evolution was more widely recognised as being correct than organic evolution was.

For a large and beautiful phylomemetic tree of Indonesian batik designs, see here.

Also, amateur trees of internet memes are now being constructed - e.g. see here.


Sunday, 15 May 2011

Response to Peter Richerson

A few comments on Peter Richerson's recent article: "Culture Is an ACTIVE Part of Biology". To start with a quote:

The specific idea that cultural variation can respond directly to selection, and then influence genes by the Baldwin Effect, is harder for many people to swallow. I reiterate that entertaining this idea is just to walk down Darwin’s straight and narrow path. It all boils down to heritable variation for fitness. Richard Dawkins followed this path in the Selfish Gene where he introduced the concept of memes. One of the problems with the meme concept as it evolved is that users of the term focused far too heavily on the selfish potential of memes. But I think it is near to undeniable that cultural variants are sometimes selected to become selfish patIthogens along the lines that Dawkins suggested.

Thumbs up to Richerson for endorsing memes!

It is true that meme enthusiasts typically focus on the down-side of memes. There is a reason for that. When meme-interests coincide with DNA-gene interests, it is not easy to tease them apart. So, to see where memes and DNA-genes differ, one must look for cases where there are fitness conflicts - where the two selective forces pull in different directions.

But then:

I don’t much care for the term “meme” for the same reason that Joe dislikes it. It implies a close analogy between genes and cultural variants.

Yes - there is a "close" relationship between genes and "cultural variants" - a.k.a. memes - for some value of "close". There are many similarities.

It also tends to imply that an inheritance system has to be quite gene-like to count as an entity that can behave in Darwinian ways. None of these things is necessarily true (Henrich et al., 2008).

"an inheritance system has to be quite gene-like to count as an entity that can behave in Darwinian ways" seems to be vague. The Henrich reference says:

Memeticists like Blackmore (1999) and Aunger (2002) believe cultural representations, or as they prefer, memes must be particulate for cumulative cultural change to occur.

Sue explicitly denies holding such a position in her review of Richerson's book Not By Genes Alone. Here's what she says:

The population approach, they say, does not imply that cultural evolution is analogous to genetic evolution; nor does it depend upon “discrete, faithfully replicating, genelike bits of information.” I quite agree, but then so would Dawkins and most other memeticists.


Nor do replicators have to be “discrete, faithfully replicating, genelike bits of information”. Dawkins long ago pointed out that the copying fidelity of most memes is very low, there is often no right way of deciding where one meme begins and another ends, and most memes do not appear to be particulate – themes later taken up by both Dennett (1995) and me (Blackmore 1999).

Epidemic threshold

The epidemic threshold refers to a threshold above which agents can spread explosively and cause epidemics. It is usually expressed in terms of the ratio between birth rate and death rate. If the ratio is larger than one, then an epidemic may result. If it is less than one then an epidemic becomes extremely unlikely.

The epidemic threshold is an important concept in epidemiology. It has previously been applied to cultural evolution - for example by Seth Godin in Unleashing the Ideavirus (p.77) - though he called it the magic-number.

Many factors influence the birth rates - including their level of virulence, their lifespan, whether they are defeated by attacks from the host immune system - and so on.

Being above the epidemic threshold does not guarantee that an epidemic will result. Statistical fluctuations may result in extinction before an epidemic forms. Being below the epidemic threshold makes an epidemic extremely unlikely, though.

The epidemic threshold is a pretty important idea in viral marketing. If your shareable item is below the threshold, and you get exponential decay - with negligibe viral effect. If it is above it, you could see exponential growth - as your marketing work is done for you by the masses.


Epidemic failure and success.

This diagram shows the output from a very simple model of an epidemic. Infected hosts are plotted against time, from a fixed initial "seed" population. Different plot lines reflect different birth and death rates for the parasite.

Economic significance

The result of this is typically a non-linear R.O.I. on the marketing budget:
  • Below the epidemic threshold the effects are so small that you might as well not have bothered.
  • Above the epidemic threshold you see explosive growth - then it is more likely that the marketing budget has been well spent.

Much the same model applies equally well to mutualists as well as pathogens.

Big seed marketing

A related idea is big seeding - which is another strategy for avoiding extinction.

Tipping points

Malcom Gladwell once wrote a book called The Tipping Point - which referred to essentially the same concept. For more about that, see our post on that topic.

Note that the term epidemic threshold is also - rather confusingly - used to describe a number of hosts which need to be infected before an outbreak is classified as an epidemic. This post is not about that concept.


Memetic hijacking

Memetic hijacking involves a meme becoming active in a host and overriding the influence of its DNA-genes. The idea is named after the hijacking of vehicles - where an agent who is not supposed to be in charge siezes control.

