Basic Memetics

Sometime between 2000 and 2003 I became fascinated in psychology and sociology and read much on the subjects. This has lead to many musings and observations over the years. I had always been interested in how people acquire ideas, especially beliefs, and how it was that different people could come to different conclusions. One branch of sociology that captured my imagination more than any other was memetics.

Memetics is the branch of sociology that deals with the study of memes. The theory of memetics states that meaning comes in discrete quanta upon which a Darwinian algorithm may act. Cultural institutions, knowledge, and the like are made up of ideas called memes. These memes are passed from one person to another by both communication and imitation and so spread through society. Some memes are more likely to be transmitted than others. Sometimes it is because passing on the idea gives some benefit to the transmitter (e.g. makes them look smart, makes people like them, makes people buy their products). Sometimes it is because the meme is more readily noticed to the receiver (e.g. promises higher income, warns them of danger, promises better sex). Sometimes it is because simpler memes are easier to remember, and fewer errors are made in transmission. Whatever the reason, some memes become common and others go “extinct”. It is said that those memes were “selected” against. Often, when a misunderstanding occurs in transmission, it is said that a “mutation” happens, creating a new meme. This new meme may be either more likely or less likely to spread, depending on its nature. Over time, a group of many interrelated memes (called a memeplex), such as a culture or a language, may evolve through this mutation and selection process. This is why many use the story of genetic evolution to help explain memetic evolution.

Memes and Genes: Genes are inherited directly from parents. With the exception of mutations, the genes of the offspring are identical copies of the genes of the parents. Memes, on the other hand are not transmitted directly. A potential host must observe the current host’s behavior to “reverse-engineer” the meme in his own mind. When different memes cause the same behavior, the underlying memes may not be the same. Genes come from only two parents (or one in some species) and deliver their genes at conception. Memes can come from anybody at any stage of life. Genes have copies of themselves in every cell. To learn the genetic makeup of an individual, one need only take one cell. For nearly all types of organisms, this observation will not cause any harm and will certainly not alter the genetic makeup (single-celled organisms are an exception). To learn the memetic makeup of an individual, one cannot choose from billions of copies; there aren’t any. Any questioning, scanning, or “dissection” of the mind will affect what is there. The act of observing alters the memetic makeup. This is why study by observing behaviors and artifacts is preferred.

Memes compete for resources (records and human memory) and so only some survive. Accents cannot be transmitted by being carved in stone. Practices deemed private will not be discussed and will die out in the individual that holds them. Some practices are simply inferior to others and are forgotten in favor of others. Some memes are spread almost exclusively from parent to child. Upbringing makes a huge difference to the way people turn out later in life. Smokers tend to spawn smokers. Children tend to be of the same religion as their parents. These memes prosper by encouraging the one who holds the meme to have as many children as possible. Taboos against birth control may be just such an example of this. In theory, those who shun birth control have more kids, and therefore more become carriers of the taboo. Over time, those with the taboo should outnumber those that do not.

Benefits of The Paradigm: Memetics can yield useful insights that more conventional viewpoints that focus on the motives of individuals and groups cannot. One notable parallel that can be drawn is that between insecticides and censorship. Insecticides are used to kill certain types of insects, and keep them from spreading. Censorship does much the same thing to certain types of ideas. The point is to keep them from spreading. Unfortunately, there is a downside to using too much insecticide. Because of variation within the insect population, some insects are more resistant to insecticide than others. These resistant variants will invariably be the ones that escape the effects of the insecticide and continue to reproduce and spread. Then, because they are no longer competing with resources with their less resistant brethren (they are all dead, thanks to the farmer), the resistant variants now easily dominate, replenishing the numbers of the entire insect population in the next generation. If less insecticide or less potent insecticide was used, these resistant variants would never dominate over the others, and the insecticides would continue to be useful. This is how insecticides create “superbugs” (a similar phenomenon occurs with bacteria and antibiotics).

In the same way, censorship, if too strict or far-reaching, can generate “superideas” by killing (halting the spread of) the less virulent ideas, leaving only the strongest and most censorship-resistant ideas to dominate. If the censorship were applied more gently, it would still have retained its usefulness. It is believed by many that this is one of the reasons that religions often flourish secretly in those countries where practitioners of said religions are heavily persecuted. This effect would have been very easy to miss by one not using the memetic approach. It is a great paradigm.

Criticisms: There are many criticisms of memetics. One is that it has yet to make testable predictions. Culture is so complex that it cannot be known ahead of time which memetic effects will be strongest and so which memes will win. Whatever happens, it is compatible with the theory. Another problem is that cultural evolution may have a property known as high-sensitivity to initial conditions. This means that very small changes now can yield enormous changes in the future. Since it is impossible to know the exact state of every mind in the community in perfect detail, predictions will become exponentially inaccurate with time. This is the same reason that weather can’t be forecast accurately more than a week into the future. Feedback is also a problem if the scientists are themselves part of the same system they study. There is also the phenomenon of individual learning, as opposed to receiving information from others. It is not always clear how much individual learning there is in the system to compete with the memetic effects, and this must be taken into account.

Another criticism that has been raised against memetics is the possibility that “blending” may occur between memes sometimes. When learning a language, does one adopt the accent for a particular sound most often heard by the individual in question, or does one blend the sounds together, creating a unique sound? While it is obvious that larger memeplexes such as accents blend by mixing and matching individual memes, it is generally thought that the individual memes themselves are fundamental units of information that cannot blend; one must adopt either one or another. Now, however, it is being questioned whether a such thing as fundamental units of contextual information (meaning) exist.

At first glance, it would seem that there is no real cause for concern; the phenomenon of meme-blending should not kill memetics any more than the phenomenon of incomplete dominance should kill genetics. However, there is a difference: during incomplete dominance, the underlying genotype remains unchanged and one or both alleles may be passed on to the offspring. With meme-blending on the other hand, the original memes are lost and reverse-engineered, creating a brand new meme. It would be as if the offspring of someone with strawberry-blonde hair also had strawberry-blonde hair by reverse-engineering up some genes for strawberry-blondeness, rather than inheriting either a redhead gene or a blonde gene to combine with a gene from the other parent. Memetics may still be model worth using, though. If meme-blending happens infrequently or not at all, it can be considered mere statistical noise. The one thing that is clear is that much more study is needed.

Please discuss.