Breaking the Spell

by Daniel C. Dennett

  Researchers have several other terms for the intentional stance. Some call it “theory of mind” (Premack and Woodruff, 1978; Leslie, 1987; Gopnik and Meltzoff, 1997), but there are problems with that formulation, so I am going to stick with my more neutral terminology.5 Whenever an animal treats something as an agent, with beliefs and desires (with knowledge and goals), I say that it is adopting the intentional stance or treating that thing as an intentional system. The intentional stance is a useful perspective for an animal to take in a hostile world (Sterelny, 2003), since there are things out there that may want it and may have beliefs about where it is and where it is heading. Among the species that have evolved the intentional stance, there is considerable variation in sophistication. Faced with a threatening rival, many animals can make an informationally sensitive decision either to retreat or to call the other’s bluff, but there is scant evidence that they have any sense of what they are doing or why. There is some (controversial) evidence that a chimpanzee can believe that another agent—a chimpanzee or a human being, say—knows that the food is in the box rather than in the basket. This is second-order intentionality (Dennett, 1983), involving beliefs about beliefs (or beliefs about desires, or desires about beliefs, etc.), but there is no evidence (yet) that any nonhuman animal can want you to believe that it thinks you are hiding behind the tree on the left, not the right (third-order intentionality). But even preschool children delight in playing games in which one child wants another to pretend not to know what the first child wants the other to believe ( fifth-order intentionality): “You be the sheriff, and ask me which way the robbers went!”

  Whatever the situation is with nonhuman animals—and this is a topic of vigorous and hotly debated research6—there is no doubt at all that normal human beings do not have to be taught how to conceive of the world as containing lots of agents who, like themselves, have beliefs and desires, as well as beliefs and desires about the beliefs and desires of others, and beliefs and desires about the beliefs and desires that others have about them, and so forth. This virtuoso use of the intentional stance comes naturally, and it has the effect of saturating the human environment with folk psychology (Dennett, 1981). We experience the world as not just full of moving human bodies but of rememberers and forgetters, thinkers and hopers and villains and dupes and promise-breakers and threateners and allies and enemies. Indeed, those human beings who find perceiving the world from this perspective difficult—those suffering from autism are the best-studied category—have a more significant disability than those who are born blind or deaf (Baron-Cohen, 1995; Dunbar, 2004).

  So powerful is our innate urge to adopt the intentional stance that we have real difficulty turning it off when it is no longer appropriate. When somebody we love or even just know well dies, we suddenly are confronted with a major task of cognitive updating: revising all our habits of thought to fit a world with one less familiar intentional system in it. “I wonder if she’d like…,” “Does she know I’m…,” “Oh, look, this is something she always wanted….” A considerable portion of the pain and confusion we suffer when confronting a death is caused by the frequent, even obsessive, reminders that our intentional-stance habits throw up at us like annoying pop-up ads but much, much worse. We can’t just delete the file in our memory banks, and, besides, we wouldn’t want to be able to do so. What keeps many habits in place is the pleasure we take from indulging in them.7 And so we dwell on them, drawn to them like a moth to a candle. We preserve relics and other reminders of the deceased persons, and make images of them, and tell stories about them, to prolong these habits of mind even as they start to fade.

  But there is a problem: a corpse is a potent source of disease, and we have evolved a strong compensatory innate disgust mechanism to make us keep our distance. Pulled by longing and pushed back by disgust, we are in turmoil when we confront the corpse of a loved one. Small wonder that this crisis should play so central a role in the birth of religions everywhere. As Boyer (2001, p. 203) stresses, something must be done with a corpse, and it has to be something that satisfies or allays competing innate urges of dictatorial power. What seems to have evolved everywhere, a Good Trick for dealing with a desperate situation, is an elaborate ceremony that removes the dangerous body from the daily environment either by burial or burning, combined with the interpretation of the persistent firing of the intentional-stance habits shared by all who knew the deceased as the unseen presence of the agent as a spirit, a sort of virtual person created by the survivors’ troubled mind-sets, and almost as vivid and robust as a live person.

