Breaking the Spell

by Daniel C. Dennett

  Part of what is tantalizing about this research avenue is how nonreductionistic it is! McClenon and Hamer have worked independently of each other, so far as I know. Neither mentions the other, in any case, and neither is treated by Boyer or Atran or the other anthropologists. This is not surprising. The collaboration between geneticists and neurobiologists on the one hand and anthropologists and archeologists and historians on the other, pioneered by Luigi Luca Cavalli-Sforza and his colleagues, is a recent and spotty trend. False starts and disappointments are bound to outnumber triumphs in the early days of such interdisciplinary work, and I make no promises about the prospects of the specific hypotheses of either McClenon or Hamer. They make a vivid and accessible example of the possibilities in store, however. Recall Dawkins’s point quoted in chapter 3: “If neuroscientists find a ‘god center’ in the brain, Darwinian scientists like me want to know why the god center evolved. Why did those of our ancestors who had a genetic tendency to grow a god center survive better than rivals who did not?” (2004b, p. 14). We now have one eventually testable answer to Dawkins’s question, and it invokes not just biochemical facts but the whole world of cultural anthropology.6 Why did those with the genetic tendency survive? Because they, unlike those who lacked the gene, had health insurance! In the days before modern medicine, shamanic healing was your only recourse if you fell ill. If you were constitutionally impervious to the ministrations that the shamans had patiently refined over the centuries (cultural evolution), you had no health-care provider to turn to. If the shamans had not existed, there would have been no selection advantage to having this variant gene, but their accumulated memes, their culture of shamanic healing, could have created a strong ridge of selection pressure in the adaptive landscape that would not otherwise have been there.

  This still doesn’t get us to organized religion, but it does get us to what I am going to call folk religion, the sorts of religion that have no written creeds, no theologians, no hierarchy of officials.7 Before any of the great organized religions existed, there were folk religions, and these provided the cultural environment from which organized religions could emerge. Folk religions have rituals, stories about gods or supernatural ancestors, prohibited and obligatory practices. Like folk tales, the sayings of folk religion are of such distributed authorship that it is better to say that they have no authors at all, not that their authors are unknown. Like folk music, 8 the rituals and songs of folk religion have no composers, and their taboos and other moral injunctions have no legislators. Conscious, deliberate authorship comes later, after the designs of the basic cultural items have been honed and polished for many generations, without foresight, without intent, by nothing but the process of differential replication during cultural transmission. Is all this possible? Of course. Language is a stupendously intricate and well-designed cultural artifact, and no individual human designers get to take credit for it. And just as some of the features of written languages are clearly vestigial traces of their purely oral ancestors, 9 some of the features of organized religion will turn out to be vestigial traces of the folk religions from which they are descended. By vestigial traces, I mean this: the preservation over many generations of a folk religion—its self-replication in the face of inexorable competition—demands adaptations that are peculiar to an oral tradition and that are no longer strictly necessary (from a reverse-engineering point of view) but that persist simply because they haven’t yet been sufficiently costly to be selected against.

  5 Memory-engineering devices in oral cultures

  The total corpus of Baktaman knowledge is stored in 183 Baktaman minds, aided only by a modest assemblage of cryptic concrete symbols (the meanings of which depend on the associations built up around them in the consciousness of a few seniors) and by limited, suspicious communication with the members of a few surrounding communities.

  —Fredrik Barth, Ritual and Knowledge Among the Baktaman of New Guinea

  Humans, it appears, are the only animals that spontaneously engage in creative, rhythmic bodily coordination to enhance possibilities for cooperation (e.g., singing and swaying when they work together).

  —Scott Atran, In Gods We Trust

  Every folk religion has rituals. To an evolutionist, rituals stand out like peacocks in a sunlit glade. They are usually stunningly expensive: they often involve the deliberate destruction of valuable food and other property—to say nothing of human sacrifices—are often physically taxing or even injurious to the participants, and typically require impressive preparation time and effort. Cui bono? Who or what is the beneficiary of all this extravagant outlay? We have already seen two ways rituals might pay for themselves, as psychologically necessary features of divination techniques, or hypnotic induction procedures in shamanic healing.10 Once they were established on the scene for these purposes, they would be available to be adapted—exapted, as the late Stephen Jay Gould would say—for other uses. But there are other possibilities to explore.

  Anthropologists and historians of religion have theorized about the meaning and function of religious ritual for generations, usually from blinkered perspectives that ignore the evolutionary background. Before we look at speculations about rituals as symbolic expressions of one deep need or belief or another, we should consider the case that can be made for rituals as memory-enhancement processes, designed by cultural evolution (and not by any conscious designers!) to improve the copying fidelity of the very process of meme transmission they ensure. One of the clearest lessons of evolutionary biology is that early extinction lies in the future of any lineage in which the copying machinery breaks down, or even just degrades a little. Without high-fidelity copying, any design improvements that happen to occur in a lineage will tend to be frittered away almost immediately. Hard-won gains accumulated over many generations can be lost in a few faulty replications, the precious fruits of R & D evaporating overnight. So we can be sure that would-be religious traditions that have no good ways of preserving their designs reliably over the centuries are doomed to oblivion.

