News and Views

Psychology: Atheism and moral intuitions

How robust is the perceived association between immorality and atheism? Studies across 13 countries demonstrate that immoral behaviour is intuitively associated with atheism: people routinely assume that an immoral person is likely to be an atheist, and this effect is consistent across a wide range of societies, though with notable variation.

Social cues are powerful indicators of socially relevant information: the way people dress, their accents, their facial expressions, their attitudes on abortion, their political party — these and other observations generally give us a better-than-chance guess about how they might interact with us. Although many of these cues depend on the context (for example, a facial tattoo might be a cue of aggression in one society but a cue of status in another), some social information usually means the same thing to all observers. These aspects of behaviour should represent dimensions of behaviour that have historically been beneficial for observers to attend to. Deriving their hypotheses from a cultural evolutionary theory of religion, a recent study by Gervais and colleagues1 provides evidence for one nearly universal heuristic — atheists will act immorally because they do not fear punishment from gods.

Cultural evolutionary theories of religion have proposed a crucial role for belief in punishing and for morally involved gods in facilitating cooperation in large-scale societies2. Because atheists don't believe that they are being watched by a morally concerned being, they face fewer perceived costs of non-cooperation (that is, believers are deterred by divine, in addition to secular, modes of punishment). Thus, prejudice toward atheists is rooted in distrust, reflecting their reduced incentive to cooperate3.

Although anti-atheist attitudes are well documented in Western samples, it was unclear how atheists are perceived across the world. To assess worldwide prevalence of anti-atheist biases, the authors utilized the conjunction fallacy, a well-known psychological bias4. The classic conjunction fallacy presents a vignette about Linda, who was a philosophy major and was concerned with social justice and discrimination. Participants then answer whether they think it is more likely that Linda is a bank teller, or that she is a bank teller and a feminist. Although logically impossible (that is, the likelihood of two phenomena occurring together cannot exceed the likelihood of either one alone), this bias occurs because the conjunction (bank teller and feminist) better matches people's impressions of Linda than the profession alone. Across 13 diverse countries, Gervais and colleagues demonstrated that participants are likely to commit this fallacy when asked if an immoral target (for example, a serial murderer) is more likely to be an atheist as well as a teacher (as opposed to just a teacher), but not when asked if the target is more likely to be religious. Follow-up studies suggest that this effect is not limited to extreme moral violations and ruled out the alternative hypothesis that immorality is associated with disbelief in anything (for example, evolution). Finally, in a case where immoral behaviour conforms to a religious stereotype, participants tended to assume that a priest who molests children is, in fact, a priest who doesn't believe in God.

Image: DNY59/E+/GETTY

This study marks an important advance in explaining the prevalence of anti-atheist attitudes. Traditional social psychological theories tend to focus on prejudice as a function of group membership. In this view, prejudice toward atheists should be prevalent mainly among believers. The cultural evolutionary framework utilized by Gervais and colleagues departs from this approach, proposing that distrust of atheists is rooted in intuitions that religion is necessary for moral behaviour. That is, because atheists lack belief in supernatural monitoring, they pose a cooperative threat that transcends group membership. Consistent with this framework, anti-atheist attitudes across several diverse samples seem to reflect these intuitions about religion and morality rather than out-group derogation. Perhaps most convincing is the finding that these effects are robust even in nations with non-religious majorities.

Although we agree that the conjunction fallacy is a useful tool for examining implicit biases, it remains unclear how information about atheism affects perceptions in natural settings. Atheism is rarely the only piece of information known about interaction partners, and it is possible that, when included with the social information that individuals collect naturally, atheism will be perceived as less indicative of immoral behaviour. Swan and Heesacker5 argued that individuating information doesn't override the atheist label, but does make targets seem more favourable — indeed the individuated atheist target was rated similarly to the unindividuated Christian target. Hall and colleagues6 utilized fake profiles, allowing them to manipulate single dimensions of a target to examine how religious beliefs or behaviours affect trust in a naturalistic social perception setting; this method may be useful in testing alternative explanations.

Further, the consistency of this effect is impressive, but the trend is not uniform across countries (the effect was stronger in the United States than Finland, for example). Rather than attribute these differences to noise, we anticipate that national and individual differences may be instructive in understanding anti-atheist prejudice. There are reasons to anticipate functional shifts in the degree of distrust felt toward atheists, as past research has shown that prejudice may vary in accordance with the perceived prevalence of atheists7. Further, we suspect that there are multiple distinct prejudices toward atheists, driven by separate functional threats8. Individuals concerned about mate retention may hold sexual stereotypes of atheists, while those concerned with safety may hold aggression-related stereotypes about atheists. Finally, atheists are not seen as all bad — they may even be considered more fun3. These tendencies too, may vary across cultures, as religious groups differ in the emphasis they place on religious beliefs, thoughts, and practices9,10.


  1. 1.

    et al. Nat. Hum. Behav. 1, 0151 (2017).

  2. 2.

    et al. Behav. Brain Sci. (2016).

  3. 3.

    , & J. Pers. Soc. Psychol. 101, 1189–1206 (2011).

  4. 4.

    & Psychol. Rev. 90, 293–315 (1983).

  5. 5.

    & Secularism Nonreligion 1, 32–42 (2012).

  6. 6.

    , , , & Psychol. Sci. 26, 1368–1376 (2015).

  7. 7.

    Pers. Soc. Psychol. Bull. 37, 543–56 (2011).

  8. 8.

    , & in Handbook of Social Psychology (eds Fiske, S. T., Gilbert, D. & Lindzey, G.) 761–796 (John Wiley & Sons, 2010).

  9. 9.

    & J. Pers. Soc. Psychol. 81, 697–710 (2001).

  10. 10.

    , & Front Psychol. 6, 1352 (2015).

Download references

Author information


  1. Adam B. Cohen and Jordan W. Moon are in the Department of Psychology, Arizona State University, 950 S. McAllister, Tempe, Arizona 85287, USA.

    • Adam B. Cohen
    •  & Jordan W. Moon


  1. Search for Adam B. Cohen in:

  2. Search for Jordan W. Moon in:

Competing interests

The authors declare no competing interests.

Corresponding author

Correspondence to Adam B. Cohen.