Norenzayan et al. present an admirable and highly plausible theory on the development and spreading of prosocial religions. Their approach represents a significant advance in the psychological study of religion, integrating many central ideas into a unified theory and generating a host of testable hypotheses. We specifically applaud the theory for explicating how the effects of cultural factors and cognitive factors intertwine in bringing about specific types of religions.
That said, most of the theory's central arguments about the relationship between mentalizing, cognitive biases, and religion remain to be empirically tested and should be stated with less certainty than is currently the case. Although the evidence that the authors cite is abundant and compelling, it is indirect and must be acknowledged as such. Aside from the difficulty of proving evolutionary hypotheses, in order to properly test the theory, it needs to be complemented by direct evidence of a link between mentalizing, the cognitive biases, and religious beliefs.
The authors write that the “cognitive science of religion has begun to show that religious beliefs are rooted in a suite of core cognitive faculties” (sect. 2.1, para. 1), and that individual differences in mentalizing, teleological thinking, and dualism partly explain religious and other paranormal beliefs. They cite several papers and books in support of these arguments. Closer analysis of the references, however, reveals that rather than offering empirical evidence, they provide only theoretical or indirect information about the relationship under scrutiny. For example, although Barrett (Reference Barrett2004), Kirkpatrick (Reference Kirkpatrick1999), Lawson and McCauley (Reference Lawson and McCauley1993), Guthrie (Reference Guthrie1993), and Boyer (Reference Boyer2001) provide conceptual backgrounds for the authors’ theoretical position, they are mainly hypothetical discussions of the types of relationships that we might expect to find. As well, many of the sources present background information on the main concepts but do not directly deal with the associations between mentalizing, cognitive biases, and religious beliefs, and some do not mention religion at all (Frith & Frith Reference Frith and Frith2003). Other references present evidence that is empirical but indirect. For example, Waytz et al. (Reference Waytz, Gray, Epley and Wegner2010), Norenzayan and Shariff (Reference Norenzayan and Shariff2008), and Bloom (Reference Bloom2012) deal with the question of whether religion has effects on morality or on prosocial behavior; however, they do not address the cognitive mechanism through which these associations might be realized. Although we cannot criticize the citing of any of these references, the length of the reference list may give readers an incorrect impression that the relationships between mentalizing, cognitive biases, and religiosity are well established.
Further, there are several studies whose results are not in line with the argument that individual differences in mentalizing and cognitive biases partly explain belief in God and in paranormal events, but discussion of these studies is missing. For example, promiscuous teleology is not necessarily related to God beliefs (Lombrozo et al. Reference Lombrozo, Kelemen and Zaitchik2007). Strong mentalizers do not experience supernatural agency more strongly (Barnes & Gibson Reference Barnes and Gibson2013), and do not attribute their life events to God, any more often than others (Banerjee & Bloom Reference Banerjee and Bloom2014). Similarly, the evidence for the argument that reduced mentalizing abilities, as found in autistic spectrum disorders, predict reduced belief in God, and that schizotypal tendencies are associated with “hyperreligiosity,” is not as clear as Norenzayan et al. suggest. Individuals with autism spectrum disorder may consider religion to be important (Schaap-Jonker et al. Reference Schaap-Jonker, Sizoo, van Schothorst-van Roekel and Corveleyn2013), and schizotypy is not necessarily linked with any kind of religiosity (Diduca & Joseph Reference Diduca and Joseph1997; Maltby et al. Reference Maltby, Garner, Alan Lewis and Day2000). Linking religious beliefs with schizotypy and contrasting them with autistic traits and nonbelief is therefore an oversimplification.
The few existing papers that have thus far supported a positive relationship between mentalizing, cognitive biases, and religiosity (and that are cited) are problematic for a number of reasons. Firstly, the relationships have tended to be weak. For example, when mentalizing, autistic traits or promiscuous teleology have predicted religiosity (or vice versa), the beta coefficients have ranged from 0.07–0.18 (Kelemen et al. Reference Kelemen, Rottman and Seston2013; Norenzayan et al. Reference Norenzayan, Gervais and Trzesniewski2012; Willard & Norenzayan Reference Willard and Norenzayan2013; see also Lindeman et al. Reference Lindeman, Svedholm-Häkkinen and Lipsanen2015), implying that the ability of these predictors to explain variance in religiosity has ranged from trivial to small.
Secondly, mentalizing is a large-scale, multidimensional construct. It can refer to the capacity to understand that other people have minds different from one's own (the theory of mind proper); the capacity to understand what others feel or think; the ability to experience and share the emotions of others; the ability to interpret communicative signs, detect intentionality, and understand social outcomes; the mirror neuron system; and the ability to think about thinking (i.e., metacognition). It can pertain to attending, perceiving, recognizing, describing, interpreting, inferring, imagining, simulating, remembering, reflecting, and anticipating, making it a slippery concept, potentially extending beyond manageable bounds (Allen Reference Allen, Allen and Fonagy2006). The available instruments do not capture this variance in mentalizing among healthy adults. In most studies on supernatural beliefs, mentalizing has been operationalized either with the Empathy Quotient (Baron-Cohen & Wheelwright Reference Baron-Cohen and Wheelwright2004), or with the Eyes test (Baron Cohen et al. Reference Baron-Cohen, Wheelwright, Hill, Raste and Plumb2001), which according to recent studies may not tap mentalistic abilities (Valla et al. Reference Valla, Ganzel, Yoder, Chen, Lyman, Sidari and Wong2010). However, comprehensive and valid methods to assess mentalizing among healthy adults are not easily available. Hence, there is a pressing need for the field to develop better methods for studying how mentalizing and various biases are related to religiosity today.
In sum, we congratulate Norenzayan et al. for the much-needed theoretical synthesis, but we caution against relying too strongly on evidence that, at least for part of the theory, is currently scarce and contradictory. At present, the only thing that can be said is that there “may be” a relationship between mentalizing, certain cognitive biases, and religiosity. Evidence that would warrant using the wording “is” or even “probably is” does not yet exist.
Norenzayan et al. present an admirable and highly plausible theory on the development and spreading of prosocial religions. Their approach represents a significant advance in the psychological study of religion, integrating many central ideas into a unified theory and generating a host of testable hypotheses. We specifically applaud the theory for explicating how the effects of cultural factors and cognitive factors intertwine in bringing about specific types of religions.
That said, most of the theory's central arguments about the relationship between mentalizing, cognitive biases, and religion remain to be empirically tested and should be stated with less certainty than is currently the case. Although the evidence that the authors cite is abundant and compelling, it is indirect and must be acknowledged as such. Aside from the difficulty of proving evolutionary hypotheses, in order to properly test the theory, it needs to be complemented by direct evidence of a link between mentalizing, the cognitive biases, and religious beliefs.
The authors write that the “cognitive science of religion has begun to show that religious beliefs are rooted in a suite of core cognitive faculties” (sect. 2.1, para. 1), and that individual differences in mentalizing, teleological thinking, and dualism partly explain religious and other paranormal beliefs. They cite several papers and books in support of these arguments. Closer analysis of the references, however, reveals that rather than offering empirical evidence, they provide only theoretical or indirect information about the relationship under scrutiny. For example, although Barrett (Reference Barrett2004), Kirkpatrick (Reference Kirkpatrick1999), Lawson and McCauley (Reference Lawson and McCauley1993), Guthrie (Reference Guthrie1993), and Boyer (Reference Boyer2001) provide conceptual backgrounds for the authors’ theoretical position, they are mainly hypothetical discussions of the types of relationships that we might expect to find. As well, many of the sources present background information on the main concepts but do not directly deal with the associations between mentalizing, cognitive biases, and religious beliefs, and some do not mention religion at all (Frith & Frith Reference Frith and Frith2003). Other references present evidence that is empirical but indirect. For example, Waytz et al. (Reference Waytz, Gray, Epley and Wegner2010), Norenzayan and Shariff (Reference Norenzayan and Shariff2008), and Bloom (Reference Bloom2012) deal with the question of whether religion has effects on morality or on prosocial behavior; however, they do not address the cognitive mechanism through which these associations might be realized. Although we cannot criticize the citing of any of these references, the length of the reference list may give readers an incorrect impression that the relationships between mentalizing, cognitive biases, and religiosity are well established.
