Sunday, 17 August 2014

Cultural kin selection

A major breakthrough in evolutionary biology in the 1960s took the form of the development of a theory that could account for much cooperative behaviour in nature. That theory takes on new significance and importance when applied to cultural variation. In particular, nepotism, kinship, relatedness and kin selection all have direct parallels in cultural evolution.

Kin selection

Genetically related individuals frequently cooperate and behave altruistically towards one another. This phenomenon is modeled by a branch of evolutionary theory known as "kin selection". This models cases where organisms engage in behaviour that favours relatives over non-relatives. Kin selection explains parental care, nepotism, eusociality among the social insects, problematical adopted children - and many other phenomena. The original explanation of kin selection invoked shared genes. J. B. S. Haldane was one of the first evolutionary biologists to understand the idea in the 1950s. It was subsequently studied and modeled by William Hamilton in the 1960s. Hamilton said:

The existence of altruism in nature can be explained by thinking about the replication of genes. We need to descend to the level of the gene, rather than the individual, in order to see that the gene exists surrounded by copies of identical genes that exist in all its relatives - in particular in its close relatives, its siblings, who have a half chance of carrying a copy of that particular gene, its offspring, which also have a half chance, parents: a half-chance, cousins: one eighth, etc. Seeing this swarm of genes that exists around a particular one, we can then ask what is the behavior caused by this gene that is most likely to cause the propagation of this set of copies in the relatives around it.
A broadly similar argument applies to memes. This raises interesting possibilities for what we will be calling "cultural kin selection".

Cultural kin selection

Just as genetic kin can be expected to cooperate, so it seems reasonable to expect memetic kin to cooperate - on much the same grounds.

Our understanding of altruism needs to be augmented by considering the reproduction of memes. We need to descend to the level of the meme, in order to see that an individual meme is surrounded by copies of itself in the form of the meme's parents, offspring, siblings and cousins - its memetic kin. Seeing this swarm of memes that exists around a particular one, we can then ask which of the behaviors that could be promoted by this meme would be most likely to cause the propagation of the swarm of copies of itself that surround it.

That kin selection can be usefully applied in the cultural realm is an old idea. Boyd and Richerson discussed the idea in 1980. Paul Allison and Francis Heylighen noted it in 1992. Anthropologists had previously distinguished between "biological kinship" and "social kinship" (Hawkes, 1983) or between "natural kin" and "nurtural kin" (Watson, 1983) - but they mostly lacked a coherent theory about the evolutionary basis of these categories. Cultural kin selection helps to explain why these traditional anthropological categories are as useful as they are.

Suicide terrorism represents a good example of cultural kin selection. All the memes in the suicide bombers are extinguished, but copies of their memeplexes in other individuals are promoted by the publicity generated by their actions. Suicide terrorists believe that they are part of a brotherhood and that their actions help their relatives. This is not far from the truth - though the "brotherhood" is a cultural - not an organic one - and the bombing typically promotes associated memes - not organic genes. Cultural kin selection is involved many types of human social behaviour - including political, military, religious and professional groupings.

Parental resource allocation

Parental investments in offspring are one of the the most prominent manifestations of kin selection in the organic realm. These so not require particularly advanced cognition or much in the way of recognition of kin. Parents usually provide their offspring with more than just a genetic inheritance. They often provide them with material resources, to help their offspring get off to a good start in life. Parental investment typically involves one or more of these two types of contribution:
  • Resource boluses - allocated at birth;
  • Resource trickles - supplied over an extended period of time;
Parents sometimes provide a resource bolus - to help give their offspring a good start in life. For example, this can take the form of albumen in a large egg, or stored fats in a large nut or seed.

Another strategy is to provide a resource trickle over a more extended period of time. This requires an extended association between the parent and their offspring after it is born. Such extended relationships do exist - but they are not that common: many organisms simply abandon their offspring. Trickle feeding is the approach taken by strawberry plants, for example. They reproduce sexually using seeds - but also employ vegetative reproduction - using "runners". In this latter case, the baby strawberry plants are attached to their parents by fibrous stalks that provide them with nutrients while they are getting established. Among mammals, 'brooding' is common and maternal affection for offspring is fairly widespread. Humans also use trickle-feeding techniques with their own offspring. Maintaining a connection between parent and offspring in this way allows parents to dynamically allocate resources between their offspring - depending on their perceived viability.

Cultural parental resource allocation

Resource boluses and resource trickles are both found in cultural evolution. Resource trickles seem to be more common than in the organic realm. The world of finance provides many examples. One example is sales people: it is common to put new recruits on a salary after teaching them how to do their job. This allows them to support themselves while they are learning their new trade - but before they are able to earn a healthy commission. Another example is franchises. When a new franchise starts up, it is sometimes supported economically for a while by existing ones - while the new establishment finds its feet. Offices, factories, farming and mining operations often behave in a similar way.

Maintaining a connection over which funds can be transferred is simple and cheap in the modern world, and a resource trickle provides more dynamic control over the flow of resources. Cultural parents don't normally perish during childbirth - and so parents are often around to supply a resource trickle.

Religious memeplexes are among those that make extensive use of extended parental investments. Religions are often old and highly evolved elements of culture, with considerable adaptation to the human psyche. It is common for them to feature extended indoctrination periods. Sometimes the indoctrination takes place by trusted family members during childhood years - when the human mind is at its most impressionable. There's an extended association between the religious memes in the adult teachers, and their cultural offspring inside the children.


A common example of cultural parental investment involves teaching. Teachers sometimes have extended relationships with their students. While they are together, memes are planted in the minds of the student - and the teacher cares for and nurtures them. Teachers don't just deliberately expose people to their memes, they check to make sure they are installed properly, and provide additional exposure if they are not. The memes are repeated and reinforced until they are well established in their new host. If one teaching method fails, another teaching method can be tried. Teachers often act as though they care for the future welfare of the memes they implant in students. They also act as though want the students to keep coming back - so that more memes can be installed. The meme's eye view pictures the memes inside the teacher influencing their behaviour - so that they ensure that the memetic offspring are well established in the minds of the students. If the student is inspired to subsequently go on to teach others, that is better still.

Teachers often teach others to teach. That's a case of memes not just caring about their immediate offspring - but trying to ensure that they become long-term ancestors. Obviously, being linked with memes associated with teaching others is a way for memes to improve their own fitness. Teaching memes are thus in demand - many other memes act as though they want to be associated with them.

Cultural kin recognition

Outside relationships that involve teaching, cultural kin selection acting on memes inside different humans demands that memes somehow "recognize" each other while they are inside other human bodies. For mental symbionts to identify other mental symbionts while they are inside human bodies is a non-trivial feat. If you think of memes as software, that may help to understand how such a thing is possible. How this feat is actually performed is interesting. Memes typically use the same psychological apparatus designed for recognizing relatives in order to to identify copies of themselves inside others.

Memes often subvert their host's kin recognition for their own ends by making non-relatives appear to be relatives. A concrete example of subversion of host kin recognition may be found on the battlefield. Shared military uniforms indicate shared nationality memeplexes. These memeplexes are prepared to sacrifice themselves for copies of themselves in other bodies – and they manipulate their hosts to achieve that end – to fool their hosts into believing that their fellows are their genetic kin – not just their memetic kin. This is why military uniforms are often designed to cover the entire body and make soldiers appear to be identical clones of one another - to act as a superstimulus to the kin recognition apparatus in the human brain. Shakespeare expressed the feeling of brotherhood associated with warfare - writing:

We few, we happy few, we band of brothers; For he to-day that sheds his blood with me shall be my brother.
The human hosts may not literally be fooled into thinking that non-relatives are really kin. However, kin recognition is part of the human psyche - extending down into unconscious realms. While people may not be consciously fooled, part of their brain is still thinking: "kin" - and acting accordingly.

Religions subvert kinship kinship recognition systems as well. Monks are "brothers" in a "brotherhood", and the priests are called "Father". Nuns are "sisters", in a "sisterhood" and the head nun is their "Mother Superior" - or their "Reverend Mother". Then there's the holy "father" - who plays the paternal role. As with the military, the monks and nuns often wear identical uniforms - so they look like kin. Church is all about family - but it the relationships involved are not organic - they are cultural.

Kin recognition has had a rocky ride in mainstream biology in recent years. It has turned out to not be as widespread as was originally imagined. Gardner and West's 2007 article "The Decline and Fall of Genetic Kin Recognition" covers the controversy - suggesting that markers used for kin recognition would tend to rapidly reach fixation - and become useless as kin-specific markers - unless high levels of mutation or selection oppose this. They suggest that selection for marker diversity which is caused by parasites may help to explain why kin recognition is widespread among humans. The theory of gene-meme coevolution suggests another answer to this question - that humans use memes as markers that act as proxies for DNA relatedness that are both highly variable and easily identifiable. In both cases, rapidly evolving symbionts would act to promote altruism by providing a rapidly-changing source markers to act as signals associated with relatedness.

Perhaps the most important thing to say about cultural kin recognition is that it is not a prerequisite for cultural kin selection. David Hales in 1997 claimed that:

memetic kin altruism can only function if memes can induce individuals to distinguish between memetic kin and non-kin.
However, advanced cognition which is capable of recognizing other individuals is not required in order to distinguish kin from non-kin. A strawberry plant doesn't need to "recognise" its own offspring - because it is joined to them by runners. Similarly, many organisms don't disperse their offspring very far from home - in which case, being nice to your neighbours is often much the same as being nice to your relatives.