  We can observe today the birth and swift death of cults, as the early adherents lose faith or lose interest and drift away, leaving hardly a trace after a few years. Even when members of such a group fervently want to keep it going, their desires will be thwarted unless they avail themselves of the technologies of replication. Today, writing (not to mention videotape and other high-tech recording media) provides the obvious information highway to use. And from the earliest days of writing, there has been a keen appreciation of the need not only to protect the sacred documents from damage and decay, but to copy them over and over, minimizing the risk of loss by ensuring that multiple copies were distributed around. For many centuries before the invention of movable type, which made possible for the first time the mass production of identical copies, roomfuls of scribes, shoulder to shoulder at their writing desks, took dictation from a reader and thus turned one frail and dog-eared copy into dozens of fresh new copies—a copy machine made of people. Since the originals from which the copies were made have mostly turned to dust in the meantime, without the efforts of these scribes we would have no reliable texts for any of the literature of antiquity, sacred or secular, no Old Testament, no Homer, no Plato and Aristotle, no Gilgamesh. The earliest known copies of Plato’s dialogues still in existence, for instance, were created centuries after his death, and even the Dead Sea Scrolls and the Nag Hammadi gospels (Pagels, 1979) are copies of texts that were composed hundreds of years earlier.

  A text inked on papyrus or parchment is like the hard spore of a plant that may lie undamaged in the sand for centuries before finding itself in suitable conditions to shed its armor and sprout. In oral traditions, in contrast, the vehicle—a spoken verse or sung refrain—lasts for only a few seconds, and must enter some ears—as many as possible—and imprint itself firmly in as many brains as possible, if it is to escape oblivion. Getting registered in a brain—getting heard and not
iced above the competition—is less than half the battle. Getting rehearsed and rehearsed, either in the privacy of a single brain or in unison public repetition, is a life-or-death matter for an orally transmitted meme.11

  If you want to brush up your memory of the order of worship in your church’s Sunday service, or check to see whether one should stand or sit down during the closing benediction, there is almost certainly a text you can consult. The details are printed in the back of every hymnal, perhaps, or in the Book of Common Prayer, or, if not there, at least in texts that are readily available to the priest or minister or rabbi or imam. Nobody has to memorize every line of every invocation, every prayer, every detail of the costumes, music, manipulation of sacred objects, and so forth, since they are all written down in one official record or another. But rituals are not by any means restricted to literate cultures. In fact, the religious rituals of nonliterate societies are often more detailed, typically much more demanding physically, and just plain longer in duration than the rituals of organized religions. Moreover, the shamans don’t go to official shaman-seminaries, and there is no Council of Bishops or Ayatollahs to maintain quality control. How do the members of these religions keep all the details in memory over the generations?

  A simple answer is: They don’t! They can’t! And it is surprisingly hard to prove otherwise. Whereas members of a nonliterate culture may be well-nigh unanimous in their conviction that their rituals and creeds have been perfectly preserved by them over “hundreds” or “thousands” of generations (a thousand years is only about fifty generations), why should we believe them? Is there any evidence that supports their traditional conviction? There is a little.

  Much of the excitement that accompanied scholars’ discovery of the Nambudiri ritual tradition turned on the fact that although texts delineating Vedic rituals exist, the Nambudiri have not used them. Exclusively by non-literate means, they have sustained this elaborate ritual tradition with astonishing fidelity (as gauged by the centuries-old śrauta śutras). [Lawson and McCauley, 2002, p. 153]

  So at first it appears that the Nambudiri are perhaps a uniquely lucky oral culture, having some evidence in support of their conviction that they have preserved their rituals intact. If it were not for the Vedic texts, presumably unknown to them and never consulted over the years, there would be no fixed yardstick against which to measure their confidence in the antiquity of their traditions. But, alas, the story is too good to be entirely true. The Nambudiri tradition may be oral, but they are not illiterate (some of their priests teach engineering, for instance), and it is hard to believe they have kept themselves entirely isolated from the Vedic texts. “It is known that during their six-month initiation period of the training, preparation and rehearsals leading up to the actual event, use is made of notebooks, prepared by the senior AcAryas who have already taken part in previous rituals….”12 So the Nambudiri are not really an independent benchmark of how accurate oral transmission can be.