The term applies to memes that are deleterious - from the perspective of host genes.

Just as viruses hijack host cells, so memes hijack host brains.

As with viral infections, memetic hijacking can be temporary or permanent. The host's memetic immune system can sometimes reassert itself, or the host can remain permanently infected - a persistent memetic infection.

Memes that are deleterious to their hosts lie on a spectrum between near-neutral hitchhikers to full-blown hijackers.


  • Pornography - one of the more common and obvious sources of memetic hijacking;
  • Cults - cults are rich sources of negative memes;
  • Computer games - much computer game use appears to be memetic hijacking;
  • Causes - a modern secular way for memes to take resources from human hosts;
Extreme cases of memetic hijacking can result in people infected with necrotrophic memes (sometimes called memeoids).

Saturday, 14 May 2011

Search magnet

"Search magnet" is a conventional term in internet marketing. It refers to a co-meme which is effective at attracting the attention of those searching.

If using Hofstadter's terminology, a "search magnet" is a co-meme that attracts the attention of searchers to the scheme's bait. On the internet most searches are textual searches - so search magnets are often textual. Sometimes searches have a visual component - for example, where multiple items are presented to the user and they select one. So, we also have:

  • Text search magnets - uses commonly-searched-for keywords;

  • Image search magnets - looks visually interesting;

  • Audio search magnets - sounds interesting;

Usually the prime function of the search magnet co-meme is to attract attention. However, it can do double duty - and be combined with the bait, or other co-memes.

Case studies

  • Will it Blend is a classic use of search magnets.

    Blendtec's Will It Blend? viral videos combined many popular consumer electronics products with the Blendtec Total Blender in a novel way - by "blending" them. The show features the Blendtec founder, Tom Dickson, attempting to blend unusual items to help him show off the power of his blender.

    The full story is here - but for our purposes note that the result was a large number of short videos with names featuring popular consumer-electronics products.

    Dickson says that the campaign has been a great success for Blendtec:

    The videos were placed on the internet in early November. Within just a few short days, we had millions of views. The campaign took off almost instantly. We have definitely felt an impact in sales. Will it Blend has had an amazing impact to our commercial and our retail products.

  • Another example of the use of search magnets is Ray William Johnson - an internet comedian who is best known for his stratgey of memetic hitchhiking on viral videos. His YouTube channel has details. Here is a sample video:

Missed opportunities

By contrast, the OldSpice YouTube channel is probably a failed example of a search magnet campaign. The campaign may have succeeded in other ways - but they made hundreds of videos, and did a really weak job of sprinking search-friendly keywords into the video titles.

Sunday, 8 May 2011

Unleashing the Ideavirus (review)

My review of Seth Godin's book Unleashing the Ideavirus.

The tipping point


The idea of a tipping point is one that comes from physics. Adding a small amount of weight to a balanced object can cause it to suddenly and completely topple. The point at which it does so is known as the tipping point. References to the idea of a tipping point with reference to water buckets tipping over date back to the 1800s.

Chaos theory

Chaos theory has the closely-related concept of a catastrophe. This is dealt with by catastrophe theory. There is also the related concept of self-organised criticality. Before catastrophe theory came along catastrophes were treated as a form of runaway positive feedback.

Since chaos theory is so well established it would make a lot of sense to refer to "tipping points" as "catastrophe points" or "critical points". However, the term "catastrophe" has powerful and not always appropriate negative implications.


The "tipping point" phrase was first applied in a sociological context by Morton Grodzins who studied American neighborhoods in the 1950s and 1960s. He discovered that most of the white families remained in the neighborhood as long as the comparative number of black families remained very small. But, at a certain point, when "one too many" black families arrived, the remaining white families would move out en masse in a process known as white flight. Inspired by the physics term, he referred to that moment the "tipping point".


The term is a fairly general, but it appears to have an important application to memetics and cultural evolution.

The term is frequently applied with memes that "go critical". For viral content to explosively spread virally, people need to share it more frequently than they forget about it. For most wannabe-memes, they don't reach the required critical threshold - and are quickly forgotten about. However, other memes find themselves on the other side of the threshold - and spread virally - increasing in an exponential fashion. This effect is similar to what happens when a nuclear reaction goes critical. The critical threshold is now referred to as the epidemic threshold - a topic which I have a post about.

There are several possibilities here for memes to cross the threshold:

  • Mutation - One is that your meme needs to mutate. An example of this would be the geddan meme - which had to cross the divide between video-game glitch and popular dance before making the big time.

  • Hitchhiking - Another possibility is that it needs to find the right hitchhiking partner. An example of that is Keyboard Cat. The original keyboard cat video languished in obscurity - until it was used in a mashup by Brad O’Farrell - and then it exploded onto the internet - with thousands of copies being made, all of which used memetic hitchhiking with other viral content.

Memes can lie dormant for other reasons as well. Sometimes, they have not wandered into the right niche yet. Sometimes, they just need to find the right vector, who will get a proper infection started. Lastly, they may need environmental changes before they can thrive. Legal changes, changes in moral behaviour or the extinction of competing forms can all help a meme go over the tipping point.

One example of an environmental change is when there is a competitive shakeout in a contested market, usually leading to one winner. Initially the market is small, and several contenders aim to reap the rewards of dominating it. As time passes, one competitor draws ahead, and then users switch to their product or service in a rapid cascade.

We see this with VHS vs Betamax and HD-DVD vs Blu-ray, Facebook vs MySpace - and so on.

Malcom Gladwell's The Tipping Point

The idea of tipping points was popularised by Malcom Gladwell in his popular book, The Tipping Point - How Little Things Can Make a Big Difference.


Here's Meme Molly, explaining the The Tipping Point concept.

Here's a Guinness advert - illustrating Malcom Gladwell's point.

See also: Pot Noodle 'Tipping Pot' (Guinness Spoof)


Wednesday, 4 May 2011

Viral Video Chart

For internet meme enthusiasts, the Viral Video Chart offers a fine way of tracking trending YouTube videos.

It tracks shares of such videos on social media sites, and then builds lists of the most popular ones. There's also a search facility. It beats the native YouTube facilities in a number of areas. The site is well worth checking out if you not familiar with it - and are interested in viral videos.

URL: http://viralvideochart.com/. They also have a meme-blog and a social media blog.

Payloads and penetration

Following up on my description of Hofstadter's reduction of memeplexes into co-memes, I think we can identify some more useful co-memes.

To recap, Hofstadter proposed an interesting meme classification scheme, involving groups of memes known as schemes - which contain cooperating memes, which are now commonly called co-memes.

The common types of co-memes are usually considered to be:

  • bait - attractive incentive to adopt the meme;
  • hook - promoting replication - being "hooked" is all about passing it on;
  • threat - of punishment, to assist hook execution;
  • vaccime - innoculation to protect against rival memes.

I think this list can be usefully expanded upon.


Memetic hitchhiking notes that some kinds of content can be divided into two parts - a delivery mechanism, and a payload. The delivery mechanism contains all the elements designed to do the propagation, while the payload is the deliverable cargo - part whose transmission is the true purpose of the message. It is usually some form of advertising.


In organic biology, pathogens often contain mechanisms designed to bypass host immune defenses. The malaria parasite uses a mosquito's hypodermic penetration equipment to get inside its host. Many other pathogens rely on being eaten by their hosts - since the body's defenses in the digestive tract are weaker. They work rather like Trojan horses. For example many flukes get past their host's defenses in this way.

In the domain of culture, schemes may contain elements designed to lower host memetic immunity before infection takes place.

"Trust us" is a simple example. "I'm on your side" and "let me help" work similarly.

Sometimes a one-two punch technique is used - first the host is softened up with emotional imagery, and then the target message is delivered. So, many religions have a history of using young ladies to make their converts. "Booth babes" are a more modern example of a similar technique.

Reducing host immunity seems to be a concept which is mostly distinct from Hofstadter's concept of bait.

Immunosuppression is the most common technical term for reduction in the effectiveness of the host's immune system. Immune evasion is the most common technical term for pathogens evading the host's immune system. I want an umbrella term that includes both types of technique. Immune penetration - or just penetration for short - seems to be the most suitable term to me.


Schemes may also contain free-riding memes and memetic junk.

Agner Fog in Cultural selection

Another classification scheme has been proposed by Agner Fog in Cultural selection (1999):

  • Bait
  • Hook
  • Indoctrination
  • Protection against rival meme complexes
  • Reward and punishment
  • Taxation

Tuesday, 3 May 2011

Meme alleles

Q: When is a meme not a meme?

A: When it is an allele!

In the organic world "gene" refers to a locus on a chromosome, and "allele" refers to particular genetic sequence that can occurs at that locus.

However, not many use the term "allele". People can't pronounce it. They can't spell it. They don't know what it means. So - most of the time - instead of using the word "allele" they just use the word "gene" instead.

Some aspects of the conceptual distinction between genes and alleles carries across into cultural evolution. So, the idea of there being certain "slots" with different mutually exclusive instances of heritable information that fit into those "slots" applies in both organic and cultural evolution.

