The streets of small rural towns in United States are typically tranquil places, but they are especially still on Sunday mornings. On those mornings, people dutifully report to their local places of worship, pay their tithes, listen to the sermon, sing the hymns, and bow their heads for the prayers. Religion is a cultural bulwark; it is a glue of the community.

But southern piety, like all incarnations of faith, harbors a dark history. One spotted with racism and contempt for minorities, and incredulity that women might do much else than sing in the choir. What’s worse is that there is no shortage of scripture from which to justify these iniquities. And God, at least in the past it seems, provided ready dispensation. More insidious are the acts of violence and aggression committed under the auspices of religion: the crosses burned, the houses of worship destroyed, and the lives taken, all with the apparent sponsorship of no less than the “creator of the universe“. Religion, it would seem — all brands included, but some more than others — has a real problem on its hands.

Cultural firebrands like Christopher Hitchens and Richard Dawkins have leveled scathing criticisms of the faithful. At the heart of the discussion, though, is a fundamental issue: causality. Does religion cause one to be good or bad? There are persuasive arguments in both directions. According to Sam Harris, religion is a wellspring of horror. Islam is “the mother lode of bad ideas,” and Christianity is not much better. Given the association between religious belief and intolerance, Harris argues that we should inoculate ourselves against the disease of faith (which Dawkins has described as a kind of mental virus). When religion does inspire acts of compassion and self-sacrifice, Harris argues that it simply gives us “bad reasons for being good when good reasons are available.”

At the same time, religion seems to be flexible, accommodating scriptural readings that sanction either tolerance or violence (Harris acknowledges this point). The forces that push around these interpretations — informing what they give rise to — are geopolitical and historical; more disconnected from what’s actually written in the holy books. Indeed, compelling arguments can be made on both sides. And yet, authors in the two camps have sometimes failed to appreciate a deeper layer of complexity that makes an answer to this question elusive.

The Swamp of Faith

Imagine that we did not know what caused malaria, yet we have observed that living in close proximity to a swamp seems to increase the incidence of new cases. We suspect that swamp gas might be the cause.¹ Someone proposes testing our idea systematically and we know that there are 100 swamps in the region. With the flip of a coin we decide which swamps will be drained and which ones will be left standing. Within a period of time, we observe that new cases of malaria have plummeted in areas where the swamps have been drained. Our swamp gas interpretation seems vindicated.

Except for the fact that it isn’t; we never even really tested our hypothesis. All that we’ve demonstrated is that there is something about swamps that seems to cause malaria. To know if it was the gas, we would have needed to remove only the gas, and leave the other swamp qualities untouched (that’s essentially impossible). Plus, we know that malaria is transmitted via mosquitoes, not gas, and the prevalence of disease fell because draining the swamp robbed the mosquito larvae of the stagnant water needed to reproduce. This example illustrates a simple reality: causal interpretation is tricky. Even when we’re reasonably careful — as in the case of our rudimentary swamp experiment — our causal attribution can be incorrect.

There is no wand to wave in order to make the disconcerting parts of religion disappear in some places and not others so that we can observe the effects. Religion, and figuring out what part of it might create bad behavior, doesn’t lend itself easily to experimental manipulation. We are left then with the next best option. We do the best we can to try to observe whether religious beliefs co-occur with certain behaviors (either good or bad). We try to make sure the religious beliefs came first, since we don’t want to mistakenly conclude that religion causes behavior if it could be the other way around. Finally, we’re careful to account for any variable that might predict both the religiosity of individuals, and also how they behave. It is this last one, though, that complicates life.²

People do not just stumble into religion. While it’s true that most people inherit their religious beliefs from their parents and community, to some extent people also select themselves into certain religions, especially in diverse societies with some degree of religious freedom. It’s more complicated than even this, though. Religious affiliation is one thing, but the fervor with which one practices religion is something different. It’s possible to be nominally Christian or Islamic. Alternatively, there are true believers, dogmatically observant of every word of holy text. These people are maximally obedient, and they acquiesce to even the parts of scripture that trouble modern, secular intuitions.

It’s reasonable to argue that religion radicalizes (mainly) young men, prompting them to detonate the suicide vest, burn the cross, or harass their gay victim. It’s reasonable too, though, to contend that certain personality styles are attracted to violence and aggression. The fuses of these people are shorter, their tolerance for frustration is low, and their ability to delay gratification is lacking. The fantasy of paradise after the push of a button is a siren song that’s hard to resist — the exploding vest is hardly even felt. Of course, one shouldn’t confuse this with the argument that all jihadism stems from a cocktail of low self-control and an authoritarian personality. We use these traits only to illustrate a point.³

Harris rightly points out that the contrary can be true: years of planning and a significant ability to delay gratification are required to execute a complex plot like the 9/11 hijacking and World Trade Center attacks. Still, if people drawn to terrorism or extreme levels of religious piety are systematically different from those who are not, then a selection effect is at work. This makes it difficult to specify religion as a potent causal force for behavior.

It can of course be true that selection is occurring and that religious indoctrination exerts causal effects on behavior. To illustrate, consider a less inflammatory example: football (the American variety). Does participating in football make one violent? The play is physical; it’s a game that embodies aggression and competition. Participation in football — perhaps through a number of mechanisms — may predispose one to rely on force rather than words to solve conflicts. Of course, not everyone is interested in football. It might simply be that people more prone to violence are more attracted to football. We’ve tested this, but we did it using data collected from twin siblings. What we found were some strong selection factors in play (both genetic and environmental) that explained why some adolescents played football and others did not, and also why some individuals were more violent than their peers. When we used techniques that allowed us to pull out these selection effects, adolescents who played football appeared to, nonetheless, be more prone to violence.