  Compare the problem here with the ongoing research on the evolution of languages. Using complex and sophisticated probabilistic analyses, linguists can deduce features of extinct oral languages whose last speakers have been dead for millennia! How can this be done with no tape recordings to consult and no texts in the language they are extrapolating? The linguists make heavy use of the enormous corpus of textual data in other, later languages, tracing linguistic shifts from Attic Greek to Hellenistic Greek, and from Latin to the Romance languages, and so forth. Finding common patterns in these shifts, they have been able to extrapolate back with some confidence to what languages must have been like before writing came along to fossilize some of them for later ages to study. They have been able to extract regularities of pronunciation shift and grammatical shift, and juxtapose them on patterns of stability, to arrive at highly educated and cross-confirmed guesses about how, say, Indo-European words were pronounced long before there were written languages to preserve the clues like fossil insects in amber.13

  If we tried to do the same extrapolation trick with religious beliefs, we would first have to establish benchmarks for stability and shift in them, and so far this has not proved feasible. What little we know about early religions is almost entirely dependent on surviving texts. Pagels (1979) offers a fascinating perspective on the Gnostic Gospels, for instance, early competitors for inclusion in the canon of Christian texts, thanks to the fortuitous survival of written texts that have been passed on as translations of copies of copies…of the originals.

  We cannot, then, just take it on faith that nonliterate religious traditions still extant in the world are as ancient as advertised. And we already know that in some such religions there is not a tradition of obsessive preservation of ancient creed. Fredrik Barth, for instance, found lots of evidence of innovation among the Baktamans, and as Lawson and McCauley (2002, p. 83) dryly note, “Perfect fidelity to past practice is not an unwavering ideal for the Baktamans.” So, whereas we can be quite sure that people in oral traditions have had religion of one sort or another for thousands of years, we shouldn’t ignore the possibility that the religion we see (and record) today may consist of elements that have been invented or reinvented quite recently.

  People run and jump and throw stones pretty much the same way everywhere, and this regularity is explained by the physical properties of human limbs and musculature and the uniformity of wind resistance around the globe, not a tradition somehow passed down from generation to generation. On the other hand, where no such constraints ensure reinvention, items of culture will be able to wander swiftly, widely, and unrecognizably in the absence of mechanisms of copying fidelity. Different strokes for different folks.14 And wherever this wandering transmission occurs, there will automatically be selection for mechanisms that enhance copying fidelity whenever they arise, whether or not people care, since any such mechanisms will tend to persist longer in the cultural medium than alternative (and no less costly) mechanisms that get themselves copied indifferently.

  One of the best ways of ensuring copying fidelity over many replications is the “majority rule” strategy that is the basis for the uncannily reliable behavior of computers. It was the great mathematician John von Neumann who saw a way of applying this trick in the real world of engineering, so that Alan Turing’s imaginary computing machine could become a reality, permitting us to manufacture highly reliable computers out of unavoidably unreliable parts. Practically perfect transmission of trillions of bits is routinely executed by even the cheapest computers these days, thanks to “von Neumann multiplexing,” but this trick has been invented and reinvented over the centuries in many variations. In the days before radio communication and GPS satellites, navigators used to take not one or two but three chronometers aboard their ships on long voyages. If you have just one chronometer and it starts running slow or fast, you’ll never know it is in error. If you bring two and they eventually begin to disagree, you won’t know whether one is running slow or the other is running fast. If you bring three, you can be quite sure that the odd one out is the one in which the error is accumulating, since otherwise the two that are still in agreement would have to be going bad in exactly the same way, an unlikely coincidence under most circumstances.

  Long before it was consciously invented or discovered, this Good Trick was already embodied as an adaptation of memes. It can be seen at work in any oral tradition, religious or secular, in which people act in unison—praying or singing or dancing, for instance. Not everybody will remember the words or the melody or the next step, but most will, and those who are out of step will quickly correct themselves to join the throng, preserving the traditions much more reliably than any of them could do on their own. It doesn’t depend on virtuoso memorizers scattered among them; nobody needs to be better than average. It is mathematically provable that such “multiplexing” schemes can overcome the “weakest link” phenomenon, and make a mesh that is much str
onger than its weakest links. It is no accident that religions all have occasions on which the adherents come together to act in public unison in rituals. Any religion without such occasions would already be extinct.15

  A public ritual is a great way of preserving content with high fidelity, but why are people so eager to participate in rituals in the first place? Since we are presuming that they are not intent on preserving the fidelity of their meme-copying by constituting a sort of social computer-memory, what motivates them to join in? Here there are currently a welter of conflicting hypotheses that will take some time and research to resolve, an embarrassment of riches in need of culling.16 Consider what we can call the shamanic-advertising hypothesis. Shamans the world over conduct much of their medicine in public ceremonies, and they are adept at getting the local people not just to watch while they induce a trance in themselves or their clients but to participate, with drumming, singing, chanting, and dancing. In his classic Witchcraft, Oracles and Magic Among the Azande (1937), the anthropologist Edward Evans-Pritchard vividly describes these proceedings, observing how the shaman cleverly enlists the crowd of knowing onlookers, turning them into shills, in effect, to impress the uninitiated, for whom this ceremonial demonstration is a novel spectacle.

  It may be supposed, indeed, that attendance at them has an important formative influence on the growth of witchcraft beliefs in the minds of children, for children make a point of attending them and taking part in them as spectators and chorus. This is the first occasion on which they demonstrate their belief, and it is more dramatically and more publically affirmed at these séances than in any other situation. [Evans-Pritchard, 1937 (1976 abridged edition, pp. 70–71)]