The Lifecycle of Memes

By Henrik Bjarneskans, Bjarne Grønnevik and Anders Sandberg

Abstract

Memes, self reproducing mental information structures analogous to genes in biology, can be seen as the basis for an explanatory model of cultural and psychological behaviour. Their properties and effects are evolutionary conditioned and ultimately seeks to promote their replication. To survive in a context the memes must meet certain conditions. We abstract a model of these conditions and use it to analyse three well-known memes: the "Kilroy was here" graffiti, urban legends and Christianity.

1. What is a Meme?

Memes were originally described by Richard Dawkins in his book The Selfish Gene (1976) [1]
a unit of cultural transmission, or a unit of imitation.
- - -
Examples of memes are tunes, ideas, catch-phrases, clothes fashions, ways of making pots or building arches. Just as genes propagate in the gene pool via a process which, in the broad sense can be called imitation. If a scientist hears, or reads about, a good idea, he passes it on to his colleagues and students. He mentions it in his articles and his lectures . If the idea catches on, it can be said to propagate itself, spreading from brain to brain.
What makes the meme concept so powerful is its close analogies to the theory of natural selection. Natural selection occurs whenever the following conditions exist (Dennet 1990):
  1. Variation: a continuing abundance of different elements.

  2. Heredity or replication: the elements have the capacity of creating copies or replicas of themselves.

  3. Differential "fitness": the number of copies of an element that are created in a given time varies, depending on the interaction between the features of that element (whatever it is that makes it different from other elements) and features of the environment in which it persists.
This is a quite general definition which is not limited to biology, and suggests that memes are subject to natural selection: they vary (due to "mutations" in transmission or mental storage, plus deliberate changes), they replicate (by definition) and have differing fitness. This leads to phenomena of competition, co-evolution, population dynamics and adaptation surprisingly similar to their biological counterparts. The set of shared memes form the memepool (in analogy with the genepool).

It should be noted that human decisions are part of the memetic selection process; from the perspective of the memetic ecology humans and human behaviour corresponds to the climatic and geological environment of biological life. In the meme perspective, it is more accurate to say that the message has evolved into its form in order to encourage people to spread it than to say that people have selected or 'bred' the message into its form.

Although people often make the decision to spread a meme or not consciously, this process is influenced by the meme. Some memes are viewed as important, and hence spread to others after a conscious and sometimes rational evaluation; some memes exploit aspects of cognition or emotion to bias their hosts to spread them. Natural selection favours memes that are good at reproducing, which suggests that in time there will exist many memes that are very efficient replicators. Their accuracy is irrelevant for their survival, only their ability to replicate and find new hosts; memes that interest people and encourage them to spread the meme will thrive at the expense of less attractive versions.

What makes the meme perspective so interesting is that it suggests that some of what we have learned from biology can be applied to human psychology. Dawkins points out that "a cultural trait may have evolved in the way it has simply because it is advantageous to itself" . Gross (1996) says

The main shift in thinking that needs to take place is to look at the spread of the legend not so much from the point of view of the people who propagate the warning, but from the point of view of the warning itself.
In memetics, ideas are viewed as almost independent creatures in a symbiotic relationship with human minds and cultures.


1.1 Defining Memes

The meme concept is somewhat slippery to define, and there is an multitude of definitions ranging from the very wide to the very narrow. The definition of meme we will use in this essay is
A meme is a (cognitive) information-structure able to replicate using human hosts and to influence their behaviour to promote replication.
This is a somewhat strict definition, since it excludes many structures able to replicate without influencing host behaviour or using non-human hosts such as chimpanzees, dolphins and computers. It can be seen as a subset of the more general memes described by Dawkins.

Memes do not only influence behaviour to promote replication, but many of the most successful memes have other side-effects (for example, being able to invoke various emotions) or promote their replication by being useful or through other features (like parasiting on other memes, e.g. parodies and imitations); using a biological analogy one could say symbiotic memes spread mainly using their usefulness, while parasitic memes compel the host to spread them. This compulsion can be more or less subtle, ranging from explicit orders like in chain letters ("Send ten copies of this letter to your friends") to implicit influences that link with our attitudes like the "Save the whales" meme described in (Hofstadter 1985, p. 55).

It is quite common that memes are confused with ideas/thoughts. Both are cognitive structures, but an idea is not self-replicating and is spread passively (i.e. for extrinsic reasons) if it is spread beyond its initial host at all. The difference is sometimes hazy; the idea "Isn't it time for us to eat?" can easily spread in a small group, but will not spread well outside the group and will disappear once the question is settled, while a meme usually can spread generally and does not have any limited lifespan.

It should also be noted that memes often form meme complexes, groups of memes mutually supporting each other and replicating together. The dividing line between a meme and a meme complex is yet again diffuse. In this text we will not try to distinguish between the two.

1.2. Issues in Memetics

1.2.1. The Meme - Gene Analogy

Much fuss has been made over this analogy since it was introduced by Dawkins in his "The Selfish Gene". When Dawkins introduced this analogy it was to give us a meaningful comparison in the light of which we would better understand the concept of memes. This was done to help our initial phase of understanding; unfortunately, many memeticists has not left this area. Many writers have scrutinised the comparison with the gene to see if it really is analogous or not, i.e. Hans-Cees Speel (1996). Although this has provided interesting reading we feel that it is a bit beside the point. The importance of memes lies not in whether they are mental copies of the genes and obey the same laws as the genes do or don't, but rather in how they work and what they are capable of (and not capable of). We do not feel that you can reach a complete understanding of this only by comparing them to other things. You have to study the idea of memes in it self.