Memetic relatedness

Relatedness is not always so easy to calculate in the case of culture. Here is Peter Richerson (2010) expressing scepticism on the topic:

In the case of culture, the analog of kinship is very hard to estimate. Having two parents with equal genetic contribution makes the calculation of relatedness easy. In cultural transmission, one, two, a few, or many people in your social network are possible sources of culture. People may use different parts of their network for different cultural domains. No one has proposed a way to estimate cultural relatedness in the face of such problems.
It is not true that no one has proposed a way to estimate cultural relatedness. Paul D. Allison did it in 1992. John Evers did it in 1998. However, it is true that the concept of "memetic relatedness" between people does face some practical problems. We currently have no practical way to sequence all a person's memes - to recover all the cultural information stored in an individual brain. We can, however measure the occurrence of particular memes, via questionnaires and similar methods. In the organic realm, basic a calculation of relatedness on shared genes works quite well - because of the mechanics of meiosis. However, in the cultural realm, memes are not dished out so evenly and uniformly - and one sample of memes might give one estimate of relatedness, while another sample might produce a different estimate. That is a practical problem for estimating memetic relatedness between people - both for scientists and for memes attempting to track their own offspring. However, there are statistical techniques designed to deal with this sort of issue. If you randomly sample some memes from one person and then see if they are present in another person (and repeat the process the other way around) that will probably produce a reasonably usable figure for the memetic relatedness between them. The issue isn't a show-stopping problem.

However, while calculating memetic relatedness between people is not easy, it is often not necessary. Consider for a moment the similar case of kin selection among organic parasites. Some braconid parasitoids attack caterpillas. They crawl into the caterpillar's brain, form cysts, and manipulate its behaviour. These parasites die - but the behavior they induce helps their kin to reproduce. This is an example of kin selection. However, there's no calculation of the additional level of relatedness between the hosts that arises as a result of them sharing parasite genes. Such a calculation would be irrelevant and unnecessary. Kin selection acting on memes is similar. Many of the important relationships are between sets of memes or memeplexes. Calculating relatedness between memes is pretty trivial. Often such relatednesses are either one or zero - i.e. either the memes are either identical copies, or they are not. This idea can be expanded to memeplexes without much difficulty. That is enough to support the theory of kin selection in the cultural domain. Averaging relatedness over all the memes in a single host is often not necessary in order to model the dynamics involved.

Hamilton's rule

The "holy grail" for students of cultural kin selection seems to have been to derive an equivalent of Hamilton's rule. There have been a number of attempts to do this. An early attempt was made in a paper by John Evers (1998) called "A justification of societal altruism according to the memetic application of Hamilton's Rule". This paper derived a variant of Hamilton's rule applied to memes - adapted to deal with horizontal transmission. It was based on the idea of using a figure for the "fraction of shared memes" in place of Hamilton's relatedness. However, John doesn't really go into the difficulties associated with this idea. Also, adjusting Hamilton's rule to deal with horizontal transmission doesn't seem to be particularly urgent to me.

More recently, David Queller (2011) wrote a paper titled "Expanded social fitness and Hamilton's rule for kin, kith, and kind" - which attempted to roll kin selection, "kith" selection and "green beard" effects into an extended version of Hamilton's rule. David Queller's work is not based on memetics. It attempts to cover all kinds of social effect - not just cultural ones. While this work is interesting and general, its generality works against it in some respects. It doesn't allow predictions to be based on shared memes - and that is one of the main virtues of cultural kin selection. David Queller's use of the term "kith selection" clashes with Gordon Rakita's prior use of the term in a potentially confusing manner. I'm inclined to label "kith selection" as unnecessarily-obscure jargon.

Of course, applying an unmodified version of Hamilton's rule directly to pair-wise interactions between cultural creatures is still perfectly possible. In the organic realm, kin selection theory makes use of the idea of "genetic relatedness" - and idea that gives a rough estimate of the proportion of rare genes that are likely to be shared between two individuals. Part of the attraction of Hamilton's rule is that it allows a cheap calculation of this "relatedness" - based on easily accessible information about geneaology. In practice a lot of relatedness figures between cultural creatures are either 0 or 1. The lack of memetic meiosis complicates this approach. Also, in cultural kin selection, there are additional difficulties after relatedness has been calculated. Genes typically affect behaviour in the organic realm fairly directly. However, for the "puppet masters" of memetics, memes must manipulate their hosts in a highly indirect manner. Memes face difficulties associated with accessing host sense data and with controlling host motor outputs. These are broadly similar to the difficulties parasites face in manipulating their hosts. As with parasites, the resulting host behaviour is the result of a battle with host's DNA genes and all the other memes that the host carries. Most memes don't get things their own way. This indirect control over behaviour acts as a further confounding factor which makes it harder for approaches based directly on using Hamilton's rule to produce useful answers.

I think it is best to avoid obsessing over memetic versions of Hamilton's rule. The original Hamilton's rule works in the cultural domain in an unmodified form. The only problem is that the complex and indirect nature of meme expression can mean that it is harder and more complex to apply. Of course, it is still possible to adopt the meme's eye view - and ask how a meme could act so as to affect the propagation of the surrounding swarm of copies of itself. That is still an extremely useful approach - even if you don't directly use Hamilton's rule.

Related content

This article is based on an excerpt from my forthcoming "Memes" book.

No comments:

Post a Comment