These notes were written in the late 1990s (1998, I think), following an interesting debate between Peter Godfrey-Smith and Daniel Dennett at the annual Colloquium hosted by the Department of Philosophy at UNC Chapel Hill. The topic was the potential role of memes in explaining cultural evolution and human behavior. I'm somewhat agnostic on whether the gene/meme analogy will prove theoretically fruitful but, like Dennett, I think the idea is well worth exploring.
Update, March 2018: Karolin Lohmus has done me the honor of translating my notes into Estonian. You can find her translation here.
As a point of entry, let me take an argument that Peter made. He said he thought that traditional social science explanations were likely to work better than “meme's eye” explanations for the bulk of cultural changes. He further said that although there may be some subset of cultural phenomena for which meme-based explanation would be superior, this did not license the conclusion that the explanation of cultural change ought to be unified under the hegemony of memes. He seemed to envisage “peaceful co-existence” between the two sorts of explanation, with memes confined to a rather small sub-domain.
In reply, Dennett said that where “social science” explanations do work well (for my purposes here I'll take the notion of social science explanation to be more or less synonymous with rational choice explanation, though that is perhaps an over-simplification), they can be subsumed or carried over wholesale into a meme-based view; and that leaves the explanation of the subset of cultural phenomena that are not amenable to rational choice explanation as definite “value added” on the part of the meme theory.
Some people then said (and Peter seemed to agree) that this move on Dennett's part amounted to “theft versus honest toil.”
I have a different take on this. I think the most fundamental objection to Peter's line of argument is that rational choice explanations simply take for granted the existence of rational choosers (persons with highly developed sets of preferences and highly developed means of calculating the satisfaction of those preferences), while meme theory holds out the possibility (“the promise” would perhaps be too strong at this point) of explaining the emergence of such choosers, as very particular sorts of meme colonies. If it proves possible to make good on this (the task has two aspects: explaining the historic emergence of reasonably rational, social persons in the first place; and explaining the recurrent creation of such persons out of the material provided by each generation of baby humans) then this would provide a principled means of “appropriating” rational choice explanation of some aspects of cultural evolution, and putting it in its place within the context of a broader and more fundamental meme-based theory.
(A word on the notion in the background here: There exists a wide explanatory gap between the traditional, biological level of Darwinian analysis - which takes us a long way, but by no means all the way, in the explanation of the emergence of humanity as we know it - and the level of rational choice analysis which just presupposes fully-formed rational choosers. The claim is that neither biology nor social science, as traditionally understood, holds out any hope of closing the gap. This creates a potential space for meme theory to claim - although whether that theory will prove capable of spanning the gap, in anything like its present form, remains to be seen.)
Of course there remains a good deal of honest toil to be done on behalf of meme theory in this regard. Not only is it necessary to explain the emergence of rational choosers; it is also necessary to explain the (observed degree of) robustness of such choosers (such highly structured meme complexes) in the face of potential subversion by “opportunistic” memes.
The problem here may be illustrated by reference to the prediction Dennett admitted to having made, that the massive publicity gained by Heaven's Gate would lead to a bout of replication of the suicide cult meme. In making that prediction Dennett was implicitly reckoning that the suicide cult meme would be able (for a significant number of hosts) effectively to overcome the biologically inculcated prejudice in favour of staying alive, and the ordinary day-to-day meme-set of calculation of rational advantage. That would imply that the latter meme-set is rather fragile. But if it's (potentially) fragile, how legitimate is it to serve up rational choice explanations that presuppose it (as part of a meme-based view)? We would need a more extended explanation of the form: “X happened because it was the rational thing for the people in question to do, given their preferences and beliefs - and here's why we're justified in supposing that people would continue to do the rational thing (to retain their integrity as rational choosers) in the circumstances.”
Actually this is a bit more complicated. Given that we are constituted as reasonably rational agents, this creates a particular “playing field” upon which memes must operate: for the most part they will be obliged not just to contradict our rational choices, to “get us to do what they want”, but rather to shape those choices. The suicide cult meme would presumably not get very far if it merely implanted a desire to commit suicide, in isolation from the rest of our belief-desire economy. We'd think, “Wait a minute, why on earth do I want to do that?”. It must implant a set of beliefs that rationalizes the desire to commit suicide (certain benefits will accrue). Nonetheless, the required beliefs in this case are aberrant; they do not cohere with our broader set of “healthy” beliefs, and in that sense we can talk of such a meme facing the task of compromising our rationality.
Thus the “more extended explanation” that I mentioned has to be augmented: “X happened because it was the rational thing for the people in question to do, given their preferences and a set of beliefs reasonably appropriate to their situation - and here's why we're justified in supposing that people would continue to do the rational thing in the circumstances, and not to be prey to aberrant beliefs.”
Heaven's Gate leads me to a further point, in response to a question from the audience at Chapel Hill: Someone asked the (good) question, “In the case of memes, how can we identify the counterpart to the enhancement of fitness in the case of biological mutations?” We want to avoid the circularity that turns “survival of the fittest” into the tautologous “survival of the survivors”. It's tricky: We want a concept of fitness that does not simply collapse into “differential replicative success”, but that nonetheless remains very close to the latter. (E.g., we do not want to say that a mutation is fitness-enhancing when it provides a solution to a design problem facing an organism that is superior from an engineering point of view, if the cost of this solution is too high. We want the overall cost-benefit balance to be “right” before we talk of enhanced fitness, where the costs and benefits ultimately come down to replicative prowess.)
