The Bactra Review: Occasional and eclectic book reviews by Cosma Shalizi   61

Thought Contagion

How Belief Spreads Through Society

by Aaron Lynch

New York: Basic Books, 1996

The Case for the Meme's Eye View

``Meme'' is a word Richard Dawkins coined in The Selfish Gene (Oxford UP, 1976), by analogy with ``gene,'' to label cultural replicators --- ideas or behaviors capable, at least in principle, of propagating forever from one individual to another. The logic of the theory is straightforward --- neo-Darwinism abstracted from its original setting in population genetics --- and, after a few hundred pages of Dawkins hammering home the analogous message about genes, quite persuasive. Memes (by definition) influence how their carriers act. In particular, a meme can make its vectors behave in ways which make it more or less likely that the meme will be passed on to others. Those memes which tend to make their carriers pass them on better will spread at the expense of rival memes, and we should expect memes to produce complex adaptations --- behavioral adaptations --- which tend to promote their propagation. They are somewhere between cultural genes and viruses of the mind, so the metaphors used in talking about them are drawn from both genetics and epidemiology.

Dawkins is an extremely persuasive writer, as are some of those whom he has inspired to write about memes, most famously Douglas Hofstadter and Daniel Dennett. The notion of memes has led to a great deal of buzz and hand-waving and speculation, especially on the net, and even a decent sermon on tolerance. It makes a first-rate mind-toy. But some people want more, specifically an actual science of memetics, and at this point, if not before, they meet opposition. Memetics, an intelligent adversary might say, would not even be wrong. After all, social scientists and humanists have been looking at the transmission of folk-tales, myths, rumors, texts, mores, etc. for centuries. If it makes biologists and their sympathizers feel better to call all these things ``replicators,'' well and good; no doubt they can even fit some numbers to the replicator equation, if they have nothing better to do. But really, this is telling M. Jourdain he's been speaking prose, and does not add to knowledge. Memeticists would look at, e.g., 18th century Europe, observe the number of Newtonians increasing and the number of Cartesians dwindling, and conclude that there was selection for Newtonian memes and against Cartesian ones. But this would not explain why Newtonianism triumphed over Cartesianism, and would be entirely post hoc. The theory, to dignify it by the name, amounts to the tautology that what spreads, spreads.

One of the reasons I don't buy this is that very similar attacks are made against conventional, genetic natural selection, and are groundless. Believing in memes does reverse how one looks at beliefs and culture --- a pointed noted by many people, but most forcefully put by Dennett in his lecture on ``Memes and the Exploitation of the Imagination'' (parts of which are recycled in Consciousness Explained and Darwin's Dangerous Idea). Normally we take it as given that people believe an idea if they think it is true and follow it if they think it is right. From the meme's eye view, the fundamental tautology is that a meme spreads if it's good at replicating. Most of the time, memes replicate by seeming true or right (or, I suppose, beautiful), and then memetics doesn't add much to our understanding. But as Dennett says

[t]he theory becomes interesting only when we look at the exceptions, the circumstances under which there is a pulling apart of the two perspectives; only if meme theory permits us better to understand the deviations from the normal scheme will it have any warrant for being accepted.

Aaron Lynch's Thought Contagion is the first book I know of which attempts to meet this challenge. Lynch is an ex-physicist, and while I wish I could claim this shows in his work, it doesn't, really. He has written a mathematical paper on ``Units and Events of Replication'', which is excellent and quite devoid of metaphor, but there is no math in Thought Contagion, and it should be comprehensible to those with no scientific or technical background at all.

The plan of the book is as as follows. The first chapter introduces the idea of memes, which Lynch also calls ``thought contagions,'' and limits to ``actively contagious ideas,'' i.e. those whose effect on their carriers is to enhance their own probability of transmission. It also describes their modes of propagation, which Lynch claims are seven. Memes can spread, he says, by increasing the number of children of their believers; by increasing the likelihood that the infection is passed on to children; by encouraging the conversion of others or proselytizing; preserving believers by ``influenc[ing] adherents to live longer, or make them avoid dropping out''; by sabotaging or attacking competitors; by seeming true or at least plausible; and by seeming advantageous to believe or at least spread. Most of Lynch's emphasis falls on the first three modes.

The second chapter is a brief discussion of how memetics relates to the existing social sciences, as well as to sociobiology (poorly), evolutionary psychology, cognitive science (read: Hofstadter and Dennett) and the psychohistory of Isaac Asimov's Foundation novels.

Upon this plinth Lynch has set four figures which display the memetic art: ``Family Plans'' (family structure and the number of children), ``Sexually Transmitted Beliefs'' (memes relating to sex --- who, how, when, where and why), ``Successful Cults'' (western religions, including Islam but not Zoroastrianism, or indeed anything which started east of Mesopotamia), ``Prescription Beliefs'' (those relating to health and disease, with street gangs tossed in for good measure) and ``Controversy'' (miscellaneous burning issues of contemporary America --- talk radio, gun control, abortion). An epilogue considers what memetic theory has to say about itself, and after (to my mind) unconvincing invocations of Gödel and infinite regress, concludes that this is `` `formally undecidable'. ''

I closed the book with mixed feelings, much more mixed than I had anticipated --- I'd been expecting something either uniformly excellent or (just possibly) really horrible, and it was neither.

On the credit side of the balance-sheet I enter the following: The book is short, and a very easy read --- I went through it at something like a page a minute. If Lynch has any peculiar ideological axes to grind, he doesn't grind them here. (He is not religious, and is too intelligent to be taken in by Rush Limbaugh.) He succeeds at writing about ideas sine ire studioque, like an epidemiologist writing about germs, or an ethologist writing about animals. The tone throughout is detached and level-headed, with a few flashes of dry wit. Both the ``dawn of a new age'' and ``sinister manipulation by unseen agents'' strains of rhetoric, which mar much writing on memes, are absent. He at least professes to respect the existing social sciences, and one supposes the humanities as well, and, notwithstanding the extent to which these are merely the ``vain babblings, and oppositions of science falsely so called'',this is a good thing.

The debit side of the ledger begins here, because there are places where the book would have benefited from a closer acquaintance with those fields, especially history. The discussion of polygamy makes it intensely puzzling how polyandry could survive at all, much less persist for millennia, surrounded by polygamous cultures, as it has in parts of the Himalayas; that of hostility to Hispanic immigrants in modern America and complaints that they don't learn English (pp. 168--9) misses the fact that this is precisely our pattern of hostility to immigrants, formerly directed against speakers of Italian, German, Polish, Yiddish, etc. What particularly hit my buttons were the remarks on capitalism, socialism and communism in ``Memetics and Economics'' (pp. 18--21). The economics is not good --- his argument implies that the prevalence of salaried employment should be inversely related to economic growth --- and the history worse.

Related to this flaw, Lynch has a tendency to frame memetic justifications of the conventional wisdom about society, which he takes as a given. The most striking instance of this his argument that ``families with fathers present afford perhaps the strongest immunity to street gang memes'' (p. 149). This, he says, is because fathers are usually more able than mothers to physically protect their children against recruiting gangs. Now, it might be going too far to say that the conventional wisdom is always wrong, but it takes only a modest degree of skepticism and investigation to realize that it often is, especially when it is reduced to betting in favor of one middle-aged dad mano a mano against two or three or ten young thugs. Memetics should explain why conventional wisdom takes the form it does, rather than trying to rationalize it. * --- I hasten to add that this is just a tendency, and not even an over-riding one.

All this talk of conventional wisdom brings us to another flaw, which is the lack of consideration of what might be called professional or official sources of memes and beliefs --- churches, governments, schools, journalists, etc. Maybe (as is hinted in chapter two) these things can't be handled well by the natural-selection framework of memetics --- but then, so much the worse for memetics. In any case, it weakens some parts of the book, like ``Jingoism and Pacifism'' (pp. 169--173) considerably.

I have two more technical criticisms. One is about Lynch's emphasis on the transmission of memes within families, from parents to children --- call it familial transmission (Lynch doesn't). The problem with this is that it's slow, on the same time-scale as genetic evolution, whereas non-familial transmission of memes can be much faster. Consider two memes, A and B, which are respectively familial and non-familial. Meme A is under positive selection pressure, and the ratio of carriers to non-carriers increases by 5% per generation. Meme B is also selected for, and the ratio of its carrier to non-carriers increases by 5% per year. Over, let us say, 500 years (twenty generations), the ratio A/non-A will increase by a factor of about 2.7, but the ratio B/non-B would have been amplified by a factor just over 72 billion --- ``would have,'' since that growth rate couldn't be sustained. Unless the selection pressures on non-familial transmission are very weak or rapidly fluctuating, I'd expect them to trump those on familial transmission, and in discussing families patterns and sex Lynch leans on the latter.

The other criticism is that it is by no means obvious that ideas, behaviors and the like are stable replicators, as Lynch assumes both here and in ``Units and Events.'' But I've gone over this ground elsewhere, and shan't repeat the arguments here. Suffice it to say that the problem isn't addressed here, but is not necessarily fatal.

But that is the last of the debits, and there is one more large entry on the credit side: Lynch can be remarkably insightful. Some of his analyses --- of AIDS; of ``contagious cures,'' especially homeopathic ones; of what would happen to nootrophics if they actually worked; of breast-fetishes among straight men; of talk radio --- strike me as brilliant, almost certainly correct, and amenable to empirical testing.

This perhaps sounds more negative than it ought; the balance, for me, is strongly in the black. I have some technical problems with Lynch's book; there are very few numbers; it remains largely speculative and conjectural. But there is every reason to think that the theory can survive fixing those problems, and that the conjectures can be empirically, quantitatively tested. In advancing them, Lynch has claimed a large and specific territory for memetics, where it is supposed to be at least as illuminating as the conventional social sciences and history. None of this has been done before, or at any rate not so well, and it's a valuable service. It now remains for those devoted to memetics to ask those empirical questions and justify their claim to that territory. The time has come, in other words, to put up or shut up.


Disclaimer: Mr. Lynch was kind enough to discuss this review, in the most civilized way, when it was first written two years ago, pointing out some mis-understandings and mis-statements of his position; but I persist in disagreeing with him on some points, and I have no stake, financial or otherwise, in the success of this book.
xi + 192 pp., a few graphs, bibliography, no notes or citations of any kind (a barbarism to lay at the publisher's feet), index
Cultural Evolution, Memes, etc.
Currently in print as a hardback, US$24, ISBN 0-465-08466-4, forthcoming in December 1998 as a paperback, US$14, ISBN 0-465-08467-2, LoC HM251 L95 (and, as the author notes elsewhere, a new subject heading: ``Contagion (Social psychology)'' ).
7--8 December 1996, then lost on a floppy disk, found, revised and released 30 September 1998