Meme theory is probably the most well-known of the anthropological theories bouncing around the academy today, but it didn't develop in anthropology departments and is consequently not very well-known by anthropologists themselves. Since few anthropologists care about the problems meme theory purports to explain, they're also not very interested in approaching it. This is not entirely true; there are some anthropologists who have looked at meme theory, including especially Maurice Bloch, a Franco-British anthropologist at the London School of Economics. Dan Sperber, a brilliant anthropologist who has branched out into linguistics (introducing the influential theory of relevance into pragmatics) and much else, has also written extensively about the notion of memes.* But it is certainly the case that you won't learn anything at all about the idea behind memes - even a debunking of it - on an anthropology course.
The idea behind meme theory is that a lot of the things that people do come in packets that can be identified almost independently of surrounding cultural stuff, in the same way that genes can be at least somewhat understood when isolated from a genome (but not really). Memes are cultural bits that humans transmit to one another, allowing them to do things they couldn't or simply didn't do before. The standard criticism, certainly the one we find in Bloch, is that these cultural bits actually end up fitting together in unique ways when put in different contexts, so that dan dan mian is really quite different to spaghetti bolognese, even if they're part of a related tradition. Italian pasta is different to Chinese noodles because of the way these are used in the broader culture of Italy and China. Often there's an implicit holism to the claim that memes aren't real because they're different in different contexts. (I'm not sure if Bloch is endorsing this holism.) It should be mentioned, by the way, that meme theory has come up in the past, in the form of "traits". Listing the traits of a "primitive" human society was a common practice in nineteenth century anthropology (and placing them on an evolutionary scale was another), something that received the same criticisms from anthropologists that meme theory has.
Well, genes work differently depending on other things going on in the genome, and we can still identify and track genes in different organisms and see how their functions and attributes change. The idea that the HOX gene in tiktaalik is the same as in a human is a useful fiction designed to illustrate that the genes are both functionally and in origin effectively the same. In reality they are slightly different sequences that do slightly different things, but that originate from the same ultimate source - a common ancestor. The sequences do different things depending on other bits of content in the genome as well, which is precisely what Bloch is saying about the introduction of noodles to Italy. I think that if we bear this in mind and don't turn the word "meme" into the overused jargon that it has become in popular circles, there's no problem with this use of meme theory. Perhaps "trait" would be a more neutral word, but either way, the point is the same, and both come with considerable baggage. What's important isn't the label (that's not important at all), but the idea.
It was Dawkins who coined the term and the general concept, but meme theory is taken furthest and most realistically by Dan Dennett. Here's my very rough and ready reading of Dennett's general position on memes, comparing them with genes and beliefs: when we say that someone believes something, or that the HOX gene is the same in tiktaalik and humans, we are simplifying complex physical processes into easily parseable statements. "Believing" that Jesus is Lord is actually a very complicated process, and there isn't one single, say, neuron, that corresponds to this belief. Saying that someone believes that is an abstraction - a useful fiction - that we use to understand a process we can't see and can only have evidence of when the person acts. Genes and beliefs correspond (or appear to) with what is actually going on in the processes that form the organism or the processes that generate action respectively, but that doesn't mean they're really, really, really, real.** Every time we talk we talk about things that aren't metaphysically real entities - ie, things that are always unambiguous.
By the same reasoning as is applied to genes and beliefs, "culture", or what humans do, can be talked about using a variety of simplifying concepts, among them the concept of memes, which is like a gene but for human so-called "cultural traits". The difference is that human actions are even more complicated than genes - which is to say that human action, the basic bit of what a meme is, is even further removed from the physical processes that are actually going on. When we say something as ordinary as, "the man raised his hand", we're simplifying an enormously complicated physical process dependent on a huge range of variable factors into a short and intuitive sentence. Anyway, memes are a simplification of the even more complicated problem of tracking what people do over thousands of years and across all of earth's continents.
Here's an example of this kind of meme (actually, there are examples everywhere, because all words are, by this reckoning, memes). I was in a burger shop in Oxford (Pepper's Burgers on Walton Street) last year at about two in the morning after an evening in the pub ordering a gigantic burger with blue cheese. On the side of the joint was a magazine - I think it was about life in Oxfordshire and the fun stuff you can do there. A fairly dull magazine. The title of it was Ocelot. The word "ocelot" comes from Nahuatl (ōcēlōtl in the original Nahuatl), the language of the Nahuas (the tribes that founded the Aztec triple alliance, amongst other achievements), and it means a type of large, aggressive felid.