The circle can be broken via probabilistic reasoning: fitness is enhanced where, given the environment and so on, the probability of differential replicative success is raised. This leaves open the possibility that superior fitness may be nullified by bad luck (or a mediocre fitness score compensated by good luck) in any particular case: it's just that we don't expect this to happen very often. Or rather: we expect it to happen all the time, but we rely upon the Law of Large Numbers to ensure that it will happen on a significant scale - large enough to influence the trajectory of the evolutionary process - with extreme rarity.
What goes for biology here can equally well go for memes. The fitness of a meme is high if its probability of replicative success is high (although it may happen to bomb all the same). “Constructing” such probabilities may not be easy, but then neither is it always easy in biology.
Here's a sample argument: the meme of religious martyrdom is fitter than that of the suicide cult. This is because martyrdom is part of a larger and more powerful meme complex, which may be analysed along the same lines as phenomena of “exaggerated display” in biology, which at first appear paradoxical since the display features appear to disadvantage the organism in question. Martyrdom is a religion meme saying “Look how powerful I am: I can afford to throw away some of my hosts and still thrive. Don't you want to be part of something that powerful?” Or “Look how compelling I am: I can instill in some of my hosts the desire to lay down their lives in my cause. You don't have to go that far, but shouldn't you too be compelled?” The suicide cult meme is weaker since it lacks the accompaniment of a Cause for which the host is called upon to die; without a “big idea”, its transmission, for the most part, depends on personal exposure to a highly charismatic local leader. Both the martyrdom meme and the suicide cult meme have uphill work to do in recruiting any hosts, but the martyrdom meme has a definite advantage in its articulation with a Grand Cause meme.
Chemistry “gave birth to” and sustains the biological. But once the biological level has emerged, a space is created for specifically biological concepts and explanations, which cannot meaningfully be reduced to sentences of Chemistry. Similarly, might one say that if memetic evolution gives rise to a population of rational choosers, there emerges a space for rational choice explanations that cannot meaningfully (or perspicuously, at any rate) be reduced to sentences of Memetics? In the same way that a “chemical” account of the Darwinian process of natural selection is impossible (just at the wrong level), so perhaps a memetic account of, say, the microeconomic competition process may be impossible.
It seems to me that such a possibility does not threaten the memes project. We could have peaceful coexistence of social science explanation and meme-based explanation, but, from this perspective, we should not think of the two sorts of explanations as being “at par”, and simply covering different sub-domains of human culture, but rather as articulated in a particular way. And of course, just as there are areas in biology where the chemistry does “rise to the surface”, so to speak, there would be areas of social explanation where discussion of rational behaviour has to be intertwined with talk of memes.
I suppose that good starting points for meme-based analysis would be areas of human culture where rational choice explanations have definite problems. Standard Darwinian evolution in biology is - as Dennett has stressed - a bodge, with no foresight. A marvellously successful bodge, but still a bodge, and this leaves definite tell-tale signs. In the same way one presumes that the construction of rational choosers out of human biological raw material, via memetic evolution, would also be something of a bodge job, leaving odd loose ends and gaps where rational choice explanation breaks down.
It's not just that some aspects of culture fall outside of the purview of rational calculation - as in Peter's example of accents, which, for the most part, might be considered as “don't cares” of cultural evolution. (It would be quite disappointing to the meme ethusiast if memetic analysis turned out to be good only for explaining the trajectory of “don't care” cultural items.) It's also that some aspects of cultural evolution might exhibit definite “irrationality”, from a standard viewpoint.
Here's one attempt to spot a few such points of leverage.
I wish I had time to pursue some of these myself. Maybe someday.
Here is a further thought, related to but not integrated with the previous ones.
The general form of a rational choice explanation is: people did X because they perceived it as being in their best interests to do X (because doing X was seen, in the light of the beliefs of the people concerned, as most conducive to the satisfaction of their preferences).
But for such an account to figure as an “explanation” (an answer to a question we care to ask) certain conditions are required. Specifically, the set of beliefs in question, or the set of preferences, must be (a) somewhat unfamiliar to those for whom the account constitutes an explanation, but at the same time (b) not excessively outlandish.
“Why did personal computers plus word processing software displace electric typewriters in the early to mid 1980s?” A rational choice explanation of this would be redundant, for people alive today who understand the terms of the question. The advantages of word processing, once personal computers became cheap enough, are simply too obvious. We may feel we have been offered a real explanation, however, when we are given a persuasive account of some cultural phenomenon in terms of a set of beliefs that is unfamiliar (by reason, perhaps, of our distance - either temporal, or in geographical or social space - from the decision situation in question), or a set of preferences that is unfamiliar (for the same sort of reason). Why did some distant tribe carry out such-and-such a cultural practice? Perhaps the anthropologist can explain this to us in terms of the particular beliefs and preferences of the tribe in question, so that once the explanation is grasped, we say to ourselves, “Ah yes, I can see that if they thought this, and wanted that, this practice would make sense.”
But note that if the beliefs or preferences that are supposed to be operative are too weird, we are not going to find the putative explanation satisfactory. Suppose someone tries to explain the Heaven's Gate suicides simply by reference to the beliefs and desires of the cult members (“they thought they would achieve this advantage by killing themselves”). Are we going to say, “OK, now I understand that given their beliefs this was their best course of action”? Surely not. We're going to ask, “But how come they ended up believing that crazy stuff?”. Similarly, if a rational choice account were to appeal to masochistic preferences, we would generally be disinclined to accept it as an explanation, and inclined to demand a further account of where such self-destructive preferences are supposed to have come from.
Perhaps it is at such points that a specifically memetic analysis can “interface” with a rational choice account.