I briefly mentioned before a study that said that children brought up in a religious environment had a harder time distinguishing whether stories that had fantastical elements were real or fictional. Unfortunately, I did not have access to the full paper but reader Raven did have access and he kindly sent me a copy and I have now had a chance to look at it. [Update: Ben Finney in the comments has kindly provided a link to the full paper.]

The paper Judgments About Fact and Fiction by Children From Religious and Nonreligious Backgrounds by By Kathleen H. Corriveau, Eva E. Chen, Paul L. Harris (that appeared in the journal Cognitive Science (2014) 1–30) reported on the results of two studies. In both they interviewed 5- and 6-year old children, some of whom had been exposed to religious teaching either because they attended a parochial school or went to church with their family or both, and those children who went to public schools and had no religious exposure.

The children were presented with two boxes marked ‘real’ and ‘pretend’ and asked to decide which character or story went into which box. After first establishing that the children could tell the difference by answering correctly when they were presented with names they were already familiar with (such as Goldilocks and Thomas Edison), the children were then ready for the actual study and the experimenters went on to the next stage.

Here is what they did in Study 1.

The experimenter then presented nine characters: three in a realistic story context, three in a religious story context, and three in a fantastical story context. The religious stories were adapted from the Old and New Testaments of the Bible, in which an ordinarily impossible event was brought about via divine intervention. In the fantastical stories, the same Bible stories were modified to exclude any reference to divine intervention, so that the impossible event was effectively presented as magical rather than miraculous. Finally, in the realistic stories, the Bible stories were modified such that the improbable event was made plausible due to human intervention (see Appendix A for the full script of the nine stories in all three contexts). (p.6)

Here’s what they did in Study 2.

The experimenter then presented eight story-based characters. As in Study 1, all of the characters were adapted from the Old and New Testaments of the Bible, but all were presented in the fantasy context only and with a name that was not tied to the biblical story (for example, in the story of the parting of the Red Sea, Moses was called John). The 8 stories were modified to exclude any reference to divine intervention, so that the impossible event was presented as violating ordinary causal principles but not as a miracle. Nevertheless, in half of the stories, the causal violation was the familiar violation from the Bible story (e.g., the parting of the sea). In the remaining half of the stories, the causal violation was comparable in terms of its causal implausibility, but it was unfamiliar (e.g., the parting of a mountain). In addition, for half of the stories, the word “magic” was explicitly used. In the remaining half, the word “magic” was not used. This resulted in four within-subjects story conditions. (p. 14)

So what did the authors conclude? From Study 1:

Our central question concerned children’s judgments about the status of story characters in religious stories. Children with exposure to religion—via church attendance, parochial schooling, or both—judged such characters to be real. By contrast, children with no such exposure judged them to be pretend. This sharp discrepancy between children with and without exposure to religion lends no support to the hypothesis that children are “born believers” (Barrett, 2012) with a natural credulity toward extraordinary beings with superhuman powers. Indeed, secular children responded to religious stories in much the same way as they responded to fantastical stories—they judged the protagonist to be pretend. (p.12)

From Study 2:

Across all four story types, secular children were more likely than religious children to categorize the protagonist as pretend. This pattern of results undermines the hypothesis that religious children associate fantastical stories with familiar Bible stories. (p.17)
…

The presence versus absence of a religious education is associated with children’s conviction that ordinarily impossible events can or cannot occur in a realistic story. (p.18)

The authors draw some conclusions from both studies about the effects of religious influence on children’s thinking.

How exactly do religious children come to have a broader conception of what can actually happen than secular children? On the one hand, exposure to religious teaching might encourage children to entertain the idea that some agents are endowed with a special or superhuman power that can override ordinary causal regularities. This extraordinary power might be best exemplified by God, but it might extend beyond God to a variety of other agents, including those encountered in narratives. On this hypothesis, even if children have no natural inclination to believe in divine or superhuman agency, religious instruction can readily lead them to do so. An alternative possibility is that children are disposed to credulity unless they are taught otherwise by their families. Thus, secular children are schooled in the idea that natural laws preclude any kind of miraculous or magical outcome. For example, their parents might cast doubt on any invocation of non-natural powers. According to this interpretation, secular children would likely differ from religious children in thinking about what can happen even when no agent, divine or otherwise, is present. More specifically, on this second hypothesis, secular children would be more confident than religious children that certain outcomes are impossible, simply because they violate the laws of nature, irrespective of the presence or absence of any particular agent.

It is too early to adjudicate firmly between these two hypotheses, and indeed both might have some validity. Nevertheless, recent findings lend more support to the first hypothesis than the second. When asked to say what events could or could not occur in real life, children from 4 to 8 years of age as well as adults differentiate sharply between ordinary events (e.g., eating an apple) and impossible events (e.g., walking on water; Shtulman & Carey, 2007). Moreover, this differentiation is observed whether the events fall into the physical, biological, or psychological domain (Shtulman, 2009). Given this early emerging, widespread and stable tendency to judge a variety of events in different domains as impossible, it is reasonable to think of children as adopting a default stance of doubt toward violations of ordinary causality. In other words, it is more plausible that a religious upbringing overcomes children’s pre-existing doubts about whether ordinarily impossible events can occur than that a secular upbringing suppresses children’s natural inclination toward credulity. (p.23)

We live in an age when even otherwise sensible people suffer from all manner of magical thinking such as superstitions, ghosts, witchcraft, demons, angels, fairies, vampires, zombies, communicating with the dead, and the like, all of which violate the normal laws under which we operate. Being indoctrinated with religion at an early age is like a gateway drug that causes people to be more willing to succumb to such beliefs later in life.

From there it is a short step to magical thinking in other contexts, such as that cutting taxes on the rich will lead to prosperity for everyone and that doing nothing about climate change will cause the problem to go away by itself and that the US health care system is the best in the world.