Further, there are several studies whose results are not in line with the argument that individual differences in mentalizing and cognitive biases partly explain belief in God and in paranormal events, but discussion of these studies is missing. For example, promiscuous teleology is not necessarily related to God beliefs (Lombrozo et al. Reference Lombrozo, Kelemen and Zaitchik2007). Strong mentalizers do not experience supernatural agency more strongly (Barnes & Gibson Reference Barnes and Gibson2013), and do not attribute their life events to God, any more often than others (Banerjee & Bloom Reference Banerjee and Bloom2014). Similarly, the evidence for the argument that reduced mentalizing abilities, as found in autistic spectrum disorders, predict reduced belief in God, and that schizotypal tendencies are associated with “hyperreligiosity,” is not as clear as Norenzayan et al. suggest. Individuals with autism spectrum disorder may consider religion to be important (Schaap-Jonker et al. Reference Schaap-Jonker, Sizoo, van Schothorst-van Roekel and Corveleyn2013), and schizotypy is not necessarily linked with any kind of religiosity (Diduca & Joseph Reference Diduca and Joseph1997; Maltby et al. Reference Maltby, Garner, Alan Lewis and Day2000). Linking religious beliefs with schizotypy and contrasting them with autistic traits and nonbelief is therefore an oversimplification.
The few existing papers that have thus far supported a positive relationship between mentalizing, cognitive biases, and religiosity (and that are cited) are problematic for a number of reasons. Firstly, the relationships have tended to be weak. For example, when mentalizing, autistic traits or promiscuous teleology have predicted religiosity (or vice versa), the beta coefficients have ranged from 0.07–0.18 (Kelemen et al. Reference Kelemen, Rottman and Seston2013; Norenzayan et al. Reference Norenzayan, Gervais and Trzesniewski2012; Willard & Norenzayan Reference Willard and Norenzayan2013; see also Lindeman et al. Reference Lindeman, Svedholm-Häkkinen and Lipsanen2015), implying that the ability of these predictors to explain variance in religiosity has ranged from trivial to small.
Secondly, mentalizing is a large-scale, multidimensional construct. It can refer to the capacity to understand that other people have minds different from one's own (the theory of mind proper); the capacity to understand what others feel or think; the ability to experience and share the emotions of others; the ability to interpret communicative signs, detect intentionality, and understand social outcomes; the mirror neuron system; and the ability to think about thinking (i.e., metacognition). It can pertain to attending, perceiving, recognizing, describing, interpreting, inferring, imagining, simulating, remembering, reflecting, and anticipating, making it a slippery concept, potentially extending beyond manageable bounds (Allen Reference Allen, Allen and Fonagy2006). The available instruments do not capture this variance in mentalizing among healthy adults. In most studies on supernatural beliefs, mentalizing has been operationalized either with the Empathy Quotient (Baron-Cohen & Wheelwright Reference Baron-Cohen and Wheelwright2004), or with the Eyes test (Baron Cohen et al. Reference Baron-Cohen, Wheelwright, Hill, Raste and Plumb2001), which according to recent studies may not tap mentalistic abilities (Valla et al. Reference Valla, Ganzel, Yoder, Chen, Lyman, Sidari and Wong2010). However, comprehensive and valid methods to assess mentalizing among healthy adults are not easily available. Hence, there is a pressing need for the field to develop better methods for studying how mentalizing and various biases are related to religiosity today.
In sum, we congratulate Norenzayan et al. for the much-needed theoretical synthesis, but we caution against relying too strongly on evidence that, at least for part of the theory, is currently scarce and contradictory. At present, the only thing that can be said is that there “may be” a relationship between mentalizing, certain cognitive biases, and religiosity. Evidence that would warrant using the wording “is” or even “probably is” does not yet exist.
ACKNOWLEDGMENT
This work was supported by the Research Funds of the Academy of Finland (No. 265518).