ABSTRACT
The replicability and importance of the correlation between cognitive style and religious belief have been debated. Moreover, the literature has not examined distinct psychological accounts of this relationship. We tested the replicability of the correlation (N = 5284; students and broader samples of Canadians, Americans, and Indians); while testing three accounts of how cognitive style comes to be related to belief in God, karma, witchcraft, and to the belief that religion is necessary for morality. The first, the dual process model, posits that analytical thinking is recruited in overriding intuitions related to supernatural beliefs. The second, the expressive rationality model, posits that analytical thinking is recruited in supporting already-held beliefs in an identity-protective manner. And the third, the counter-normativity rationality model, posits that analytical thinking is recruited to question beliefs supported by prevailing cultural norms. In Study 2, we tested the replicability of our results in a re-analysis of published data. The association between analytic thinking style and beliefs was replicated. We conclude that whereas the counter-normativity rationality model was contradicted by the data, both the dual process and expressive rationality models received varying degrees of empirical support, but neither model fully accounted for all the patterns in the data.
Author contributions
A.B. assembled the relevant datasets, conducted all analyses and wrote the paper with C.W., H.S., and A.N.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Moreover, the entire philosophical tradition of natural theology has aimed at providing demonstrative arguments to prove God’s existence and thus license great religious belief in the face of skepticism (Chignell & Pereboom, Citation2020).
2 The Indian participants in Study 1 were roughly 75% Hindu (the rest of participants were mostly Christian or Muslim); and although karmic beliefs are more strongly endorsed by Indian Hindus than other Indian subpopulations, they are still commonly endorsed by Indian Christians and Muslims (White et al., Citation2018).
3 These are the same two cases in which LOO model evaluation indicated no evidence for difference in predictive performance between regressions with and without the conservatism by analytical cognitive style interactions.