Science on Religion

Exploring the nexus of culture, mind & religion

Is atheism linked to intelligence?

Not so long ago experts predicted the imminent collapse of religion in modern western cultureSocial_Psychology_Quarterly_Cover_March_2010. Religion – often synonymous in these discussions with superstition, magic, and delusion – would at last give way to the autonomy of human reason and the power of the experimental method of natural investigation. But something happened on the way to religion’s funeral. People kept on believing. Recent neuroscientific and evolutionary research has suggested that either many of the hallmarks of religion are, or are byproducts of, adaptations that helped our earliest ancestors survive. The more that has been learned about the way our brains and our genes function it is the evolutionarily novel lack of religious belief and practices that is truly puzzling. A recent study by Satoshi Kanazawa provocatively titled, “Why liberals and atheists are more intelligent,” suggests that people that are more intelligent are able to overcome the limitations of our evolutionary past. In other words, the smarter you are the more likely you are to be an atheist.

Kanazawa proposes what he calls the “Savanna-IQ Interaction Hypothesis” as an explanation for the “origin of values and preferences” in human mental and social life. The hypothesis consists of a combination of the “Savanna Principle” and a theory of the evolution of general intelligence previously proposed by the author that suggests general intelligence evolved as a “domain-specific adaptation for the domain of evolutionarily novel, nonrecurrent problems” such as flash floods, drought, and lightning strikes (35). The Savanna Principle is the idea that the “brain has difficulty comprehending and dealing with entities and situations that did not exist in the ancestral environment” (the African savanna). The combination of these concepts leads Kanazawa to propose the “Savanna-IQ Interaction Hypothesis” that the constraints placed on human understanding by the ancestral environment should be stronger among those with less general intelligence and also that more intelligent people should be “better able to comprehend and deal with evolutionarily novel . . . entities and situations than less intelligent individuals” (36).

In order to test his hypothesis, Kanazawa conducted two studies, one of data from the National Longitudinal Study of Adolescent Health (Study 1) and the other from the General Social Surveys (Study 2) in order to identify any correlation between general intelligence and three “evolutionarily novel values.” While Kanazawa offers political liberalism and monogamy as well, we will look at what he has to say about the third novel value, atheism.

Before reviewing the specific findings related to atheism and intelligence we should note that Kanazawa takes it as settled that, “it is evolutionarily familiar and natural to believe in God, and evolutionarily novel not to be religious.” Kanazawa maintains this based on recent research in evolutionary psychology that suggests a hyperactive perception of intentionality in the brain that is adaptively paranoid about unknown causes and thus attributes quasi- or supernatural causation as an explanation.

While this line of research and the theories that attend it are far from unanimously accepted by students of the neurological dynamics of religion it should also be noted that there is a good deal of theoretical and experimental work needed to connect the tendency to perceive supernatural agents and the related but distinct idea of belief in a single supernatural agent such as the God usually denied by modern atheists. While it may be natural for our brains to suggest the actions of spirits or even gods it is not clear that belief in a monotheistic deity – the one usually denied by modern atheists – is similarly “natural.” In other words, whatever the state of the neuro-psychological evidence, Kanazawa’s line of argument might well benefit from greater sophistication in the understanding of religion.

In Study 1, Kanazawa analyzed data from the National Longitudinal Study of Adolescent Health (Add Health) in order to determine if adolescent intelligence (IQ) accurately predicted future adult beliefs and values (“religiosity”). Kanazawa followed the Add Health data in relying on Peabody Picture Vocabulary Test (PPVT) scores to determine “general intelligence.” The raw PPVT scores (0-87) “are age-standardized and converted to the IQ metric, with a mean of 100 and a standard deviation of 15” (42).

While acknowledging that the PPVT is “properly a measure of verbal intelligence, not general intelligence,” Kanazawa pools an impressive collection of research suggesting that the PPVT is related closely enough to general intelligence as to be applicable to his study. Unfortunately, the author does not even discuss the work of researchers who argue exactly the opposite point on the relationship between the PPVT and IQ (such as Maxwell & Wise, 2006).

After controlling for age, sex, race, education, religion, and number of times married, Kanazawa found a “clear monotonic bivariate relationship between adolescent intelligence and adult religiosity” (43). In other words, as religiosity increases intelligence decreases and as intelligence increases religiosity decreases. In fact, adult subjects who identify as “not at all religious” had a mean intelligence as adolescents of 103.09 while the “very religious” had a mean intelligence of 97.14.

While Kanazawa is correct that numerically these scores are significantly different from each other, he does not mention that despite this, the means are well within one-half of the standard deviation for IQ. Given that the range from 85 to 115 is often considered “average” intelligence, it remains an open question how significant these scores can actually be when translated into meaningful interaction with the world. The extent to which such scores can represent actually meaningful differences in intelligence would seem to be therefore in some question.

In Study 2, Kanazawa seeks to overcome the obvious generational limitations of his first study by comparing intelligence to religiosity in the General Social Survey data from 1972 to 2004. After controlling for age, sex, race, education, earnings, religion, and survey year, Kanazawa found that “more intelligent individuals have a significantly weaker belief in God . . . and significantly less intense religiosity” (47). On this basis, Kanazawa argues that the trend discovered in Study 1 is replicated in this far more demographically diverse sample.

However, only the data from Study 2 includes responses to the question of belief in God. The data used in Study 1 does not offer a measure of theistic belief. Rather, the first Study’s only religious variable is the extent to which subjects consider themselves religious people. For this reason, while it may be argued that “religiosity” has the predicted relationship to intelligence (as measured in the two studies) it remains to be demonstrated that atheism specifically – the intended “evolutionarily novel value” highlighted even in the paper’s title –  has such a relationship.

In addition, the measure of intelligence in Study 2 is a relatively simple multiple choice synonym quiz, which is most properly a measure not of general intelligence per se but of verbal intelligence. While it may be true that verbal intelligence (especially when measured by sophisticated tools like the PPVT) is “highly correlated with general intelligence” it is by no means certain. Furthermore, it could be argued that a high level of verbal intelligence is itself an evolutionarily novel feature. In light of the intelligence measure selected for Study 2, especially one might ask how many synonyms our ancient savanna dwelling relatives would have had any use for in the ancestral environment.

The Savanna-IQ Interaction Hypothesis remains an intriguing theory despite the potential limitations of the paper under review here. Most notably, people that are more intelligent were found to be no more and no less likely to value evolutionarily familiar concepts and objects such as marriage, family, children, and friends. This finding is entirely in keeping with Kanazawa’s hypothesis.

While offering a wealth of interesting observations Kanazawa seems not to have supplied an answer to his own provocative question. Further research is required to determine first if, and then why, atheists are more intelligent.

Satoshi Kanazawa, “Why liberals and atheists are more intelligent,” Social Psychology Quarterly 73(1): 33-57. DOI: 10.1177/0190272510361602

Add comment

Security code


0 #2 C. Caldwell-Harris 2011-05-03 21:17
Some other points about this article:

The point that Kanazawa's data set didn't even contain atheists is quite intriguing, as is the very small effect size (IQ of 97 vs. 103). I suppose he would say that this is almost half of standard deviation which is an amount that intelligence researchers consider important.

I'm puzzled by the reference to liberalism in the title of the paper. Is being liberal evolutionarily novel? The contemporary left's view on egalitarianism is more similar to the reverse dominance observed in many hunter gatherer groups (as discussed in the David Sloan Wilson's Darwin's Cathedral).

It makes more sense to me that taking on culturally newer ideas may co-occur with intelligence.

I do appreciate Kanazawa's paper on general intelligence, but some of his evolutionary psych papers have been ridiculed by bloggers, see:
0 #1 C. Caldwell-Harris 2011-05-03 21:16
Thank you for this insightful and informative critique.

This point of your critique is intriguing: ..."good deal of theoretical and experimental work needed to connect the tendency to perceive supernatural agents and the related but distinct idea of belief in a single supernatural agent such as the God usually denied by modern atheists."

However, when interviewed and surveyed, atheists emphasize that primary thing they reject is supernaturalism (see Hunsberger & Altemeyer, 2006 and my study,

Several of the atheists I surveyed expressed annoyance that the questions focused on God, deities, and religious belief, when they saw nonbelief in God as logically derivative from their commitment to scientific naturalism: "I don't just not believe in God. I don't believe in witches, fairies, ghosts or any supernatural beings."


New religion surveys online

Check out, a website filled with fascinating, research-grounded surveys about religion, morality, and belief. Sign up to get incisive feedback about your religious motivations and inner life – and help researchers learn more about science, religion, and culture in the process.

You are here: Home Research News Research Updates Is atheism linked to intelligence?