This is why we have chosen to make a detailed study of what we can call "the memetic life-cycle", to try to discover its inner memetic workings. Our aim is to find a model that, whether analogous to the genetic life-cycle or not, is sound and supported by studies of existing memes.

1.2.2. Definitions

In the field of memetics there are a couple of different definitions of "host", "vector" and "meme" around, and there is a tendency to make these wide to the point of being meaningless. We want operational definitions that are usable and still distinct. Therefore, in this paper, we are going to use the words "host" and "vector" as such (meme has already been defined above):

Host = A host must be able to possess at least the potential capacity to elaborate on the meme and to perform those cognitive tasks connected to the meme that we normally refer to as "understanding". This means that only humans can be hosts (animals can perhaps become hosts for simpler memes, but we will not discuss this here), at least until the development of artificial intelligences reaches further.

Vector = A vector is anything that transports the meme between hosts without the capacity to reflect on the meme. Examples are a wall, a voice, an email-program, or a picture. Can a human be a vector? Yes she can, if she lacks the cognitive capacity (or interest) to elaborate on a specific meme. Then she is just a non-reflective carrier of the meme, much the same as a book. Note though that the human vector is still a potential host - or inactive host (Grant, 1990) - for the meme, should she suddenly choose to analyse the meme (in its widest sense) or achieve the contextual understanding which would make this possible.

1.2.3. The Conscious Meme

It is worth noting that although the terminology used in genetics and memetics sometimes seems to indicate that genes and memes act upon their own conscious will, this is of course not the case. Genes and memes are not conscious, and they do not have a will as such to act upon. But it is practical and economical to speak as if they do, since their behaviour follows such patterns.

This way of speaking can be seen as lazy shorthand; "a meme wants X" means "the fitness of a meme is enhanced by X".

2. The Lifecycle of Memes

Memes have a life-cycle similar to parasites (Fig 1). During the transmission phase of the meme it is encoded in a vector, such as a spoken message, text, image, email, observed behaviour or slab of stone. When a potential host decodes the meme (reads the text, hears the message) the meme may become active and infects the person, who becomes a new host (the infection phase). At some point the meme is encoded in a suitable vector (not necessarily the same medium it was originally decoded from) and can be spread to infect new hosts. Figure 1
This division of the lifecycle makes it easier to discuss memetic selection criteria, such as the list proposed by Heylighen (1994): or other, more elaborate divisions such as (Hale-Evans 1995).

In the following we will discuss the intrinsic factors of the meme that contributes to its fitness, and those external factors that mesh with them. We will look at the factors that help or hinder a meme in each of the phases of its lifecycle as depicted in Figure 1. These factors will be summarized below in an extended version of the model (Figure 2).

A successful meme will be good at exploiting these factors in its environment, while memes that cannot exploit them well will be out-competed and eventually go extinct. If one phase presents an insurmountable obstacle to the meme, it will be unable to reproduce and survive. This allows us to make estimates of memetic viability.

2.1. Transmission Phase

In the transmission phase the meme is encoded in a vector, some kind of information- carrying medium. Which medium is used strongly depends on the meme, both how it can be expressed (influenced by its complexity, the need for copying-fidelity and the requirements of its semantic form) and how it "wants" to be expressed. Often the medium is strongly linked with the meme or an actual part of it, as in the case of the "Kilroy was here" meme where part of the meme is the graffiti itself, suggesting the possibility of scrawling it on some suitable surface; "The medium is the message" as Marshall McLuhan put it. Memes are able to shift between media, sometimes with a mutating effect on the meme (such as the major differences between the Hunchback of Notre Dame by Victor Hugo and the same story told by Disney).

It is interesting to see how memes and media have co-evolved: many media have been developed as memetic vectors, able to encode memes indefinitely and with a high degree of exactness, while memes have evolved to use them and exploit their peculiarities.

Dawkins (1976) suggests three qualities of a meme that gives it a high survival value: longevity, fecundity and copying fidelity. Longevity and copying-fidelity are most significant in the transmission phase (although longevity can also be seen as the meme's ability to remain in memory for a long time), and are strongly linked with the properties of the medium. Heylighen (1994) also points out that ease of communication increases memetic fitness, either through creating salient behaviour that is easy to imitate or by being able to be clearly expressed.

2.1.1. Reproduction Ability

The more copies of itself a meme can encode in vectors, the higher fitness does it have (by definition). This implies that memes would thrive in media where it is easy to make numerous copies and distribute them widely. From a memetic standpoint, the ideal medium is a broadcast medium where many copies can be made cheaply. If it is one-to- many such as radio, then the memes of the broadcaster will spread with little feedback; this supports a less diverse memetic ecology than a many-to-many network, but also promotes the memes of the sender more strongly. In a many-to-many medium such as the Internet there are multiple meme sources, and the memetic diversity is high.

Two typical examples of the influence of reproduction ability are "xeroxlore" ("You don't have to be insane to work here, but it helps") that can be found in almost every modern office thanks to photocopiers, and Internet spams ("Make Money Fast!") that thrive in the broadcast environment of the Internet; neither meme would be possible without easy copying.