  What role, if any, does language play in this? Are we the only species of mammal that buries its dead because we’re the only species that can talk about what we share when we confront a fresh corpse? Do the burial practices of Neanderthals show that they must have had fully articulate language? These are among the questions we should try to answer. The world’s languages are well stocked with verbs for the basic varieties of belief-desire manipulation: we pretend and lie, but also bluff and suspect and flatter and brag and tempt and dissuade and command and prohibit and disobey, for instance. Was our virtuosity as natural psychologists a prerequisite for our linguistic ability, or is it the other way around: did our use of language make our psychological talents possible? This is another controversial area of current research, and probably the truth is, as it so often is, that there was a coevolutionary process, with each talent feeding off the other. Plausibly, the very act of verbal communication requires some appreciation of third-order intentionality: I have to want you to recognize that I am trying to inform you, to get you to believe what I’m saying (Grice, 1957, 1969; Dennett, 1978; but see also Sperber and Wilson, 1986). But, like the fledgling cuckoo, a child can get under way quite cluelessly, achieving successful communication without having any reflective appreciation of the structure that underlies all intentional communication, without even recognizing, really, that she is communicating at all.

  Once you’ve started talking (with other people), you will be bathed in new words, some of which you more or less understand; some of these objects of perception, such as the words “pretend” and “brag” and “tempt,” will help draw and focus your attention on cases of pretending and bragging and tempting, giving you plenty of inexpensive practice in folk psychology. Whereas chimpanzees and some other mammals may also be “natural psychologists,” as Nicholas Humphrey (1978) has called them, since they lack language they never get to compare notes or discuss cases with other natural psychologists. The articulation of the intentional stance in verbal communication not only heightens the sensitivity, discrimination, and versatility of individual folk psychologists, but also magnifies and complicates the folk-psychological phenomena they are attending to. A fox may be cunning, but a person who can flatter you by declaring that you are cunning as a fox has more tricks up his sleeve than the fox does, by a wide margin.

  Language gave us the power to remind ourselves of things not currently present to our senses, to dwell on topics that would otherwise be elusive, and this brought into focus a virtual world of imagination, populated by the agents that mattered the most to us, both the living but absent and the dead who were gone but not forgotten. Released from the corrective pressure of further actual encounters in the real world, these virtual agents were free to evolve in our minds to amplify our yearnings or our dreads. Absence makes the heart grow fonder, or—if the absent one was somewhat frightening in reality—more terrified. This still doesn’t get our ancestors to religion, but it gets them to persistent—even obsessive—rehearsal and elaboration of some of their habits of thought.

  Chapter 4 Extrapolating back to human prehistory with the aid of biological thinking, we can surmise how folk religions emerged without conscious and deliberate design, just as languages emerged, by interdependent processes of biological and cultural evolution. At the root of human be
lief in gods lies an instinct on a hair trigger: the disposition to attribute agency—beliefs and desires and other mental states—to anything complicated that moves.

  Chapter 5 The false alarms generated by our overactive disposition to look for agents wherever the action is are the irritants around which the pearls of religion grow. Only the best, most mind-friendly variants propagate, by meeting—or seeming to meet—deep psychological and physical needs, and then these are further refined by the incessant pruning of selection processes.

  CHAPTER FIVE Religion, the Early Days

  1 Too many agents: competition for rehearsal space

  I might repeat to myself slowly and soothingly, a list of quotations beautiful from minds profound—if I can remember any of the damn things.

  —Dorothy Parker

  What start as useful luxuries that give you an edge in a fast-moving world have a way of evolving into necessities. Today, we all wonder how we could live without our telephones, our driver’s licenses, our credit cards, our computers. So it once was with language, and the intentional stance. What started as a Good Trick rapidly became a practical necessity of human life, as our ancestors became more and more social, more and more linguistic. And, as already noted for the simpler case of the HADD, there is the possibility of too much of a good thing. The continued experience of the presence of departed acquaintances as ghosts is not the only overshooting of the intentional stance in the lives of our ancestors. The practice of overattributing intentions to moving things in the environment is called animism, literally giving a soul (Latin, anima) to the mover. People who lovingly cajole their cranky automobiles or curse at their computers are exhibiting fossil traces of animism. They probably don’t take their own speech acts entirely seriously, but are just indulging in something that makes them feel better. The fact that it does tend to make them feel better, and is apparently indulged in by people of every culture, suggests how deeply rooted in human biology is the urge to treat things—especially frustrating things—as agents with beliefs and desires. But if our bouts of animism today tend to be ironic and attenuated, there was a time when the desire of the river to flow to the sea, and the benign or evil intent of the rain clouds, were taken so literally and seriously that they could become a matter of life or death—for instance, to those poor souls who were sacrificed to appease the insatiable desires of the rain god.

  Simple forms of what we might call practical animism are arguably not mistakes at all, but extremely useful ways of keeping track of the tendencies of designed things, living or artifactual. The gardener who tries to discover what her different flowers and vegetables prefer, or tricks a dogwood branch into thinking it’s spring and opening its buds by bringing it indoors, where it is warm, doesn’t have to go overboard and wonder what her petunias are daydreaming about. Even undesigned physical systems can sometimes be usefully described in intentional or animistic terms: the river doesn’t literally want to return to the ocean, but water seeks its own level, as they say, and lightning searches for the best path to ground. It is not surprising that the attempt to explain patterns discerned in the world has often hit upon animism as a good—actually predictive—approximation of some unimaginably complex underlying phenomenon.

  But sometimes the tactic of seeking an intentional-stance perspective comes up dry. Much as our ancestors would have loved to predict the weather by figuring out what it wanted and what beliefs it harbored about them, it simply didn’t work. It no doubt often seemed to work, however. Every now and then the rain dances were rewarded by rain. What would the effect be? Many years ago, the behaviorist psychologist B. F. Skinner (1948) showed a striking “superstition” effect in pigeons that were put on a random schedule of reinforcement. Every so often, no matter what the pigeon was doing at the moment, a click and a food-pellet reward were delivered. Soon the pigeons put on this random schedule were doing elaborate “dances,” bobbing and whirling and craning their necks. It’s hard to resist putting a soliloquy into these birds’ brains: “Now, let’s see: the last time I got the reward, I’d just spun around once and craned my neck. Let’s try it again…. Nope, no reward. Perhaps I didn’t spin enough…. Nope. Perhaps I should bob once before spinning and craning…. YESSS! OK, now, what did I just do?…” You don’t have to have language, of course, to be vulnerable to such enticing illusions. The soliloquy dramatizes the dynamics that produce the effect, which doesn’t require conscious reflection, just reinforcement. But in a species that does represent both itself and other agents to itself, the effect can be multiplied. If such a strikingly extravagant behavioral effect can be produced in pigeons by making them wander into a random-reinforcement trap, it is not hard to believe that similar effects could be inculcated by happy accident in our ancestors, whose built-in love for the intentional stance would tend to encourage them to add invisible agents or other homunculi to be the secret puppeteers behind the perplexing phenomena. Clouds certainly don’t look like agents with beliefs and desires, so it is no doubt natural to suppose that they are indeed inert and passive things being manipulated by hidden agents that do look like agents: rain gods and cloud gods and the like—if only we could see them.

  This curiously paradoxical idea—something invisible that looks like a person (has a head, eyes, arms and legs, perhaps wears a special helmet)—is different from other self-contradictory combinations. Consider the idea of a box that has no interior space to put things in, or a liquid that isn’t wet. To put it crudely, these ideas are not interesting enough to be puzzling for very long. Some nonsense is more attention-grabbing than other nonsense. Why? Just because our memories are not indifferent to the content of what they store. Some things we find more memorable than others, and some things are so interesting that they are well nigh unforgettable, and still others, such as the random string of words “volunteers trainer regardless court exercise” (pulled by me “at random” from the first newspaper story I could lay my hands on just now), could be remembered for more than a few seconds only if you either deliberately repeated it to yourself dozens of times or made up some interesting story that somehow made sense of these words in just this sequence.