Why did it matter that we used twins? It would have been far easier to examine one child per family, since that’s what most social scientists do. The problem is that this age-old approach to social science is incapable of accounting for the selection effects that we accounted for in our paper. In a perfect world, we would have randomly assigned some kids to play football, others not, and then observed whether the two groups were different in their violent behavior. Because that’s infeasible, twin and sibling studies are the next best thing.

Clearly, we cannot randomly assign some children to be raised in Islamic homes, Christian homes, Jewish homes, Hindu homes, and so forth to see how and to what extent it would impact their behavior. Even if we could do it, such an approach would only reveal whether a certain faith was more broadly harmful than another, it would not tell us what aspect of the faith was deleterious. Yet the designs that we might otherwise use are not sufficient to investigate the effects of religious indoctrination. You can correlate things all you like, but unless you account for the selection effects you’re uncovering associations (which admittedly are very important), but not saying anything definitive about causality. Arguments about the purported effects of religion on behavior cannot be given a pass.

If you’re arguing that religion causes a behavior, then it’s incumbent on you to try to demonstrate that empirically, using data that can reveal more than just a correlation. This is no small task.

Smoking Cigarettes and Saying Your Prayers

We are not using selection bias as an endorsement of religion, or an attack on those who argue that it causes harm in the world. Cigarette smoking provides a perfect example. People do not randomly take up smoking. Most people know that smoking is a hazardous behavior, a fact made clear through a variety of studies (including twin studies). But some do it anyway, perhaps in part because they have poor self-control, especially over decisions that affect their long-term well-being. Religion may be quite similar in its effect.

Sam Harris has asked us to imagine that we could edit the Quran so as to remove any mention that homosexuality is a sin. Is it reasonable to suspect that some people would then feel differently about homosexuals? Probably. But others might be inclined to turn to a religious doctrine that licensed, even demanded, the persecution of some other minority group (their focus would simply shift).

A reasonable critique could be that the discussion here lacks the geopolitical nuance needed to deal with this topic (Islam and terrorism in particular). Yet, if we think that certain geopolitical events, economic sanctions, invasions, and military occupations predict fanaticism and violence, what we’re really saying is that if an invasion had never occurred, or sanctions were never levied, or an occupation did not take place, then the behavior in the region would be different (and remember, occupations often take place in response to terrorism, not always before it, despite what some terrorists tell themselves and their followers). This is still a causal argument, but we’ve changed the unit of analysis and the variables. We cannot run experimental trials of occupations. Thus, we have to put the pieces of the puzzle together by accounting for all of the non-random factors that predict certain actions (why the invasion happened, what predicts sanctions being levied, and so on).

Religions and countries have long, complicated histories. So maybe the easiest way of sorting this out is to simply listen to what extremists say about their own motivation. If someone tells you that they are committing atrocities because their religion tells them to, why not take them at their word? But even that is surprisingly tricky, as our own motivations and causal attributions can deceive us.

In The Blank Slate, Steven Pinker discusses patients who have had their corpus callosum severed, disabling communication between the left and right hemisphere of the brain. If you ask these people to perform some action by flashing them a card with a written command, but you restrict that request to the right hemisphere, the patient will comply. When you question the patient about their behavior, but restrict the question to the left hemisphere, they will invent a reason for their previous act. In the absence of a clear justification for an action, the brain supplies its own. Pinker observes:⁴

The spooky part is that we have no reason to think that the baloney-generator in the patient’s left hemisphere is behaving any differently from ours as we make sense of the inclinations emanating from the rest of our brains. The conscious mind — the self or soul — is a spin doctor, not the commander in chief.

Every Islamist has a motivation. Every Westboro Baptist protestor has scripture and verse to back up their hateful signs. This is not an argument against the negative influence of religion, it is a call for careful theorizing about why people do what they do.

As we close, consider the following from Sam Harris:⁵

A belief is a lever that, once pulled, moves almost everything else in a person’s life. Are you a scientist? A liberal? A racist? These are merely species of belief in action. Your beliefs define your vision of the world; they dictate your behavior; they determine your emotional responses to other human beings. If you doubt this, consider how your experience would suddenly change if you came to believe one of the following propositions:

1. You have only two weeks to live.
2. You’ve just won a lottery prize of one hundred million dollars.
3. Aliens have implanted a receiver in your skull and are manipulating your thoughts.

These are mere words — until you believe them.

The passage is stirring; who could argue with it? Surely, our behaviors would change in the wake of any one of these revelations. And yet, can there be much doubt that before beliefs ever have a chance to shove around behavior, they must inevitably pass through the mesh screen of our personality?⁶ After all, our personalities — products of our central nervous system — are all we have to shape, process, and steer our beliefs and our behaviors.

Make no mistake, we should be wary of the effects that beliefs might have. But we should be cautious in assigning wholesale causal influence to those beliefs when attempting to understand why the faithful among us are either “bringing in the sheaves” or punishing the sinful.

Endnotes

[1] Kenneth J. Rothman and Sander Greenland propose this example in a wonderful article on causality entitled “Causation and Causal Inference in Epidemiology,” published in The American Journal of Public Health.

[2] Luke Galen presents a very nice, detailed, overview of the complexities of linking religion to behavior.

[3] For a much more thorough discussion pertaining to this topic, see the work of Adam Lankford.

[4] Page 43 in The Blank Slate

[5] Page 12 in The End of Faith by Sam Harris

[6] We’re not suggesting that Sam Harris would disagree with this point, or that that it runs counter to his arguments. Indeed, he has made it clear that he believes personality traits matter.