2.1.2. Copying Fidelity

While mutation leads to evolution, it also risks to destroy or degrade the meme. Just as in biological organisms a balance has to exist between evolvability and copying fidelity (Kelly 1994, p. 537). Many memes have properties that enhance their copying fidelity, often by making errors in encoding noticeable (note the analogy to error-detecting and -correcting codes cf. (Biggs 1985)).
2.1.2.a. Non-alteration Policy
Some memes explicitly forbid any alteration of themselves, such as the instructions given in chain letters, the copyright notices of public domain information (only unaltered copying is allowed) and many sacred formulae, that warn of awful consequences if they should be altered. They increase their copying fidelity by making the host careful in encoding them well.
2.1.2.b. Structure
Cohesiveness may also improve copying fidelity by introducing patterns that support the copying and transmission process. Poetry becomes stable by using rhyme and meter, since mutations create obvious errors that can be corrected. Many memes contain noteworthy features such as humour, commonly known symbols or repetition (cf. the common use of groups of three in myths). Incidentally, these properties also make memory encoding and retrieval easier.
2.1.2.c. Simplicity
Another way of achieving high copying fidelity and a low mutation rate is simplicity. A short meme is less likely to be changed, and may even be "brittle" ?any change makes it obviously unusable to the encoder, and thus prevents mutation. Longer or larger memes can survive only in media with high copying fidelity and low copying costs.
2.1.2.d. Repetition
A fourth way of achieving a low mutation rate in the vector form is to concentrate on making potential hosts understand the meme, not to ensure perfect copying fidelity (which may not be possible in some media, such as the spoken word). By repeated exposure, a potential host will not only recreate the meme from possibly distorted encodings, but also be more likely to become infected. This is an especially useful strategy for meme complexes, large memes and memes with high abstractability since they can (and often must) be transmitted piecemeal.

2.1.3. Survival

The meme has to be able to survive in the medium, and the medium itself has to survive. Memes that can be encoded in durable vectors such as books, great art or major myths can spread almost unchanged for millennia, while memes encoded in ephemeral vectors such as the spoken word have to spread from host to host quickly and are also more likely to mutate into new variants.

2.1.4. Abstractability

Some memes can be encoded in a large variety of vectors, while others are fixed in a single medium. Canonical examples are funny stories, which can be spoken, acted, written or drawn, and graffiti, which is its own vector.

A meme that can pass from one medium to another can spread more easily and change into forms more able to infect new hosts, but is of course more sensitive to mutations. One strategy that works well together with high abstractability is the understanding- repeated exposure strategy, since it decreases the mutation rate and partially relies on the abstractability of the meme.

2.1.5. Decodability

Finally, a meme must ensure that its vector is possible to decode for new potential hosts. Most artificial media used for information and meme storage are intended to be easy to decode (Dawkins would say there is a memetic selection effect here: the memes of creating media that have low decodability would not be replicated as much as the memes of creating high-decodability media, which would tend to dominate), but many natural media such as behaviour are not necessarily easy to decode. Observed behaviour can be viewed as a medium of meme transmission; this is essentially social learning theory (Bandura 1977) restated in a memetic framework. Many behaviours (e.g. gestures such as applauding) are learned this way with no verbal explanation even at an early age and reinforced through positive feedback.

A classic example of behaviour transmission (although somewhat outside our strict definition of memes) is the spread of food-washing observed among Japanese monkeys: in 1952 monkeys on the island Koshima were given sweet potatoes by researchers. The potatoes were left on the beach, and while the monkeys enjoyed the food they disliked the sand. One monkey, 18-month Ima, found that she could get rid of the sand by washing the potatoes in a stream or the sea. Other monkeys observed her behaviour and repeated it as they found that it had positive results, until after a few years practically all monkeys except the oldest washed their food [2].

It should be noted that some memes increase their fitness by making decoding harder . Belonging to an exclusive group increases self-esteem; memes that can only be decoded by some hosts will thus provide them with a feeling of superiority (whose strength depends on the exclusiveness of the meme and other factors) and thus gain a certain infective advantage. The popularity of secret languages among children and secret orders among the upper classes demonstrate this.

A hard-to-decode meme requires more mental activity to decode, which makes infection more likely if the salient aspects of the meme can be reasonably sure to be decoded in the process, and suggests (through cognitive dissonance) a higher value of the meme. If the meme can also be interpreted in several or arbitrary ways, it is also more likely that potential hosts settle for interpretations that fits their attitudes the best. An example of this class of meme is alchemical writings, which are heavily cloaked in symbolism and riddles: the bait is the promise of powerful knowledge (which doesn't have to be delivered if the riddles are hard enough), the meme itself is the symbolical language of alchemy and the hook that leads to transmission is the self-enhancing feeling of spreading esoteric wisdom to the select few who can understand it.

2.2. Decoding

Decoding is when a potential host interprets and restructures the information pattern of the meme in the vector by perceiving and understanding, and thus creates a mental copy of the meme. In other words, if a meme is to infect a host, it must first be perceived by the host, then decoded to fit the host's schemata. The possibility of infection arises after decoding. Many of these decoding processes are automatic and/or subconscious.