At some point, the Nahuatl word for this felid entered English, presumably (but not necessarily) via Spanish, and due to the fact that English and Spanish speakers do not typically use the phoneme "tɬ" in their speech, and due to their not distinguishing between long and short vowels, the word ōcēlōtl became "ocelot". It has also changed to fit the stress patterns of English and Spanish, too; in Nahuatl the stress on words is nearly always on the penultimate syllable (although, if I remember correctly, it changes with the addition of a long vowel at the end, when it becomes an final syllable stress). We now pronounce it as Ocelot, where before it was more like oo-see-LOO-tl.
And then, centuries after the introduction of this word into English, a person or group of people in Oxfordshire thought it would be a good idea to name a magazine devoted to life in the county after the Aztec word for a big cat.
The process behind Ocelot magazine gaining its name must have been vastly, vastly more complicated than is presented here, depending on a huge number of mental processes - amongst others - all of them fundamentally chemico-physical. It goes back further than the Aztecs, too; they must have got the word for the ocelot from somewhere (possibly proto-Uto-Aztecan - which must also have derived the word from something else). But the only way we can realistically talk about this process is by talking as if (ōcēlōtl = a stripy felid) and (Ocelot = an Oxfordshire magazine) come ultimately from the same source and are related, or are, basically, the same meme. They're clearly different, and the English use has clearly diverged from the Aztec in a fairly consistent manner.
Anyway, my point is that this is how we have to talk about these things. It's very difficult to talk about the spread of writing systems, for instance, without assuming that they are "memes"/"traits"/"representations" in some sense, or that we can talk about them in that way. Here, of course, we see that the basic idea of memes is sound; we can talk about how our writing system ultimately come from ancient Egypt, and we can talk about the evolution of this ancient-Egypt-writing meme as it came into the minds of groups of people with different backgrounds (and consequently different needs and beliefs) to the Egyptians, ending up generating a great diversity of scripts we call "related".
For instance, by way of a huge number of interactions lost to the mists of history, people in northeastern Egypt and southern Canaan developed a script called Proto-Sinaitic, which is an abjad deriving both the sounds and shapes of its letters from Egyptian hieroglyphs. These Egypt-border peoples did not have the scribal schools and international diplomatic needs of the Egyptians, and used this simplified, basic, phonetic script for transcribing their words on monuments, potsherds, and other materia. The script simplified further over time, as more and more people used the script more of the time, resulting in a desire for greater efficiency in writing (a greater effect-effort ratio, as Sperber calls it). This became what we know as the Phoenician script, and was carried across the seas by Phoenician sailors and on land by interactions with the Egyptian empire and people in Mesopotamia - including the Greeks by sea and the Aramaeans on land.
The Greeks found that the Phoenician abjad was incapable of writing Greek words legibly, because it doesn't mark vowels (being an abjad). Greek is an Indo-European language, where marking vowels is incredibly important (because vowel changes are often used to mark changes in tense and case; see sing-sang-sung). Phoenician is an Afroasiatic language, where meaning is mostly conveyed using a consonantal base (see the Arabic example of k-t-b, which is associated with reading and related activities and not much else; "kitaab" means book, "maktab" means school, etc). The meaning is in the consonants, and the vowels are (to some extent) incidental, which is why most Afroasiatic languages are written with abjads derived from Proto-Sinaitic and why most Indo-European languages are written with alphabets derived from Greek. When the Greeks came across the Phoenician abjad, they liked it, but changed it to fit their language. The Greeks invented the world's first true alphabet due to their inability to communicate with the pre-existing abjads.
And that is how we can talk about this immensely complex process in a reasonable way, while bearing in mind that it was an enormous series of interactions between people and the contents of their brains that was actually going on, not the transfer of independently existing, really-real "memes".
We can also talk similarly about languages, or even those insipid KEEP CALM AND CARRY ON signs that have become so popular in the past few years (there's now a frickin' music album inspired by that particular meme). I'm not sure it's necessary to call these things "memes" - it's quite an ugly word, and now it really means repetitive, idiotic internet comedy more than anything else. But the basic idea behind it isn't wrong, I don't think, and being ignorant of it doesn't seem productive.
Don't get me started on Susan Blackmore's The Meme Machine, though.
* Sperber calls his improvement on meme theory the "epidemiology of
representations"; that's basically the same thing as that which I have
presented, but Sperber is more convinced of the basic reality of
"representations" (his memes) than we'd find in Dennett's outline.
Either way, it's really the same idea, improved.
** Try Brainstorms and The Intentional Stance to see the genesis of this idea (they're also both excellent books). Tadeusz Zawidski's book on Dennett, called Dennett, also has a good overview of his views on memes.