  We are all painfully aware today that our attention is a limited commodity with many competitors vying for more than their share. This information overload, with advertisements bombarding us on all sides, plus a host of other distractions, is nothing new; we’ve just become self-conscious about it, now that there are thousands of people who specialize in designing novel attention-grabbers and attention-holders. We—and, indeed, all animate species—have always had to have filters and biases built into our nervous systems to screen the passing show for things worth hanging on to, and these filters favor certain sorts of exceptions or anomalies. Pascal Boyer(2001) calls these exceptions counterintuitive, but he means this in a rather circumscribed technical sense: counterintuitive anomalies are especially attention-worthy and memorable if they violate just one or two of the basic default assumptions about a fundamental category like person or plant or tool. Concoctions that aren’t readily classifiable at all because they are too nonsensical can’t hold their own in the competition for attention, and concoctions that are too bland are just not interesting enough. An invisible ax with no handle and a spherical head is just irritating nonsense, an ax made of cheese is a bit titillating (there are conceptual artists who make a good living coming up with such japes), but a talking ax—ah, now we’ve got something to hold the attention!

  Put these two ideas together—a hyperactive agent-seeking bias and a weakness for certain sorts of memorable combos—and you get a kind of fiction-generating contraption. Every time something puzzling happens, it triggers a sort of curiosity startle, a “Who’s there?” response that starts churning out “hypotheses” of sorts: “Maybe it’s Sam, maybe it’s a wolf, maybe it’s a fallin
g branch, maybe it’s…a tree that can walk—hey, maybe it’s a tree that can walk!” We can suppose that this process almost never generates anything with any staying power—millions or billions of little stretches of fantasy that almost instantly evaporate beyond recall until, one day, one happens to occur at just the right moment, with just the right sort of zing, to get rehearsed not just once and not just twice, but many times. A lineage of ideas—the walking-tree lineage—is born. Every time the initiator’s mind is led to review the curious idea, not deliberately but just idly, the idea gets a little stronger—in the sense of a little more likely to occur in the initiator’s mind again. And again. It has a little self-replicative power, a little more self-replicative power than the other fantasies it competes with for time in the brain. It is not yet a meme, an item that escapes an individual mind and spreads through human culture, but it is a good proto-meme: a slightly obsessional—that is, oft-recurring, oftrehearsed—little hobbyhorse of an idea.

  (Evolution is all about processes that almost never happen. Every birth in every lineage is a potential speciation event, but speciation almost never happens, not once in a million births. Mutation in DNA almost never happens—not once in a trillion copyings—but evolution depends on it. Take the set of infrequent accidents—things that almost never happen—and sort them into the happy accidents, the neutral accidents, and the fatal accidents; amplify the effects of the happy accidents—which happens automatically when you have replication and competition—and you get evolution.)

  Would-be memeticists often ignore the fact that part of the “life” cycle of a meme is its moment-to-moment competition with other ideas—not just other memes, but every other idea anybody can think about—within a host brain. Rehearsal, deliberate or involuntary, is replication. We can try to make something a meme—or just a memory—by rehearsing it deliberately (a phone number, a rule to follow), or, if we just let “nature take its course,” our innate brain biases will automatically churn out rehearsals of the things that tickle them. This is probably the source, in fact, of episodic memory, our ability to recollect events in our lives. What did you have for breakfast on your last birthday? You probably can’t remember. What did you wear at your wedding? You probably can remember, because you’ve gone over it many times, before, during, and since the wedding. Unlike computer memory, which is an equal-opportunity storehouse that can readily record whatever is thrown into it, human brain memory is both competitive and biased. It has been designed by eons of evolution to remember some sorts of things more readily than others. It does this in part by differential rehearsal, dwelling on what is vital and tending to discard the trivia after a single pass. It does a pretty good job, keying on features that happen to have lined up with what tended to be vital in the past. Good advice to a potential meme is: if you want lots of rehearsals (replications), try to look important!