2.2.1. Understanding and Interest

If the host is incapable of understanding the meme or even incapable of perceiving it in the form in which it is being transmitted, the meme will not proceed to the infection phase. Memes that in some way promote their perception, by being encoded in a noticeable vector, by containing strong emotional content or otherwise arouse interest will have an increased likelihood of infecting the host.

Understanding is in this case not limited to conscious understanding, as in the case of copying a popular fashion, neither is it always necessary that the host acquires a complete and correct understanding of the meme (see the "Kilroy was here" example) if it can later be fully decoded using the repetition strategy.

2.2.2. Pro- and Contra-memes

There is a possibility that the potential host is infected by memes with the function of counteracting or helping the current meme. Such memes are often symbiotic parts of larger meme-complexes, with the function of indirectly promoting itself trough the promotion of its meme-complex. A typical example is religions, which promotes trust in the teachings of the religious authorities (pro-memes) and often contain memes denouncing competing religions as heresy (contra-memes).

Contra-memes act by making their hosts automatically reject memes that do not fit the dominant cognitive structures. Often they act by creating a strong emotional response or attributing negative traits to the meme (knee-jerk reactions). In the same way pro- memes create positive attributions of certain memes and ease their assimilation.

2.3. Infection Phase

After successful decoding the meme becomes part of the host's mental structures, and this is called infection. A person who does not remember a meme at all is not infected. A person that does remember a meme but who's behaviour is not affected has thus become a human vector. A person whose behaviour is affected by a meme has been actively infected and can potentially transmit it to other hosts.

The following factors will influence whether an active infection will occur or not.

2.3.1. Ability to Fit Cognitive Structures

The meme must fit into the present schemata of the host to be seriously considered. A meme that assumes that God exists will probably not be successful in the mind of an atheist. To convince this person, an extensive set of pro-memes would have to be successful in infecting the atheist, thus reconfiguring the cognitive structures of its host to that of a religious person.

2.3.2. Threat / Bait

Most memes use threats or temptations to make the host accept the meme. If the meme appears to provide an advantage over previous memes, it can replace them (but it should be noted that older memes may have infiltrated the motivation structure of the host, making her unwilling to switch views anyway). Threats are even more potent, since potential risks are evaluated more strongly than potential gains (Kahneman and Tversky 1982), which means that a threatening meme often can out-compete a tempting meme.

Often threats and baits are combined to further enhance the meme. Many religions use this in propagating themselves, promising the faithless a hot afterlife (threatening) and the faithful a fluffy and light one (tempting).

2.3.3. Storage

If a meme is to be spread by a host for a long time, the host must remember the meme. If a host is infected and later forgets the meme and/or stops acting out the new behaviour before the host has spread the meme on, the host has not done the meme any more good than if the host had not been infected in the first place. Thus successful memes encourage permanent or long-lasting changes in the host. Note that it is not necessary for the hosts to remember the meme itself, just change their behaviours in a way that will promote the spread of the (reconstructed) meme.

This can be achieved in several ways:

2.3.3.a. Assimilation / accommodation
If the meme can be assimilated into existing schemata, it will be supported by them. If the meme can somehow force an accommodation, it will have even better chances of becoming a vital part of a long-lasting mental structure (and this also gives it an excellent position from which to act as a contra- or pro-meme).
2.3.3.b. Elaboration
Something that a host actively thinks about is less likely to be forgotten, and also more likely to influence behaviour. Thus memes that encourage thinking or fantasising about themselves or related concepts have increased chances of survival. Rituals and ceremony are often powerful reminders of the meme.
2.3.3.c. External storage
Since human memory tends to be rather uncertain, external memory aids can also aid memes greatly not just as vectors, but as memory feedback.

If a host is infected by a scientific meme-complex he will be encouraged to read books relating to the meme complex. The host becomes likely to learn more and more about the theories rather than forgetting parts of them, and should he forget something relevant he can look it up again, the books can serve as memory feedback loops and also act as vectors for other parts of the meme-complex causing further infection.

2.3.4. Storage Time

The longer a meme infects its host, the more chances it has to be spread on to other potential hosts, but there is also an increased risk of mutation. Some memes have limited lifetimes determined by outside factors, such as millennial memes ("The world will be destroyed on Tuesday!") whose fitness decrease significantly after a set date (but before that date gain in fitness by being actual and urgent). There are examples of computer viruses that destroy themselves after a predefined goal, like making X copies of itself or residing for X days in the computer (Dawkins 1993), which are surprisingly similar to the instructions of chain letters ?after making the copies and sending them, the host is no longer instructed to spread the meme.

2.3.5. Survival

Survival of the meme depend upon whether or not the meme can handle or use the following problems in the host mind:
2.3.5.a. External Contra-Memes
An infection may be subject to contra-memes, trying to "cure" the infection after it has become active, as opposed to a contra-memes struggle to prevent an infection in the decoding phase. A contra-meme does not have to be specifically directed towards its counterpart, it only has to counteract the intention of its counterpart. Survival requires a meme to withstand their attacks. Many obviously false but firmly believed memes can be surprisingly tenacious in the face of opposing evidence (Gross 1996).
2.3.5.b. Immunity
If a meme can make the host assimilate or accommodate contra-schemata to protect itself, it is less likely to have to deal with external contra-memes or competitors. The meme complex of atheism is likely to give its host a defence against any religious meme complex. This defence is practically an immunity since a convinced atheist wont even consider listening to a religious host with an open attitude. The opposite seems to apply too.
2.3.5.c. Universality
If the meme is widely enough defined, or is a generally accepted concept, it can assimilate all other memes because it is more general and can be used as a context which can explain all necessary functions. Many religions claim to explain the worlds creation and other phenomena better than modern science.

2.4. Encoding and Spread

The meme must replicate itself if it wants to be successful, that is, to spread to and infect more hosts. Quantity is important for some memes in order to survive, especially weak memes which do not occupy the ideosphere (the universe of ideas, an analogy of the biosphere (Monod 1970)) of their host for long since they are easily forgotten. As a reaction they try to infect as many hosts in as little time as possible. A typical example of this is chain-letters.

Other memes have a more targeted area of hosts. The target hosts are the only ones necessary for the survival of the meme, but they are on the other hand crucial. This kind of meme is often evolved to survive better in a single host over time than the quality- oriented one, since the spreading of this second meme type often involves a more complex and time-consuming infection process (as in religion).

2.4.1. Hooks and motivation

A host doesn't start to spread a particular meme all by herself. She has to be motivated. To this extent, memes have a particular trait, or a co-meme, called a hook. The hook is what encourages the host to spread the meme. A common hook relies on humans altruism. It works like this (after Hofstadter 1985, p. 55):
ALTRUISTIC PREMISE: I (the host) don't want any harm to befall my friends.

MEME HOOK: "Anyone who doesn't believe in this doctrine will be tormented in the afterlife."

CONCLUSION: Since the doctrine is true, and since the premise is true, I will make sure that my friends starts to believe in the doctrine .

Another type of hook is of course the opposite, the threat directed towards the host. This is an efficient tool not only for spreading, but works also to minimise mutation ("You will be tormented if you misread the doctrine") and to guarantee a firm place in the host's ideosphere ("You will be tormented if you lose faith").

Naturally, all hooks aren't as elaborate as the ones above. These types are normally found in the larger meme complexes. But singular memes are equipped with the hook co-meme too. Take, for instance, the joke. Why do you tell a joke? Maybe because you want to make people happy, and/or you want to be appreciated or popular. The hook for telling - and spreading - a joke would then be something like this:

MEME HOOK: "If you tell people something funny they will be happy and they will appreciate you."
Hopefully, this is enough to motivate you to start spreading the meme. A hook should obviously not contradict common sense and in a meme complex it should ultimately feel like a natural extension of the meme complex. This is to make sure the hook is easily, and preferably automatically, activated. A complicated hook that demands a lot of conscious elaboration in order to be activated has doomed the meme by inhibiting the possibilities of swift reproduction.

Note that the above-mentioned meme hook is not married to a particular joke. Rather, it is implicit in the hosts disposition towards jokes in general. Whereas the hook for, say, a chain letter is explicit and tells you what to with this particular letter and this letter only.

2.4.2. Feedback

Another important determinant in how successful the spreading of a particular meme will be is the feedback. This includes direct as well as indirect feedback.

Direct feedback is the immediate response you get when trying to spread the meme. Did the new host get infected (Did she laugh at you joke)? Was the vector you chose a satisfying medium? Indirect feedback concerns matters such as how often you recognise the meme in different media and whether it seems like a lot of people inhabit the meme (and thereby strengthen your own belief in it). In other words, indirect feedback is your recognition of the spreading meme apart from your own participation in the process. You can naturally recognise some indirect feedback as being partly a result of your own reproduction of the meme, but only in the sense of your meme spreading being part of a larger meme-spreading complex. If your actions are the exclusive producer of the feedback, then the feedback is always direct.

Positive feedback strengthens the belief in the meme and encourages the spread, whether negative feedback works in the opposite direction. Strong enough negative feedback can actually kill the meme in the original host, if the meme is not equipped to handle a situation like that. Successful memeplexes are well prepared, e.g. the religious defence "If they don't believe you they work for the devil, and you should avoid further contact.". But for smaller memes, negative feedback often proves to be fatal. This is what happens when we say that a joke "wears out", we start to receive negative feedback because everybody is tired of the joke and eventually we stop telling it.

2.4.3. Survival through Vectors

As always, it is important for the meme to mutate as little as possible. But reproduction must also be as easy as possible. An optimal vector should satisfy both these needs. The Torah is an example of minimal mutation but extremely low reproduction rate. The rabbis were forced to copy the original by hand, page by page, letter by letter. If one letter got wrong they had to do the page all over again (Eco 1988, p. 569). Gutenberg changed all that, and today the printed word is one of our most common vectors.

Another element which enhances the meme's possibilities is whether it survives a transition from one media to another. Of course, this flexibility also enhances the risk for mutation. One might think that memes lie dormant while in vectors, but this is not necessarily so. With the advent of television and more interactive types of media, not to mention the Internet, the meme have got the possibility to shape its vectors to the extent that the media becomes the message, to paraphrase McLuhan. Examples of this are the various cyberchurches which exist only on the Internet and preaches technosophy, which aims to foster a spiritual appreciation for technology (Wright, 1996).

2.5. Summary of Model

In order to stay successful and reproductive, the meme has to complete the described cycle over and over again - preferably with as little mutation as possible (unless mutation is one of its particular characteristics, as in the example of urban legends, below). The meme will die if it is unable to complete the cycle.

Our ambition it writing this paper has been to analyse the obstacles and ordeals which the meme has to face in its life-cycle, and to get an understanding of its workings. The result is summarised in fig. 2.

To show how this cycle applies in real life, we will now end this study by analysing three different well-known memes or meme complexes. While reading the examples, use fig. 2 as a tool to see the different phases. Figure 2

3. Three Illuminating Examples

3.1. The "Kilroy Was Here" Meme

This meme originated during the second world war, when wharf inspector James. J. Kilroy of Quincey, Massachusetts used the slogan "Kilroy was here" to mark products he had tested and approved. The marked products appeared on many battlefields, and the signature that seemed to appear just about everywhere caught the imagination of many soldiers, who began to copy it on just about any writable surface (Funk 1950). Most likely others were intrigued by the slogan that appeared in unlikely places, so they copied it further to spread the myth.

While the meme spread well for several decades, it eventually went all but extinct in its active form. There seems to be several reasons for this:

3.1.1. Properties of the Meme

"Kilroy was here" is extremely well suited for the transmission phase, where it is encoded in a graffiti vector.

It is very easy to reproduce, and due to its brevity the copying fidelity can be very high. Its decodability is also high, since after the second world war English became a lingua franca over a large part of the world and acquired a certain status. The meme was spread by English-speaking hosts, and would thus tend to end up in areas where English was understood at least by a part of the population.

The survival of graffiti is highly variable, but by its nature it is semi-permanent and intended to be highly visible, which ensures that more potential hosts notices it.

It is uncertain how well "Kilroy was here" can be abstracted. In its original form, the graffiti vector was an integral part of the meme and crucial to hint at that it should be reproduced. Later variants appeared, such as a cartoon figure and stickers, but they do not appear to have been as fertile, mostly because they were harder to copy.

The meme's intimate connection to its vector, e.g. walls, made it poorly fit to survive in other media. Also, the meme was very sensitive for mutations. It was enough that you changed one of its smallest parts, a letter, to seriously damage the meme.

It is of great help to understand this phrase, that is to know what the English words mean together. The problem of decoding the sentence is quite an easy one, but it is harder to decode what it really means. This is probably one of its strengths. One can find ones own explanation of its meaning. Further more it is small and simple. During a time when the meme is popular, the host also gets multiple chances to try to decode it.

Since the meme is without obvious meaning it is hard to contradict, so there should be no active defence against the meme. The meaninglessness of it can also invoke wandering thoughts about the meme, and actively elaborating is connected with better remembering.

What motivated people to spread the "Kilroy was here" meme? There was never any direct host-to-host contact in the case of this meme. This meant that no host received a direct positive feedback, which is a powerful reproduction booster. And there was no obvious hook accompanying the meme.

This is one of the great meme mysteries. Perhaps that was enough motivation to spread the meme, to become part of the mystery - and also, in the beginning, to share the joke of who this much-talked-about Kilroy character was. Thus, the host created a bond with a community of Kilroy writers, most of which she would never meet, but could still belong to. A feeling of belonging may have served as a hook to motivate the conscious spread of the meme.

The Kilroy writers only way of confirming that there were others was the indirect feedback, but this is also the point. The Kilroy writers became invisible, even before each other, so that the meme seemed to live its own life mysteriously reproducing on the walls and the writers themselves could feel as privileged members of a mysterious brotherhood.

3.2. Urban Legends

Urban legends have been described (alt.folklore.urban FAQ) as stories that:
Appears mysteriously and spreads spontaneously in various forms.

Contains elements of humour or horror (the horror often "punishes" someone who flouts society's conventions)

Makes good storytelling.

The first property suggests that they are memes, able to mutate and spreading with no link to the original creator (although some urban legends attribute the story to some proper authority to gain some measure of credibility; cf. Gross (1996)). The second property in part together with the third explain why they are replicated: they fill a psychological need for entertainment, emotion, reinforcement of attitudes and attention for the storyteller.

There has been much discussion about the links between traditional storytelling, urban legends, memorats [3] and rumours. Generally urban legends are apparently realistic stories but actually have a stylised content, with a simple plot which is often very visual and easily remembered and told (af Klintberg 1978, pp. 154). This contrasts to rumours, which are short (often just a simple statement with additional information) and lack epic structure. Both urban legends and rumours can be viewed as ways of spreading information in situations without official information and by releasing the tension of social uncertainty (Mullen 1972).

It is likely that rumours may evolve into urban legends. The classic study The Psychology of Rumour by Allport and Postman (1947) suggests that a rumour will become more stylised during spreading. This may be partially the result of the experimental set-up, which is based on unilinear spread; in collective spread variations tends to appear (Peterson and Gist 1951). These results are predicted by memetic theory: in an unilinear spreading situation only transmission ability will be relevant (since the number of hosts are small), while in a collective situation the increased number of transmissions will also lead to mutation and variation. Once a rumour or memorat taps into a good epic form due to a mutation or a deliberate change its spread will be highly enhanced, and it becomes an urban myth which will spread fast.

3.2.1. Properties of the Meme

Urban legends are usually transmitted as oral tradition; different legends thrive in different social groups. The mutation frequency is rather high due to the oral spread, but this is counteracted by comparatively simple and strong storylines which can be elaborated in various ways, often based on strongly interesting subjects such as sex, death, the supernatural and embarrassment. The subjects all tend to promote listener interest and hence replication.

To be accepted as anything other than a joke or pure horror story the legends need some measure of plausibility; often this is provided by referring to apparently real people and institutions (Gross 1996), the story shows that the people involved are "normal" people or a real "friend of a friend" (thus extending the storyteller's credibility indirectly to the presumed source).

Another reason to accept or remember the story is by hearing it from several independent sources; this appears to confirm its veracity, and minor inconsistencies can be explained away as being errors in re-telling.

Memetic theory predicts that legends that fit in well with the social schemata and attitudes of their hosts will have a higher fitness than legends that do not conform, and this seems to be supported by the changes that occur in urban legends over time. Some urban legends have survived for many decades, changing to fit in with changes in popular attitude or society. The myths about people being drugged in subways originally involved white slave trade, but today warn of dealers seeking to make more people dependent on drugs.

Bengt af Klintberg points out (1978, pp. 153) that it is possible to partially classify urban myths by the way they spread. Many spread in the characteristic way of rumours: an exponential spread until saturation followed by a die-back (sometimes caused by an official denial). The same rumour or urban myth can recur in other places with a similar way of spreading; in a surprising number of cases newspapers acts as vectors. Note the similarity to epidemics. These urban myths often deal with things relating to the listener's life, something that he or she could experience. But there are also urban legends that spread in a less explosive manner. Their contents are less likely to be experienced by the listener, and are told more as entertainment than actual events. Typical examples are the horror stories told by teenagers (af Klintberg 1978, pp. 181).

To survive memetically the legends do not need to be believed, since they provide other incentives for being told, but if the storyteller believes in the legend its spread and credibility will be enhanced and can be motivated even if it lacks obvious entertainment value. This may explain the difference between the fast- and slow-spreading legends.

Another important factor is that believable legends involve the listeners much more by suggesting they could be the victims of the story; in the case of potential dangers (such as the "rat in the pizza" stories) there is an incentive of remembering the danger and passing on the knowledge.

3.3. Christianity

Religions represent some of the most powerful and elaborate meme complexes in existence today; they have evolved over millennia into countless variants and co- evolved with cultures. Making a complete memetic analysis of even a single religion is beyond the scope of this paper, so by necessity the following discussion will be rather general.

Religions tend to consist of some basic core memes (in the case of Christianity the belief in God and salvation through Christ) surrounded by symbiotic doctrinary memes (how salvation can be reached, ethical systems, the cosmology) and then an immense cloud of related memes (religious stories, doctrines, interpretations). These memes form a symbiotic whole; the core memes need symbiotic memes to provide hooks and baits, and the symbiotic memes reinforce each other and are given legitimacy by the core memes.

3.3.1. Properties of the Meme

The Christianity meme complex has throughout history been transmitted in a multitude of forms: as oral stories, through books and art, through example and through upbringing. Due to its complexity the transmission takes time and is closely linked to cultural understanding. This either requires a relatively concentrated effort to transmit the complex (mission) or to spread it by cultural diffusion and imitation (upbringing). A frequent diffusion situation is when a child is brought up in a Christian home. The Christian meme complex is presented as the truth about how the world functions. Variants of the meme have increased their fitness by encouraging a high rate of reproduction and cultural transmission (Lynch 1996).

Most major religions rely on active transmission: one or more hosts actively supports the spread of the meme, often in an interactive and deliberate way. Efficient methods for mission have co-evolved with the religion and the situation; the best missionaries gained the most converts, among which were the next generation of missionaries (and missionary teachers) who would learn and spread some of their best methods.

Classically, Christianity have used the bait of salvation (freedom from fear, personal happiness and prosperity, spiritual fulfilment, eternal life or union with God have all been promoted at various time) combined with the threat of damnation to promote interest and infection. This is however just the explicit bait, it appears likely that many Christian movements have been spread by implicit factors such as a sense of belonging, social conformity and a consistent world-view. It is worth noting that the baits and threats are mostly based on the symbiotic memes and not the core memes of the complex, which means their relative prevalence can change to fit the situation (for example the ratio of hellfire threats to salvation baits used in sermons) or they can evolve while leaving the core memes unchanged.

Religions are often better than other meme complexes (such as science) at explaining how the world works on an emotional level. They provide answers to existential questions that are emotionally appealing, creating a satisfying world model (which then becomes intellectually satisfying regardless of its consistency due to cognitive dissonance). Because religions seldom try to empirically prove themselves they cannot be disproved, which further aids their stability. A religion can spread regardless of the truth or falsity of its claims.

The Christianity meme contains an entire world-view, and seeks to cause an accommodation in the schemata of the infected host; no other memes are allowed to influence high-level planning and behaviour ("Ye cannot serve God and Mammon", Matthew 24). This is achieved by rejecting such memes or impulses as 'against God's will', 'sinful' or 'satanic'.

Like all the other major world religions, Christianity has a strong mission. It both exists as an explicit missionary order and in the form of an implicit altruistic hook (see the section about hooks and motivation). Christians are urged to set good examples to others, which also increases the likelihood of transmission through social learning.

4. Conclusion

The memetic approach is a tool that helps us to understand certain aspects of human behaviour. As with all tools, it is not necessarily the best solution at all times. That is why we equip ourselves with a lot of different psychological analytical instruments, so we will be able to choose an efficient approach for each different setting.

Critics of memetics complain about the danger of transforming everything into memes and memetics, feeling that it somehow reduces the importance of the human mind and places focus elsewhere. While some memeticists tend to go overboard with explaining everything in terms of memes, the same could be said about researchers in the fields of psychoanalysis, cognitive science or sociobiology. The importance lies in realising that psychology is full of more or less fit different explanatory perspectives, and memetics is just one of them. But by memetics you can often explain very complex cognitive structures and/or social psychological phenomenon (like Christianity) in a very general, no-nonsense way without having to entangle yourself in a web of unnecessarily complex theories.

Before our paper ends, we will inform you of our until now secret sub-goal with this paper. It is our intention that by now you, by reading this text, has been infected with one of the strongest memes on the planet: The Meta-Meme, e.g. the meme about the theory of memes. It is our sincere hope that you will tell your friends about this (yes, transmission and further infection) or maybe even let them read this paper. In either case, unless you carry a very strong vaccime (se appendix), we have made you a host. And you didn't even flinch. You should be lucky we are not after your money...

Footnotes

  1. It should be noted that the idea of self-replicating ideas and their evolutionary struggles had come up before in the writings of R. Sperry (1965) and J. Monod (1970).
  2. It is interesting to note that the real story eventually became a basis for a common urban legend, the "100th monkey effect": once 100 monkeys knew how to wash their food the ability suddenly became widespread among the monkeys, even outside Koshima. This claim was originally made in the 1979 book Lifetide by Lyall Watson where the author made up the story; since then it has become common myth in the peace and environmental movements (Amundson 1987). Compare this to the section about the evolution of urban legends.
  3. Stories relating real personal experiences, which are interpreted through the collective tradition but lack the fixed intrigue of legends. The term was introduced during the 1930's by C.W. von Sydow. (af Klintberg 1978, pp. 152)

References

Amundson, R. (1987) Watson and the Hundredth Monkey Phenomenon. Sceptical Inquirer Spring 1987, pp. 303-304.

Allport, G. W., Postman L. J. (1947) The Psychology of Rumour

alt.folklore.urban FAQ http://www.urbanlegends.com/afu.faq/

Bandura, A. (1977) Social Learning Theory Prentice-Hall, Englewood Cliffs, NJ.

Biggs, N.L. (1985) Discrete Mathematics , Oxford: Oxford Science Publications

Dawkins R, (1976) The Selfish Gene , Oxford: Oxford University Press

Dawkins, R. (1993) Viruses of the Mind. Free Inquiry , summer 1993, vol. 13 nr 3 http://physserv1.physics.wisc.edu/~shalizi/Dawkins/viruses-of-the-mind.html

Dennet, D. (1990) Memes and the Exploitation of Imagination, Journal of Aesthetics and Art Criticism , 48, 127-35, Spring 1990 http://www.tufts.edu/as/cogstud/papers/memeimag.htm

Eco, U. (1988) Focaults Pendel. Stockholm: Brombergs.

Funk, W. (1950) Word Origins and Their Romantic Stories , Funk & Wagnall Publishing Co.

Grant, G. (1990) Memetic Lexicon file:///h:/alephweb/Trans/Cultural/Memetics/meme_lex.html

Gross, D. (1996) The Blue Star Meme: Applying Natural Selection Thinking to Urban Legends http://www.lycaeum.org/drugs/other/tattoo/meme.html

Hale-Evans, R. (1995) Memetics: A Systems Metabiology http://www.apocalypse.org/pub/u/rwhe/memetics.html

Heylighen, F. (1994) Memetic Selection Criteria http://pespmc1.vub.ac.be/MEMSELC.html

Hofstadter, D. R. (1985). Metamagical Themas: Questing for the Essence of Mind and Pattern. New York: Basic Books.

Kahneman, D., Tversky, A., eds., (1982) Judgment under uncertainty: Heuristics and biases. Cambridge Univ. Press, NY.

af Klintberg, B. (1982) Harens Klagan: Studier i gammal och ny folktro. Andra upplagan. Stockholm: Norstedt & Söner.

Kelly, K. 1994 Out of Control: the new Biology of Machines , London: Fourth Estate Limited.

Lynch, A. (1996) Thought Contagion: How Belief Spreads Through Society: The New Science of Memes . New York: Basic Books. http://www.mcs.net/~aaron/thoughtcontagion.html

Monod, J. 1970 Chance and Necessity

Mullen, P. B. (1972) Modern Legend and Rumor Theory. Journal of the Folklore Institute 9:2-3, 95-109 1972.

Peterson, W., Gist, N. P. (1951) Rumor and Public Opinion. American Journal of Sociology 57, 159-167 1951.

Speel, H-C. (1996) Memetics: On a conceptual framework for cultural evolution. In: Heylighen, F. and Aerts, D. (ed.) The Evolution of Complexity. Kluwer, Dordrecht, in press. http://www.sepa.tudelft.nl/~afd_ba/hcesmem.html

Sperry R. (1965) Mind, brain, and Humanist Values. In New Views of the Nature of Man, edited by John R. Platt. Chicago: University of Chicago Press.

Wright, R. (1996) Can Thor Make a Comeback? Time vol. 148 No. 25 December 16 1996