Even Atheists Distrust Atheists?

Hemant calls attention to a new study from Will Gervais, a psychologist at the University of Kentucky who has focused much of his scholarly attention to atheism and atheists. That study finds that people will jump at the chance to view a bad person as an atheist even in the absence of any evidence of that — and that atheists do it too! The abstract:

Scientific research yields inconsistent and contradictory evidence relating religion to moral judgments and outcomes, yet most people on earth nonetheless view belief in God (or gods) as central to morality, and many view atheists with suspicion and scorn. To evaluate intuitions regarding a causal link between religion and morality, this paper tested intuitive moral judgments of atheists and other groups. Across five experiments (N = 1,152), American participants intuitively judged a wide variety of immoral acts (e.g., serial murder, consensual incest, necrobestiality, cannibalism) as representative of atheists, but not of eleven other religious, ethnic, and cultural groups. Even atheist participants judged immoral acts as more representative of atheists than of other groups. These findings demonstrate a prevalent intuition that belief in God serves a necessary function in inhibiting immoral conduct, and may help explain persistent negative perceptions of atheists.

As Hemant points out, the use of the term “intuition” here is unnecessary. We usually describe common human beliefs as “intuitive” to mean that they are somehow build-in or hardwired, but this is almost certainly a learned response built in by culture, not nature.

Gervais did a series of experiments that began by telling a story and then asking respondents what they would infer about the person in the story. Here’s the first one:

When Dax was young, he began inflicting harm on animals. It started with just pulling the wings off flies, but eventually progressed to torturing squirrels and stray cats in his neighborhood.

As an adult, Dax found that he did not get much thrill from harming animals, so he began hurting people instead. He has killed 5 homeless people that he abducted from poor neighborhoods in his home city. Their dismembered bodies are currently buried in his basement.

Hemant describes the next step in the experiment:

Then Gervais asked the subjects what they could infer about Dax just from reading those two paragraphs.

They were given two options and asked which one was more likely: that Dax was A) a teacher or B) a teacher who does not believe in God.

Five other groups were asked to choose between A) a teacher and B) a teacher who was either Christian, Muslim, Hindu, Jewish, or Buddhist.

Now, the correct answer — 100% of the time, for every group — is always Option A. No matter what option B says, it is mathematically guaranteed that Dax is more likely to be a teacher than “a teacher and something else.” (This is commonly known as the “Linda problem” or “Conjunction fallacy.”)

So what did Gervais do with the results? He looked at how many people chose Option B in each group — the wrong answer every time — and calculated the error rates.

It turns out when subjects had to choose between Dax being a teacher and Dax being a teacher who was also something else, nearly half the people chose the wrong answer when “atheist” was offered up as an option.

I’m not sure I buy that this is a really good methodology and I’d be curious to hear from Dr. X, Kate Donovan and some of the other folks with some expertise in psychology on what they think about it. I have no doubt that in this culture people really do distrust atheists, we certainly have a great deal of evidence for that. But I’m not sure how much this really helps us quantify or understand that distrust.

What Are Your Thoughts?leave a comment
  • howardhershey

    I also have a problem with the methodology. I think most people, given this choice, would *read* into the A answer that it represents teachers who are not whatever constraint on that group that the B answer uses. That is, they would think that A actually means “Teacher who does not believe in god” or “Teacher who is not a Christian, Muslim, …” in opposition to B. That said, I don’t think that the result would be much different if they actually used two non-overlapping groups.

  • http://www.pandasthumb.org Area Man


    Okay, this is a new one to me. I’ve heard of necrophilia, I’ve heard of bestiality, but I had no idea that some enterprising individual had combined the two.

    And I agree about the likely dodginess of the methodology. When I read the questions they struck me as confusing. Why not just ask directly what people think?

  • http://Reallyawakeguy.blogspot.com somnus

    I would be interested to see a study in which participants are told a story about someone who performs an act generally held to be immoral but explicitly permitted/required by a specific religion, and are then asked to determine whether the person who did so was a member of that religion or an atheist. I’d be willing to bet that a sizable percentage even of members of the religion that allows the behavior would still claim the perpetrator was an atheist.

  • rory

    I’m pretty sure these type of questions are a fairly standard way to explore subconscious bias. (Kahneman discusses them in ‘Thinking, Fast and Slow’). The idea is that since B is logically impossible, the greater the extent to which people choose it, the greater the association in their minds between the quality proposed in the setup and the group described in the answers.

    There are apparently criticisms of this type of methodology (http://en.wikipedia.org/wiki/Conjunction_fallacy#Criticism_of_the_Linda_problem) but it’s not like the researchers just made it up.

  • Nick Gotts


    No, B is not logically impossible: you can indeed be a teacher and an atheist (or Christian, or Hindu…). It is logically impossible that B should be less probable than A, but when one of two answers is logically impossible, people who notice this may look for a way to reinterpret the question so that is no longer so – such as inferring that option A really meant “teacher who is not an atheist” (or Christian, Hindu…).

  • rory

    @5, my language was imprecise. Obviously someone can be thing A and also thing B, mutually exclusive categories notwithstanding. What I meant is that it is not possible that the likelihood of the conjunction be greater than the likelihood of one component of it.

  • Pierce R. Butler

    Answer Y or N: Does Dax fly to Dublin to proposition women in elevators?

  • Martin Hutton

    re: necrobestiality

    Wow, I thought flogging your dead horse was bad….

  • a_ray_in_dilbert_space

    I think the conclusion I’d draw from this study is that there are a helluva lot of people who don’t understand set theory.

  • http://en.uncyclopedia.co/wiki/User:Modusoperandi Modusoperandi

    To be fair, atheists are pretty bad. They’re always wandering around the park downtown, pecking at crumbs on the ground, and crapping on the statue.

  • John Horstman

    Well, I see a problem right away. How is consensual incest or necrobestiality immoral? It might be super-squicky, but consent is the defining factor of sexual ethics, so the incest case is fine and there is no actual second party in necrobestiality, so it’s no more immoral than using a dildo or fleshlight. Also, cannibalism isn’t really a moral issue (though it has a whole bunch of health risks) as long as you’re only consuming corpses, and not killing living people.

    The experimental model – (X) or (X and identity category) – is an established method of measuring identity bias, though that doesn’t necessarily mean it’s overall a good method. I’ll dig into the full thing later, but with just what I cited above (values dissonance might mean various people wouldn’t find the things the researcher *thinks* are universally viewed as immoral to actually be immoral), there are potential problems. I’d also have to read the whole thing to see how Gervais is arriving at the conclusion that even atheists think atheists are less moral – does he actually give responses to each question broken down by respondent demographics (he very well may)?

  • eric

    @11: in the first case, one could at least make the argument that coercion is more likely to play a part than it is with unrelated adults. There can be all sorts of cultural pressures to defer to authority figures in your family (a parent, older sib, or in some cultures for women to defer to male family members) in ways you wouldn’t defer to a stranger. It is also much harder to walk away from your family than it is some stranger you went out on a date with; most of us are culturally something of a ‘captive audience’ to our families, we feel loyalty to them, and that may prevent us from objecting to behavior we would object to if a stranger did it. All of these factors say to me that we are justified in being legally and socially suspicious about – when it comes to close relatives – whether the ‘consenting’ part of ‘consenting adults’ is real or not.

    Now, I can see that as a decent theoretical argument but empirically I’m not sure it holds up. We legally allow all sorts of relationships where coercion is more likely than in some mythical ‘ideal’ pairing. For example, bosses dating employees is legal. You might get fired for it, but the state will not prosecute you for it.. If there were data that showed that incestual relatiionships are no more coercive (or no more likely to be coercive) than other types of squirky relationships we DO legally allow, then I’d be on board with legalizing them. For legal consistency, what the con- side would need to show is that the coercive risk is higher than what we currently legally allow. Its not clear to me that it is…but its also not clear to me that it isn’t.

  • http://drx.typepad.com Dr X

    Seems pretty solid to me.

    @John Horstman:

    “Well, I see a problem right away. How is consensual incest or necrobestiality immoral?”

    Did you read the study? First, they’re dealing with a model of morality as intuition, not morality as a logical construct that stands up to philosophical scrutiny. The soundness of the study doesn’t rest on some absolute truth about what is and what is not moral or immoral. It’s about whether common intuitions of immorality tend to be more readily associated with atheists compared with believers. They found that strong association across a variety of common moral intuitions.

    So do people associate badness with atheism more often than religion? Yes. Considerably more often. We already knew that. Do atheists also tend to do that unconsciously? It looks like they do, at least under the conditions presented in these experiments.

    How generalizable is this, i.e, how and in what ways does this have practical effects in everyday affairs of real people? Who knows? Sometimes you can produce interesting effects in a study that show you something about how the mind works in a controlled situation, but the real world effect of the phenomenon identified is negligible.

    Think of the Implicit Associations Test. You can find biases that don’t necessarily translate well into predictions of real world conduct.

  • samgardner

    Well, I’d also point out that “even atheists distrust atheists” means here that at least one person who self-identified as an atheist picked the wrong answer. I do wonder how many would make this fallacy in the clear absence of bias, and whether this was factored in.

    It is logically impossible that B should be less probable than A

    I think you typoed that, Nick, and meant that it’s logically impossible that B should be more probable than A.

  • dingojack

    I would liked some to have been asked how likely it was that Dax is:

    c) A teacher and a psychopath*.



    * Or perhaps:

    d) A [insert religion here] and a psychopath.

  • royandale

    I’ve long thought that moral behavior (altruism, community support, etc) is more likely to be a genetic trait in us (and other species) and that religious belief is secondary to this built-in propensity. But I have been dismayed by the number of people who not only question the source of my moral sense given that I’m atheist, but feel that they themselves only behave morally because of a watchful God. One of our longtime newspaper columnists here in Lexington, KY, a preacher himself, confessed as much in a recent column. His last paragraphs say it succinctly:

    “That’s another key reason I keep talking the talk and trying to walk the walk: Faith in a loving, generous God tends to overwhelm my baser nature.

    Even if you don’t need that kind of help yourself, and even if you don’t believe in God a whit, you should be glad I do believe. Be very glad.”

    That’s the kind of talk that makes me worry.

    Read more here: http://www.kentucky.com/2014/03/08/3129696/paul-prather-why-i-believe-in.html#storylink=cpy

  • dingojack

    royandale – Nice to see Fundies admitting they’re low on the moral deployment scale at the very least (I suppose).

    :( Dingo