To give a cultural example consider the idea of a furry feline animal. That represents a kind of "slot" into which "cat", "chat", "katze", "kat" or "gato" might fit - depending on which language you speak. The conceptual category of: word for a feline animal corresponds to a gene, and the specific word instances used correspond to alleles.

Words in languages behave like this, but there are other examples: preferred religion, favourite colour, choice of operating system, favourite food - and so on.

Famous examples of cultural alleles are: Pot-AY-to, pot-AR-to and tom-AY-to, tom-AR-to.

Now we get to the point of this post.

Most memes are less like genes than they are like alleles!

In common usage, if someone is talking about a meme, rarely are they talking about the idea of a "slot". Most of the time, they are talking about a particular instance - about somthing that fits into a "slot".

If you think people say "gene" when they really mean "allele" a lot, take heart. That is nothing compared to what happens in cultural evolution!

There, nobody ever says "allele" - and hardly anyone ever says "allomeme" or "meme instance". It is almost always "meme" instead.

So much for the gene-meme analogy: memes are more like alleles!

Monday, 2 May 2011

Memetic linkage

Genetic linkage refers to the concept that alleles in close proximity to each other have an increased chance of being inherited together.

The concept of linkage applies well in memetics too - leading to the idea of memetic linkage - which is the idea that nearby memes are more likely to be inherited together as a result of their close proximity.

To give some examples:

Adjacent sentences in documents tend to become linked together - and quotations may take adjacent sentences with them. Much the same thing happens with audio and video. Images may juxtapose unrelated subjects - which then become linked together.

For example, consider the girl standing in front of the CN tower. Or consider the aeroplane and the tourist in the images on the right. These days, the girl is often found with the tower, and the plane is often found with the tourist. Proximity-based linkage based on common descent is the primary reason for this.

Edward Burnett Tylor's adhesions

Edward Burnett Tylor laid the groundwork for the idea of memetic linkage in his concept of an `adhesion'.

Tylor was interested in cultural traits that appeared together, whether one trait caused the other, whether they were transmitted together, or what was happening.

Tylor presented his work on adhesions in a paper entitled:

“On a Method of Investigating the Development of Institutions; Applied to the Laws of Marriage and Descent,” which was published in volume XXVIII (1889) of the Anthropological Institute's Journal.

Tylor’s paper presented a method of “social arithmetic,” or statistical analysis of the presence of and “adhesions” between "traits". "Adhesions" broadly corresponded to correlations while the "traits" he considered were customs and institutions. He sought to compare traits with "adhesions" - to find those that seemed to occur together more frequently than the laws of chance would suggest.

The traits Tylor considered included in-law avoidance, post-marital residence rules, teknonymy, levirate marriage, couvade syndrome, bride kidnapping and exogamy.

Linkage-driven migration

An idea associated with the idea of genetic linkage is that alleles that have some kind of functional dependence on each other tend to migrate towards each other, so that linkage effects are increased, and the chances of them being divided from one another are smaller.

There are also other reasons that cause related ideas to clump together. That may mean that this effect may not be trivial to tease out experimentally.

Quantifying linkage

In genetics linkage can be quantified by measuring how far the linked alleles are from one another on a chromosome - or by looking at descendants and calculating the chances that the alleles are inherited together.

In memetics, the second option is still possible - memes frequencies can be tracked such as gene frequencies can - but a predictive theory based on the original heritable information itself is not so easy. Memes do not line up neatly in a one-dimensional space where distances from one another can be calculated easily. In podcasts and videos, temporal linkage may well be more significant than spatial linkage is. The linkage of a word to its neighbours depends little on whether the word is on one edge of the page. The linkage-distance between two memes is a more ambiguous concept than the linkage-distance between two genes. Instead you typically have to clearly specify what metric you are measuring memetic linkage with.

Criticism from Richard Dawkins

Richard Dawkins apparently denied the existence of "memetic linkage" in his 2005 book "The God Delusion". He said:

We doctors call that kind of linkage linkage, and I shall say no more about it because memes don't have chromosomes, alleles or sexual recombination.
However, he denied the existence of sexual memetic recombiantion in the same sentence - and that is surely an even more obvious concept.

Memetic linkage doesn't depend on memes being strung together in linear strings. The concept may be interpreted more loosely as an association - and then it becomes pretty clear that the idea does transfer across into the cultural realm. There is literature on "cultural linkage disequilibrium" - from Liane Gabora (1996). Memetic linkage disequilibrium is probably a better term, though.

Memetic hitchhiking

Memetic linkage is an important concept in helping to explain how memetic hitchhiking works.