Field of Science

Do kids have to be taught about the supernatural?

One of the big questions about our supernatural instincts is to what degree they are innate, and to what degree they are learned. While there is some evidence that kids are born with a instinctive belief in omniscience beings, other evidence suggests the exact opposite.

Here's a new study by Ashley King, a grad student at the University of Arizona. She ran a group of 124 children aged 4-8 through a carefully designed study designed to see whether they understood what supernatural omniscience is.

The basic set-up has been used in previous studies. Half the kids were shown a video of someone calling herself "Princess Alice" who said she had special invisible powers and  would be present through the whole ensuing test. The other half were just shown a video of the same person wearing jeans and t-shirt who simply welcomed them and said she was a friend.

After a few other tests, they were given a puzzle by the experimenter. Basically, this involved some numbers hidden on the bottom of wooden blocks that were shown to the child and shuffled behind a screen. The child then had to guess where the different numbers were - an impossible task of course.

The experimenter then was called out of the room - she told the child that she would be gone for exactly three minutes, and left a clock showing how much time had elapsed. Question is, for how long could the children resist the urge to cheat by taking a quick peek at the numbers on the blocks?

As the graphic shows, the answer is "not very long"! But the interesting thing is that there does not seem to be much difference between the kids who thought they were being watched by Princess Alice, and those who did not.

Even more interesting, Princess Alice did seem to make a difference for the older kids. If they thought Princess Alice was watching, they took twice as long to summon up the courage to peek.

Now, it has to be said that this difference was not statistically significant using the test that King used. However, the magnitude of the difference is large, and I suspect that if the data from the 7 and 8 year olds had been pooled, a difference would have been clear.

So this supports the earlier studies by Jonathan Lane, which found that only older kids really get the idea of omniscience. Younger kids assume that supernatural beings have more ordinary mental capabilities.

In other words, it seems that this aspect, at least, of our beliefs about gods does not come naturally. It has to be learned.
King, A. (2011). Development of Inhibition as a Function of the Presence of a Supernatural Agent The Journal of Genetic Psychology, 172 (4), 414-432 DOI: 10.1080/00221325.2011.554921

Creative Commons License This article by Tom Rees was first published on Epiphenom. It is licensed under Creative Commons.


  1. Take home message: don't waste your time taking kids less than 6 years-old to church. Wait till they are old enough to understand that someone is always watching.

    Or, make it a MALE KING who is watching -- who'd be afraid of a "Princess Alice" anyway? Maybe a scary male would even be understood by a 3 year-old.

  2. Hope that was cynicism, Sabio Lantz!

    This all shows that we form hypotheses(plural, all the time) from about age 3 on.
    While most children´s methodical intelligence is hampered by the idea of an all-knowing god by grown-ups telling them so, uninhibited - or in a secular society - we would come to the conclusion, that the idea of thunder being Thor´s hammer is not causal, but a story.
    To stop believing into the Easter bunny does not stop the eggs, therefore to identify it as a (double, hare AND egg) symbol of fertility and spring in Northern hemisphere is rarely painful, but to have stopped believing in someone who sneaks into your bed before puberty makes you feel etchy and uncomfortable anyway is actually helpful.
    One point where the very idea or the moderate version of religion does damage; having to think about yourself and the planning of your life and having sexual relationships is enough, an additional voyeur with absurd "commandments" is just too much - for an average child in an average childhood.
    I do NOT want to expand on the damage of the god-idea growing up in a catholic abuse family in mine here.

  3. (1) Anonymous
    I rarely respond to people who don't supply a handle to facilitate discussion. Make up a handle! Be respectful and supply a way to have easy dialogue on posts. Arghhh, "Anonymous" folks drive me crazy.

    (2) "This all shows"
    It really shows NOTHING. We must be careful not to run with the confirmation bias.
    Tom states, "Now, it has to be said that this difference was not statistically significant using the test that King used." Which translate that the study showed nothing. Then Tom says, "However, the magnitude of the difference is large,"
    But significance counts, no magnitude!
    Then Tom goes further:
    "I suspect that if the data from the 7 and 8 year olds had been pooled, a difference would have been clear."
    Just because Tom "suspects" something doesn't give it any more likelihood of being true. Tom tells us in the beginning of this post: "other evidence suggests the exact opposite." And this experiment added nothing to the argument.

    Or, Tom, am I mistaken?

    (3) Sarcasm
    Of course it was sarcasm! Sorry your childhood was so rough. I was raised in a religion, embraced one and was not hurt -- maybe I was fortunate. I left it because it didn't make sense and doing so expanded my world.

  4. Sabio, the point about Princess Alice perhaps not being scary enough is actually made by King in the paper. Previous studies have used a more stern version of Alice.

  5. You have to be careful with stats. If you're conducting a formal proof then you set a threshold and a test before the experiment and if the numbers don't pass the threshold then the hypothesis is rejected.

