American Scientist

Breaking the Spell: Religion as a Natural Phenomenon. Daniel C. Dennett. xvi + 448 pp. Viking, 2006. $25.95.

In recent years, Scott Atran (In Gods We Trust), Pascal Boyer (And Man Creates God) and David Sloan Wilson (Darwin's Cathedral) have all set forth evolutionary theories of religion. This focus is unsurprising: Like language, moral cognition, elaborate tool use and the extensive division of labor, religious belief seems to be both a typical and a distinctive feature of human intelligence—one of remarkable persistence.

Daniel C. Dennett's Breaking The Spell adds to the growing list of books on religion as a natural phenomenon but is strikingly different in tone and content from its predecessors. It is written for Americans, whose world is not secular: In the United States, it is impossible for an open skeptic to be a serious political figure, and even the skeptical take religion to be of profound moral and social significance. Dennett devotes much of his energy to trying to convince his nonsecular readers that it is legitimate to inquire scientifically into the roots of religious belief and to assess its moral consequences, good and bad. Reading this, I felt like a member of an alien species; it was a strange experience for a secular boy from a secular world.

Before I discuss Dennett's take on religion, let me explain why I think his cultural project is doomed. He wants the religious reader to take his assessment of religion's causes and effects seriously, and he labors mightily to present himself as fair-minded, if not neutral, about the truth of the central claims of such faiths as Christianity. The smart money is not on him. His intended audience will rightly regard any evolutionary model, indeed any secular model, of religion as essentially corrosive. For the so-called "genetic fallacy" (the erroneous supposition that a defect in the genesis of something is evidence that discredits the thing itself) need be no fallacy. A causal account of the origins and maintenance of belief can undermine that belief's rational warrant.

Suppose, for example, that someone were able to explain my political views about discrimination in university hiring by making the case that I held them because I am male and Caucasian. That explanation would not show that my beliefs were false, but it could show that they were not rationally based, because they depended not on the flow of information from the world to me but on my selfish interests. Given those interests, I would have those beliefs whether or not university hiring practices were fair. If, in fact, my beliefs were true, they would be so by luck. Rationally formed beliefs are counterfactually sensitive: Had the world been different, the agent's views would be different. For that reason, economic and other institutional theories of attitude are undermining. If true, they show that there is no mechanism through which an agent's beliefs about the world track states of the world.

Likewise, secular theories of religion are corrosive. Religious commitment cannot both be the result of natural selection for (for example) enhanced social cohesion and be a response to something that is actually divine. A cohesion-and-cooperation model of religion just says that believers would believe, whether or not there was a divine world to which to respond. If a secular theory of the origin of religious belief is true, such belief is not contingent on the existence of traces of the divine in our world. So although a secular and evolutionary model of religion might be (in a strict sense) neutral on the existence of divine agency, it cannot be neutral on the rationality of religious conviction.

I think this is true of all secular models of religious conviction, even the "economic model," the one that most aspires to neutrality. According to this model, which Dennett discusses in a chapter titled "The Invention of Team Spirit," religious belief is an instance of ordinary economic behavior. People join religious communities and sacrifice time, money and freedom to secure concrete rewards: immortality-despite-death, guaranteed bliss, supernatural intervention on their behalf and the like. These things are not available elsewhere; you can't just purchase them online. No wonder that the suppliers of such services stay in business. The trouble, of course, is ensuring delivery.

A community that believes in an immensely powerful and knowledgeable enforcer gets the benefits of its norms being followed without paying the costs of policing them.

• Facebook
• Twitter

We have a noncorrosive explanation of religion, but not a secular one, if we think that people are rational to expect to get what they pay for. We have a secular but corrosive theory if we think that such expectations are irrational and reflect the projection of economic decision-making beyond its proper domain. The economic model looks neutral only so long as we refuse to ask ourselves whether the expectation of return is rational. Once that question is posed, neutrality collapses. So Dennett is too optimistic if he imagines that religious readers will regard his research project as anything other than poisonous: They could be vindicated only by the project's failure.

Still, Dennett has a lot of interesting things to say about why religion might have evolved. His account is more multifaceted than those of Atran, Boyer and Wilson. In contrast with Wilson, for example, Dennett does not take religion to be a unitary phenomenon with a single evolutionary cause. Also, he does not present an integrated, systematic theory of religion; rather, he assembles the materials from which such a theory can be given for particular cases. I shall spend the rest of this review surveying some of these materials and Dennett's use of them.

Dennett has based his case in part on work of cognitive anthropologists Atran and Boyer, who in effect have argued that religion is a spandrel—a side effect of certain other cognitive adaptations. The simplest hypothesis is Atran's idea that religion is a consequence of our tendency to anthropomorphize, to project intentionality onto the world. We treat people as intentional agents—creatures that act as they do because of their thoughts and preferences. That regarding people this way is an adaptation is almost uncontroversial. As Dennett himself has persuasively argued in many of his works, it is often adaptive to treat other systems as intentional agents, especially when they are well-designed, well-functioning systems. But we habitually overuse this productive heuristic. It is harmless to talk to your cat, and it may well be productive for a hunter to conceive of his prey as actively planning to avoid or escape his attentions. But it is not adaptive to shout at and kick the step for being in the way after you have stubbed your toe on it. Likewise, we get no capacity to intervene in or predict the weather by thinking of storms as produced by divine agents. To the contrary, we get a false sense of control, which imposes a double tax: the price of the sacrifices we make, and the risks we expose ourselves to by embracing the illusion.

I agree with Dennett that this idea of Atran's partly explains the prevalence of religion. Dennett also accepts the central argument of Boyer, who thinks religious ideas spread so well because they strike a balance between strangeness (which gets them noticed and remembered) and ordinariness (which gives them intelligibility and appeal). That's why, according to Boyer, supernatural agents are transformed but familiar: They are people, animals or plants, natural objects like mountains, or artifacts like statues or knives. But they are not ordinary ordinary objects. Run-of-the-mill mountains do not eat, but this special, supernatural one does. Nevertheless, sacred mountains are mostly like other peaks; hence we can think intelligibly about them. That combination gives religious ideas both salience and transmittability. Dennett thinks that there is something importantly right about this observation and the ideas Boyer develops from it.

I am unconvinced. It is one thing for a supernatural entity to be memorable. It is another for it to be credible. Some stories get deeply etched in people's minds, perhaps because they transform the ordinary just enough to be gripping without becoming unintelligible. Famous examples—The Iliad, say—have lasted thousands of years. But people do not understand The Iliad as being literally true; still less do they organize their lives around it. And how would belief in one of the memorable fantasies humans imaginatively generate when puzzled or frightened—say, a tree that can walk (to use one of Dennett's examples)—ever be adaptive for an organism?

The best-known adaptationist ideas about religion link it to the striking fact that people must cooperate to survive. Generating resources jointly is an ancient feature of human lifeways, and we are adapted to and for cooperative social worlds. Wilson, Joseph Bulbulia and others have argued that religious belief is one of those adaptations. A community that believes in an immensely powerful and knowledgeable enforcer gets the benefits of its norms being followed without paying the costs of policing them. Dennett does not discount this hypothesis completely, but he is more inclined to endorse less obvious proposals that link religious belief to psychic benefits.

One such argument is that religion facilitates placebo effects: Perhaps the belief that you are the object of divine concern has real and crucial health benefits, particularly in a premodern world. Another is Boyer's hypothesis that religious belief simplifies choice-making in an informationally complex world.

In addition, philosophy has spawned a plethora of "ideal agent" theories: One idea from ethics, for example, is that what you ought to do is what you would want to do if you were ideally rational. Boyer points out that divine agents are often ideal agents; moreover, often they are ideal agents who are supposed to have our best interests at heart (our ancestors, for example). So in a complex and difficult world, it would be good to get their advice and follow it rather than succumb to decision paralysis. A nifty idea, although I remain skeptical: This strategy seems too costly and exploitable to me.

Dennett has long been involved in synergistic interaction with Richard Dawkins, so it is no surprise that Dawkins's memetic view of religion plays a role in Dennett's theory. Religion thrives, according to Dawkins, because its tenets and customs—its "memes"—like so many DNA or RNA-based genes, are structured to ensure that they are passed from one generation to the next (the Shaker practice of celibacy not withstanding).

Here Dennett's theory is nuanced. He points out that today's organized religions are reflective, self-conscious systems, which include not just beliefs about the supernatural but also rather strict ideas about how these beliefs are to be interpreted, warranted and fit together. Early religions may have a more or less direct biological explanation of the kind we have been discussing. But modern religions depend on massive investment in the mechanisms of cultural transmission. They cannot exist without the apparatus of holy books, seminaries, catechisms, theologians. So here a theory of cultural inheritance and cultural evolution comes into its own. Biases in preservation and transmission will be central to the explanation of the success and failure of modern religions. In contrast to Dawkins, though, Dennett does not assume that the dynamics of religious memes are virulently pathological. For him, this is an open empirical question.

I have never formed a settled view of the memetic explanation for the persistence of religion. Dawkins's image is wonderfully vivid, but the memetic view does not include an account of our cognitive biases that explains why we find supernatural explanations believable. Perhaps Atran's hypothesis (or one like it) explains this. But I have come to think that there is a plausible division of labor between meme-based views of religion and those that emphasize internal biases of the mind.

All selection processes depend on variation, and so a theory of our mental biases helps account for the cultural variations that are available. A meme-based theory is primarily external, emphasizing networks of the exchange of ideas and information. These networks and their organization help explain the transmission potential of human ideas. In small-scale social environments, people transmit many of their ideas to their kin in the next generation. Moreover, there is a reciprocal dependence on one's neighbors. In such networks, ideas and genes will be filtered in somewhat similar ways. In addition, there is limited investment in books, memory aids and other tools that help us remember and understand what we see and hear, so intrinsic biases on what we can understand and repeat play a leading role. In larger-scale societies, intrinsic memorability matters less, because cultural mechanisms are available to ease the memory burdens. So the amount of variation expands. Also, the networks are more open. Thus the transmission potential of ideas will be very different from that of genes, and ideas will be filtered very differently. In today's world, religions are not transmitted mostly to kin and to others who live in circumstances of mutual dependence. Such religions are therefore potentially more dangerous to us.

Despite my worry that Dennett is too hopeful that religious readers will be open to his arguments, I found Breaking the Spell thoughtful, informed and probing. It is surprisingly cautious: Dennett is sensitive to the limits on what we currently know and is mainly concerned to develop a research agenda. In discussing specific hypotheses about specific religions, he is not at all dogmatic. I doubt, though, that the book will reach the audience he wants. It will mostly be read by people like me, not by his decent, civil, morally serious Christian neighbors.