    However, that doesn't mean throwing everything out. It's usual to dig through analyses to see if you can find anything interesting - that's the way that new hypotheses are generated to be tested in new studies.

    But you have to remember there is nothing magical about the threshold - it's just an arbitrary number we set.

    So, in this case there seems to be a large effect, but even so it wasn't significant. Why? Well, partly because of the small numbers in each group. In other words, the design of the study makes it really hard to get statistical significance. But that does take away from the fact that there is a suggestion here of something real.

    What's more, the statistical model does not take into account the fact that a large effect was seen both 7 and 8 year olds - it only looks at each group individually. In reality, the fact that both these ages show an effect means that it's less likely to be due to chance.

    Put this together with Lanes rather similar results (the earlier study), and I think the evidence is incrementally stronger.

    Still not conclusive by any means. But it does suggest that something interesting happens at around 7-8 years old. In fact, King also showed that these children had greater cognitive skills - in particular their Theory of Mind.

  6. I know Princess Alice would have certainly made a difference for *me* - by making me more aware of the fact that *real* people might be watching me via quite mundane means. So I wonder if the effect in older kids is due to supernatural beliefs or this increased awareness.

  7. Tom, I find your take on significance quite strange. No, significance is not a magical number, but it is used for a very solid reason. The lack of significance in these results means it's likely that the older children showed this effect for a reason outside of the experimental manipulation, or that the difference between the two groups occurred simply by chance. And note that children shown Princess Alice may not have had to "summon up the courage to peek" but may have simply been inspired to wait longer before losing self-control (a documented phenomenon). This also does not demonstrate that the children believe that Princess Alice was there, but took it as an indication that they were being watch, possibly by real humans.

    However, if these results were significant, your interpretation of the results is more than a little off base. The age difference does not show that ideas about omniscience have to be "learned" but that they have to be "developed." Infants cannot see depth before they crawl, but that does not make depth "learned." Children around this age are discovering that two halves are not the same as two wholes (conservation), but this is also not learned. Children will learn them without being taught, that is: they come naturally. These studies have not shown that understanding of omniscience has to be taught, but that it develops naturally as children age.

    So what impact does this have on ideas about religious thought? Well, not much. It's no shock to anyone that younger children do not understand complex ideas, like the fact that a person who is not, and has never been, visibly in the room can see you, especially since children that young cannot even identify that a picture or recording of themselves is, in fact, them, and acts as proof of their participation in a past event. Children of this age have a physically under-developed brain that does not give them the capacity to understand religious thought; this is not evidence that ideas about gods must be learned, but that the brain must be advanced enough to comprehend the concepts involved in that sort of magical thought. By your line of reasoning, the fact that children must develop the ability (not the skill, but the ability) to do math or think about concepts such as "self" makes such ideas unnatural or taught. Recent research shows that the idea of "tabula rasa" was just plain wrong, and that children are born with many innate abilities and tendencies, including the ability to understand complex ideas like omniscience.

    Also, you must clearly distinguish between the ability to understand the concept of an omniscient being, and the belief in gods. You and I understand omniscience, but do not believe in gods. The two do not go hand in hand. Children who are never taught language will develop their own languages. They always have the tendency and ability to do so, and only learn the language of their culture because of exposure. It is very likely so that this same principle is true of religion, or a lack thereof.

    You are a very intelligent individual, and I enjoy your blog quite a bit. But please do not take preliminary results of a study outside of your field and extrapolate them to mean things they do not.

    (I would like to read this full study, but no longer have access to a university library. Would you mind sending it to me?)

  8. Hi Carla, regarding the stats, let me put it another way. A statistical test is of a carefully defined hypothesis predicated on a number of assumptions, especially about the data distribution. The p-value gives you an indication of whether that particular hypothesis is correct.

    But a failed test doesn't mean that other, related hypotheses are not correct. In this particular case, the test assumes that the ages are independent, whereas they probably are not. If you see the result in 7 year olds, you'll probably see it in 8 year olds.

    This study was also underpowered - which happened because there were so many independent variables (ages). In the pharmaceutical industry, we call this kind of study a 'dose ranging study'. You spread your net wide, trying to see which doses (in this case, ages) have an effect. You then make an educated case based on the results, and then follow it up with a large confirmatory study.

    The money to do this isn't available in psychology, sadly. and so we often have to make with these small, underpowered studies. You put them together with results from other studies and try to untangle the bigger picture.

    Yes, you are right that 'taught' is the wrong verb. But the cricital point about this study is that the prevailing belief, based on seminal research by Justin Barrett, is that children have an intuitive understanding of omniscience from a very young age. It is, if you like, the 'default' position, and it's only with age that children develop an understanding that other agents (except god) are limited.

    However, this and other research suggests that Barrett was wrong. Taken on its own it is not sufficient to make that claim, of course, But it is part of a wider body of research revealing similar things in different experimental circumstances.

    Send me an email and I'll send you the paper. I just wrote to Ashley for a copy!


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS