"Arts, Faculty of"@en . "Psychology, Department of"@en . "DSpace"@en . "UBCV"@en . "Baimel, Adam Sean"@en . "2019-07-16T15:32:12Z"@en . "2019"@en . "Doctor of Philosophy - PhD"@en . "University of British Columbia"@en . "The cognitive sciences of religion have theorized that supernatural agent beliefs are shaped by intuitively-supported psychological processes (e.g., teleological thinking, and mentalizing). And, evidence is accumulating that individual differences in reliance on these intuitions is positively related to religious beliefs and that, on the other hand, broad tendencies for questioning them are negatively related to belief. In this dissertation, I build on this literature by providing first tests of several longstanding and some novel theoretical accounts of (1) when and in what ways broad tendencies for questioning intuitions come to predict belief and (2) when and in what ways mentalizing becomes implicated in beliefs. In Chapter 2 (N = 5284 students, Americans and Indians), I examine how and when tendencies for questioning intuitions (i.e., analytical thinking) is associated with belief by testing three theoretical accounts of this relationship. In Chapter 3 (3 studies; N = 2191 students, Christian and Hindu Americans, Indians), I examine how and when intuitions for reasoning about mental states (i.e., mentalizing) come to support belief in god by testing a different set of three theoretical accounts against the evidence. In Chapter 4 (N = 2027 from 14 societies), I examine predictions generated in Chapter 3 regarding the prevalence and correlates of mentalized deity concepts in a sample that spans diverse scales of societal complexity, market integration, and religious traditions. Results demonstrate that deities are increasingly mentalized the more they are moralized (i.e., attributed with moral knowledge and capacities to punish). Throughout, this work demonstrates the necessity of taking theory-testing in the cognitive sciences of religion across cultures and religious traditions. Methodologically, this work takes an individual difference approach and employs high-powered samples. When suitable, statistical mediation analyses are conducted to test the processes by which intuitions relate to belief. Broadly, results are discussed in terms of their contributions to the refining existing accounts of how and when intuitions come to be implicated in religious beliefs across cultures."@en . "https://circle.library.ubc.ca/rest/handle/2429/71014?expand=metadata"@en . "REASONING ABOUT THE SUPERNATURAL: A CROSS-CULTURAL EXAMINATION OF HOW AND WHEN INTUITIONS SHAPE BELIEF by ADAM SEAN BAIMEL B.A., The University of British Columbia, 2012 M.A., The University of British Columbia, 2015 A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Psychology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) July 2019 \u00C2\u00A9 Adam Sean Baimel, 2019 ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Reasoning about the supernatural: A cross-cultural examination of when and how intuitions shape belief submitted by Adam Sean Baimel in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Psychology Examining Committee: Ara Norenzayan, UBC Psychology Supervisor Susan Birch, UBC Psychology Supervisory Committee Member Benjamin Grant Purzycki, Max Planck Institute for Evolutionary Anthropology Supervisory Committee Member Kristin Laurin, UBC Psychology University Examiner Katherine White, UBC Sauder School of Business University Examiner Tania Lombrozo, Psychology Princeton University External Examiner Additional Supervisory Committee Members: Steven Heine, UBC Psychology Supervisory Committee Member iii Abstract The cognitive sciences of religion have theorized that supernatural agent beliefs are shaped by intuitively-supported psychological processes (e.g., teleological thinking, and mentalizing). And, evidence is accumulating that individual differences in reliance on these intuitions is positively related to religious beliefs and that, on the other hand, broad tendencies for questioning them are negatively related to belief. In this dissertation, I build on this literature by providing first tests of several longstanding and some novel theoretical accounts of (1) when and in what ways broad tendencies for questioning intuitions come to predict belief and (2) when and in what ways mentalizing becomes implicated in beliefs. In Chapter 2 (N = 5284 students, Americans and Indians), I examine how and when tendencies for questioning intuitions (i.e., analytical thinking) is associated with belief by testing three theoretical accounts of this relationship. In Chapter 3 (3 studies; N = 2191 students, Christian and Hindu Americans, Indians), I examine how and when intuitions for reasoning about mental states (i.e., mentalizing) come to support belief in god by testing a different set of three theoretical accounts against the evidence. In Chapter 4 (N = 2027 from 14 societies), I examine predictions generated in Chapter 3 regarding the prevalence and correlates of mentalized deity concepts in a sample that spans diverse scales of societal complexity, market integration, and religious traditions. Results demonstrate that deities are increasingly mentalized the more they are moralized (i.e., attributed with moral knowledge and capacities to punish). Throughout, this work demonstrates the necessity of taking theory-testing in the cognitive sciences of religion across cultures and religious traditions. Methodologically, this work takes an individual difference approach and employs high-powered samples. When suitable, statistical mediation analyses are conducted to test the processes by which intuitions relate to belief. Broadly, iv results are discussed in terms of their contributions to the refining existing accounts of how and when intuitions come to be implicated in religious beliefs across cultures. v Lay Summary Across cultures and throughout history, the majority of humans that have ever lived have believed in god(s) and/or some other supernatural agent(s). Although vastly outnumbered by believers, there are also people who do not believe in the supernatural. The question of why many people do and some people do not believe in the supernatural has spurred a growing body of research in the cognitive sciences of religion. This research suggests that there are certain ways of thinking that promote belief in the supernatural and that when people do not think in these ways that they are less likely to believe. In this dissertation, further evidence is provided for how these differences in ways of thinking come to be related to supernatural beliefs. Far more than previous studies, this dissertation draws from the world\u00E2\u0080\u0099s vast religious diversity in asking and answering questions about psychological and cultural contributions to religious beliefs. vi Preface This dissertation is original, unpublished, independent work by the author, A. S. Baimel. The work presented in Chapter 2 was covered by UBC Ethics Certificates #H15-03085, #H15-03122, #H12-03221, and #H16-02712. The work presented in Chapter 3 was covered by UBC Ethics Certificate #H15-03085 and #H13-00671. The work presented in Chapter 4 was covered by UBC Ethics Certificate #H13-00671. The data presented in Chapter 2 was, in part, collected by Cindel White. The data presented in Chapter 4 was collected by a large team of anthropologists/psychologists at field sites spread out all over the world: B. G. Purzycki, M. Lang, C. L Apicella, Q. Atkinson, A. Bolyanatz, E. Cohen., E. Kundtov\u00C3\u00A1 Klocov\u00C3\u00A1, C. Handley, C. Lesorogol, S. Mathew, R. A. McNamara, C. Moya, C. Placek, M. Soler, T. Vardy, J. L. Weigel, A. K. Willard, D. Xygalatas. These are members of the Cultural Evolution of Religion Research Consortium lead by A. Norenzyan and J. Henrich. All analyses presented in this dissertation were conducted solely by the author, A. S. Baimel. vii Table of Contents Abstract ........................................................................................................................................................................ iii Lay Summary ................................................................................................................................................................ v Preface .......................................................................................................................................................................... vi Table of Contents ........................................................................................................................................................ vii List of Tables ................................................................................................................................................................. x List of Figures .............................................................................................................................................................. xi Acknowledgements ..................................................................................................................................................... xii Dedication ................................................................................................................................................................... xiii CHAPTER 1 INTRODUCTION .................................................................................................. 1 Intuition & belief ................................................................................................................................................... 3 Mentalizing & belief ............................................................................................................................................. 8 A cross-cultural examination of mentalized god concepts and beliefs ............................................................... 11 Summary of key questions .................................................................................................................................. 12 CHAPTER 2 INTUITION AND BELIEF ................................................................................. 14 Introduction ................................................................................................................................................................ 14 Does analytical thinking always lead to the inhibition of intuitions? ................................................................. 15 What are the sources of human intuitions? ......................................................................................................... 16 Testing the three models ..................................................................................................................................... 17 Methods ....................................................................................................................................................................... 18 Sample ..................................................................................................................................................................... 18 Measures .................................................................................................................................................................. 19 Measures of belief ............................................................................................................................................... 19 Measures of analytical thinking .......................................................................................................................... 20 Demographics ..................................................................................................................................................... 21 Results .......................................................................................................................................................................... 26 Dual process model of belief .............................................................................................................................. 26 Expressive rationality model of belief ................................................................................................................ 30 Counter-normative rationality model of belief ................................................................................................... 36 Discussion .................................................................................................................................................................... 36 Limitations and future directions ........................................................................................................................ 38 CHAPTER 3 MENTALIZING AND BELIEF ......................................................................... 40 Introduction ................................................................................................................................................................ 40 Extending the cross-cultural reach of the cognitive sciences of religions .......................................................... 41 How does mentalizing come to support belief in god? ....................................................................................... 41 When and what kinds of beliefs are supported by mentalizing? ........................................................................ 45 viii Summary of research questions .......................................................................................................................... 48 Study 1: Mentalizing and Belief in Three Samples ................................................................................................. 49 Methods ....................................................................................................................................................................... 50 Measures .................................................................................................................................................................. 51 Individual differences in mentalizing ................................................................................................................. 51 Mental state attribution to god ............................................................................................................................ 51 Generalized anthropomorphism .......................................................................................................................... 52 Belief in god ........................................................................................................................................................ 52 Results .......................................................................................................................................................................... 53 Discussion .................................................................................................................................................................... 58 Studies 2A and 2B: Mentalizing and belief in American Hindus and Christians ................................................ 60 The relative contributions of cognition and culture to beliefs ............................................................................ 61 What parts of \u00E2\u0080\u0098religion\u00E2\u0080\u0099 should mentalizing predict? ......................................................................................... 62 Methods ....................................................................................................................................................................... 63 Pre-registration ........................................................................................................................................................ 63 Sample ..................................................................................................................................................................... 63 Measures .................................................................................................................................................................. 65 Anthropomorphism ............................................................................................................................................. 65 Individual differences in mentalizing ................................................................................................................. 65 Mental state attribution to god ............................................................................................................................ 65 Belief in god and a personal god ......................................................................................................................... 66 Belief in and attributed qualities of other supernatural targets ........................................................................... 67 Analytical thinking .............................................................................................................................................. 67 Religious upbringing ........................................................................................................................................... 67 Other indicators of \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 .......................................................................................................................... 68 Demographics ..................................................................................................................................................... 68 Results .......................................................................................................................................................................... 69 Does the relationship between individual differences in mentalizing and belief in a personal god vary between different religious populations? .......................................................................................................................... 69 Does generalized anthropomorphism mediate the relationship between mentalizing and belief in a personal god? ..................................................................................................................................................................... 69 Does mental state attribution to god mediate the relationship between mentalizing and belief in a personal god? ..................................................................................................................................................................... 70 Does belief in a personal god mediate the relationship between mentalizing and mental state attribution to god? ..................................................................................................................................................................... 71 Selecting between the narrow anthropomorphism and broad anthropocentric knowledge hypotheses ............. 73 Are narrowly anthropomorphic conceptions of god supported by intuitive thinking? ....................................... 75 Do individual differences in mentalizing support belief in god after controlling for religious upbringing? ...... 76 Does mentalizing specifically predict beliefs in a personal god? ....................................................................... 77 Does mentalizing support all types of supernatural beliefs? .............................................................................. 78 Moral knowledge and agency attributions to personal gods ............................................................................... 88 General Discussion ..................................................................................................................................................... 91 Accounting for the contributions of mentalizing to belief .................................................................................. 92 Limitations and Future Directions ...................................................................................................................... 96 CHAPTER 4 A CROSS-CULTURAL EXAMINATION OF MENTALIZED DEITY CONCEPTS AND BELIEF ........................................................................................................ 99 ix Introduction ................................................................................................................................................................ 99 Methods ..................................................................................................................................................................... 101 Sample ................................................................................................................................................................... 101 Measures ................................................................................................................................................................ 101 Deity selection .................................................................................................................................................. 102 Belief ................................................................................................................................................................. 104 Mentalistic deity concepts ................................................................................................................................. 104 Moral knowledge .............................................................................................................................................. 105 Analytical decisions and methods ..................................................................................................................... 106 Results ........................................................................................................................................................................ 106 Mentalistic deity beliefs across cultures ........................................................................................................... 106 Predicting belief ................................................................................................................................................ 110 When are deities mentalized? ........................................................................................................................... 113 Discussion .................................................................................................................................................................. 117 Limitations and future directions ...................................................................................................................... 119 CHAPTER 5 CONCLUSION ................................................................................................... 121 What does analytical thinking have to do with belief? ..................................................................................... 121 What does mentalizing have to do with belief? ................................................................................................ 123 Why cross-cultural research is necessary? ........................................................................................................ 126 Limitations and future directions ...................................................................................................................... 127 Bibliography .............................................................................................................................................................. 130 Appendix ................................................................................................................................................................... 146 x List of Tables Table 2.1 Sample characteristics of the data sets .......................................................................... 22 Table 2.2 Included measures in each dataset ................................................................................. 23 Table 2.3 Summary statistics of focal measures by sample (means and standard deviations) ...... 24 Table 2.4 Correlations [95% confidence intervals] by sample of focal variables ......................... 25 Table 2.5 Predicting belief in god from cognitive reflections test score ....................................... 28 Table 2.6 Predicting belief in god from faith in intuition .............................................................. 29 Table 2.7 Testing the expressive rationality model \u00E2\u0080\u0093 cognitive reflections test ............................ 31 Table 2.8 Testing the expressive rationality model \u00E2\u0080\u0093 faith in intuition scale ................................ 34 Table 3.1 Sample characteristics of the three samples .................................................................. 50 Table 3.2 Means and standard deviations of focal variables by sample ........................................ 53 Table 3.3 Study 1 mediation models of the narrow anthropomorphism hypothesis ..................... 55 Table 3.4 Study 1 mediation models of the anthropocentric knowledge hypothesis .................... 57 Table 3.5 Participant demographics in studies 2A and 2B ............................................................ 64 Table 3.6 The contributions of analytical thinking to anthropomorphic god beliefs .................... 76 Table 3.7 Linear regression models of cultural and cognitive predictors of belief in a personal god ................................................................................................................................................. 77 Table 3.8 Predicting belief in an abstract god ............................................................................... 79 Table 3.9 The moderating effects of moral knowledge on the contributions of mentalizing and anthropomorphism to belief ........................................................................................................... 83 Table 3.10 Predicting belief in karma across samples ................................................................... 87 Table 4.1 Selected deities and belief items summaries by site .................................................... 103 Table 4.2 Means and 95% confidence intervals for the moral knowledge index by deity and site ..................................................................................................................................................... 106 Table 4.3 Testing for differences in mentalistic beliefs across deities and traditions ................. 109 Table 4.4 The contributions of moral knowledge attributions to mentalistic beliefs .................. 114 xi List of Figures Figure 2.1 The moderating effect of political orientation on the relationship between CRT and belief in god ................................................................................................................................... 32 Figure 2.2 The moderating effect of political orientation on the relationship between faith in intuition and belief in god .............................................................................................................. 35 Figure 3.1 The proposed mediation models .................................................................................. 45 Figure 3.2 Narrow anthropomorphism hypothesis mediation model ............................................ 55 Figure 3.3 Anthropocentric knowledge hypothesis mediation model ........................................... 57 Figure 3.4 Testing the narrow anthropomorphism hypothesis in two samples ............................. 71 Figure 3.5 Testing the anthropocentric knowledge hypothesis in two samples ............................ 72 Figure 3.6 Distribution and correlations of agency and experience attributions ........................... 74 Figure 3.7 Agency and experience models .................................................................................... 75 Figure 3.8 Bivariate correlations [95% confidence intervals] of mentalizing and religiosity indicators by samples ..................................................................................................................... 78 Figure 3.9 Correlations between mentalizing, anthropomorphism and belief in varied supernatural targets ........................................................................................................................ 80 Figure 3.10 Interactions of mentalizing and anthropomorphism with moral knowledge attributions \u00E2\u0080\u0093 Hindu sample ........................................................................................................... 84 Figure 3.11 Interactions of mentalizing and anthropomorphism with moral knowledge attributions \u00E2\u0080\u0093 Christian sample ...................................................................................................... 84 Figure 3.12 Mean belief in and moral knowledge attribution to karma by sample ....................... 85 Figure 3.13 The moderating effects of moral knowledge attributions on the contributions of mentalizing and anthropomorphism to belief in karma ................................................................. 88 Figure 3.14 The mediating role of moral knowledge attributions on the relationship between agency and belief in a personal god ............................................................................................... 90 Figure 4.1 Density plots of mentalistic belief (mean \u00E2\u0080\u0098think\u00E2\u0080\u0099 and \u00E2\u0080\u0098worry\u00E2\u0080\u0099) composite by site and deity ............................................................................................................................................. 108 Figure 4.2 Exponentiated mean estimates of tradition differences in mentalistic beliefs by deity ..................................................................................................................................................... 110 Figure 4.3 Predicted probabilities of belief in local deities by site as predicted by mentalistic beliefs ........................................................................................................................................... 112 Figure 4.4 Estimated odds ratios of mentalistic beliefs by tradition, deity, and moral knowledge attributions ................................................................................................................................... 115 Figure 4.5 Predicted probabilities of mentalistic beliefs by site, deity and moral knowledge attributions ................................................................................................................................... 116 xii Acknowledgements I am truly thankful for all the support I have received throughout my graduate training. Thanks to Ara Norenzayan and Sue Birch for taking me on as a graduate student and continuously providing me with all the support I could have possibly needed in the last five years. Seriously \u00E2\u0080\u0093 you have both been immensely helpful throughout this process. Thanks to Steve Heine for looking after me while Ara was away on sabbatical and starting his family, and for our shared interests in research questions that have unfortunately not made it into this dissertation. And thanks to Ben Purzycki for always looking for ways to involve me in your research, and for laughing at most of my jokes. I could not ask for a better committee, so double thanks to all of you for the time and effort you have spent thinking about this work. I thank SSHRC (Doctoral Award #6567), the Cultural Evolution of Religion Research Consortium (funded by SSHRC and the John Templeton Foundation), and the Understanding Unbelief Project managed by the University of Kent (John Templeton Foundation Grant ID #60624) for funding I received throughout my time as a graduate student. Thanks to my cohort and friends in the psychology department at UBC. And thanks to those who never asked me too many questions about my research because we had other things to talk about (Kevin Nicol, Stephanie Chou, Gena Ellett, Olga Fedorov, Seana Sterner and Emmanuelle Gabay). Thanks to Sami Berger and Corey Baimel. And, special thanks to Taylor Basso for six years of love and support. xiii Dedication To my parents, Linda and Ed Baimel. 1 Chapter 1 Introduction The human capacity to believe in things unseen is astounding. Indeed, recent estimates indicate that 83.6% of the global population is religiously affiliated and that the proportion of religiously unaffiliated individuals will decrease over the coming decades (Pew Research Center, 2015). Although these estimates cannot speak to the specific beliefs held by the religiously-minded global majority \u00E2\u0080\u0093 religions (i.e., complex cultural systems of beliefs, practices and experiences) are most clearly distinguished from other cultural institutions by their deep connection to belief and devotion to unseen supernatural forces and agents (Atran & Norenzayan, 2004). The question is, then: what can account for the global success of religion? This is critical to understanding the psychology of the lived experiences of the majority of humans that have ever lived. Importantly, substantial progress has been made in piecing together the puzzle that is the cultural evolutionary success of religion. For instance, we have come to know quite a bit about how specific types of religious beliefs such as belief in moralizing deities with the power and willingness to punish bad behaviours promote prosociality (Norenzayan & Shariff, 2008; Shariff, Willard, Andersen, & Norenzayan, 2015). Critically, the prosocial effects of beliefs in supernatural agents that are particularly concerned with interpersonal human behaviours have been demonstrated to hold across vast geographical distances and in populations of adherents of diverse religious traditions (Purzycki et al., 2016). Cultural evolutionary processes have entwined these beliefs with the ways in which individuals demonstrate their commitment to gods \u00E2\u0080\u0093 rituals \u00E2\u0080\u0093 that also have extensive prosocial consequences on tightening the bonds between even otherwise unrelated individuals (e.g., Mogan, Fischer, & Bulbulia, 2017; Xygalatas et al., 2013). Over the course of human cultural evolutionary history, committed and cohesive religious groups may have 2 had an important cultural advantage over others \u00E2\u0080\u0093 for their beliefs and practices provided a solution to the problem of sustaining cooperation in large groups (Norenzayan et al., 2016). Not forgotten by the most ambitious of the cultural evolutionary theories of religion, however, are the psychological origins of these beliefs. One prevailing view holds that many of the culturally successful forms of religious beliefs have survived on the cultural marketplace because they are particularly well-matched with and even emerge from evolved cognitive intuitions for reasoning about our social environments and the social agents that occupy it (e.g., Atran & Norenzayan, 2004; Boyer, 2001; Gervais, 2013). If in some way, religious belief is grounded in evolved cognitive mechanisms \u00E2\u0080\u0093 than this might help explain the pervasiveness of particular types of religious beliefs cross-culturally. The primary focus of this dissertation is to elaborate on the particular mechanisms by which intuitions come to support religious beliefs and particularly belief in gods. Specifically, in Chapter 2 this dissertation addresses open-questions regarding how broad tendencies for reliance on one\u00E2\u0080\u0099s intuitions come to be related to belief in god across samples of undergraduate students at a Canadian university, and a broader sample of the Canadian, American and Indian populations. In Chapter 3, I examine how and when specific social intuitions for reasoning about other minds \u00E2\u0080\u0093 mentalizing - come to be related to belief in gods and other supernatural agents/forces across samples of undergraduate students at a Canadian university, a broader sample of American and Indian populations, and targeted samples of American Christians and American Hindus samples. In Chapter 4, I provide an examination of \u00E2\u0080\u0098mentalized\u00E2\u0080\u0099 god concepts and belief in an immensely diverse cross-cultural sample. The diverse samples employed throughout are a useful contribution to the psychological literature on religion for sampling from populations typically ignored by the psychological sciences; making it less WEIRD (Western, 3 Educated, Industrialized, Rich and Democratic; Henrich, Heine, & Norenzayan, 2010). Above and beyond issues of global representation, the psychology of religion has narrowly focused on Christianity \u00E2\u0080\u0093 overlooking the immense diversity of religious beliefs and practices present across cultures (Norenzayan, 2016). Intuition & belief From our evolutionary ancestors, humans have inherited a vast array of reliably developing sophisticated cognitive systems that (1) shape our expectations of how the world around us \u00E2\u0080\u0098works\u00E2\u0080\u0099 and (2) make living in this world at the very least somewhat easier at least some of the time (Barrett, 2014; Henrich, 2015; Laland, 2017). While some of these inherited faculties seem to produce more rigid outcomes \u00E2\u0080\u0093 such as the expectation that dropped objects will pummel towards the ground, others are much more flexible, and functionally so (Baron-Cohen, 1999; Baron-Cohen, Wheelwright, Spong, Scahill, & Lawson, 2001). Unlike some intuition for \u00E2\u0080\u0098gravity\u00E2\u0080\u0099- specific religious concepts that we might encounter in the world today are far too evolutionarily novel for us to have specific cognitive mechanisms that produce them for specifically religious purposes. However, that does not mean that our evolved psychology cannot still shape and constrain the forms that successful religious concepts will (and do) take. For example, culturally-elaborated beliefs about the origins of life (e.g., it was created by a supernatural agent) and the afterlife (e.g., that there is some continuation of the self after death) are central to many religious systems across cultures. And while differences in the details can distinguish between religious traditions, the cognitive sciences of religion have argued that part of the reason why most religious systems ask and provide answers to questions about the origins of life and the afterlife in the first place is that the evolved architecture of the human mind leaves us prone to seeing design in the natural world (i.e., teleological thinking) and to thinking that our 4 minds are separate from our bodies (i.e., dualism), and can thus continue to exist after their physical death. In support of this view, developmental psychologists have demonstrated that these tendencies are early developing (but still only potential, and weak, evidence for them being evolved-systems). For example, children typically prefer and offer more purpose-based, than non-purposed based, explanations when asked questions about the existence of features of the natural world (Kelemen, 2004). And, young children are more likely to say that psychological traits rather than physical traits persist after death (Bering & Bjorklund, 2004; but this may require early exposure to Abrahamic religious contexts; e.g., see Astuti & Harris, 2008). More broadly, this view that religious concepts are \u00E2\u0080\u0098by-products\u00E2\u0080\u0099 of otherwise adaptive evolved cognitive processes has long prevailed in the cognitive sciences of religion (Atran & Norenzayan, 2004; Barrett, 2004; Bloom, 2006; Boyer, 2001). If religious beliefs are founded in intuitively supported cognitive faculties, and analytical thinking can suppress or override intuitions, does that imply that religious belief is negatively related to analytical thinking? In addressing this question, extensive correlational research has demonstrated that tendencies for overcoming the pull of one\u00E2\u0080\u0099s intuitions is associated with decreased religiosity (e.g., Study 1, Gervais & Norenzayan, 2012; Pennycook, Cheyne, Seli, Koehler, & Fugelsang, 2012; Shenhav, Rand, & Greene, 2012). The most robust of these findings are centered around the \u00E2\u0080\u0098Cognitive Reflections Test\u00E2\u0080\u0099 (CRT; Frederick, 2005). In the CRT, participants are presented with a series of logic puzzles (e.g., \u00E2\u0080\u009CIf it takes 5 machines 5 minutes to make 5 widgets, how many minutes would it take 100 machines to make 100 widgets?\u00E2\u0080\u009D) to which there is an intuitively compelling but wrong answer (i.e., 100 minutes), and a correct answer (i.e., 5 minutes). Although simple in its design, this measure reliably differentiates between those who tend to go with their default \u00E2\u0080\u0098gut\u00E2\u0080\u0099 response and those who make the effort to reason more carefully 5 and analytically about the questions - even over time and repeated tests, and controlling for cognitive ability (Stagnaro, Pennycook, & Rand, 2018). And in a meta-analysis of 31 studies (N = 15 078), Pennycook, Ross, Koehler, & Fugelsang (2016) demonstrated that individuals (N.B., mostly Americans recruited on Amazon\u00E2\u0080\u0099s M-Turk) with more correct answers on the CRT were, to a small extent, robustly less endorsing of religious beliefs (r = -.18, .95CI = [-.21, -.16]). This meta-analysis, however, did not specifically isolate the association between analytical thinking and belief in god (religious belief was broadly defined \u00E2\u0080\u0093 e.g., belief in heaven, hell, devil, and god). This work forms the core of what can be called the dual process model of religious belief. In this perspective, the human tendency for religious-like thinking emerges from the everyday functioning of evolutionary \u00E2\u0080\u0098older\u00E2\u0080\u0099 cognitive systems whose output is constrained only by more conscious, effortful, and careful reasoning (in following with Evans, 2003). At an extreme, this view holds that deliberation and rejection of human intuitions should always lead to the rejection of belief. However, more nuanced takes on cognitive and cultural precursors of non-belief or atheism emphasize that questioning one\u00E2\u0080\u0099s intuitions is just one of several routes to disbelief (Norenzayan & Gervais, 2013). Put differently still, a core prediction of this model is that it should be, all else being equal, more uncommon that individuals reason their way into their religious beliefs than it is for individuals to reason their way out of them. However, one need not look too deeply into the theological and philosophical record of treatises (i.e., especially of Christian origins) on religious belief to realize that many a religious scholar have deeply reasoned their way into their religious beliefs. Dating back to the 4th and 5th century, the careful and deeply analytical works of St. Augustine of Hippo have been the cornerstone of Christian philosophically reasoned arguments for believing in god. Today \u00E2\u0080\u0093 a quick 6 Google search for \u00E2\u0080\u009CI\u00E2\u0080\u0099m doubting my faith\u00E2\u0080\u009D will produce several hundred thousand documented cases of religious doubt and how to overcome it with some combination of both intuitive faith and reason. Although these exceptions might not break the on-average prediction of the dual process model of religious belief \u00E2\u0080\u0093 it does raise an important question as to whether there are reliably detectable circumstances under which analytical thinking can promote religiosity rather than dampen it. Much like Augustine who spent a great deal of time coming up with reasons to convince himself and others that his conversion to Christianity was the right decision (Jacoby, 2017) \u00E2\u0080\u0093 individuals are deeply motivated to justify their previously-held commitments and beliefs and sometimes go to incredible lengths to confirm their biases (Nickerson, 1998). Although overriding one\u00E2\u0080\u0099s intuitions might be a good way to reason through all the available evidence \u00E2\u0080\u0093 an alternative account suggests that better analytical thinkers might be even better at finding ways to confirm their biases regardless of the evidence. Indeed, the expressive rationality model holds that individuals deploy their analytical thinking to justify previously-held beliefs and most dramatically so when these beliefs are strong indicators of their social affiliations (Kahan, 2017). That is, in some cases rationality is deployed to strengthen confirmation biases as an identity-protective strategy. For example, Kahan & Stanovich (2016) demonstrated that belief in evolution in religious and non-religious Americans are most different (i.e., polarized) amongst analytical thinkers from either camp. No work \u00E2\u0080\u0093 yet \u00E2\u0080\u0093 has directly tested whether and how this model applies to belief in god. Indeed, already-devoted analytical thinkers \u00E2\u0080\u0093 like Augustine \u00E2\u0080\u0093 might prove to be the deepest believers, or at the very least no more or less believing than their intuitive and devoted counterparts. 7 One limitation of the extant research on the relationship between intuition and religious belief is that it has been largely limited to North American and Christian samples (with some exceptions (e.g., studies of Turkish Muslims with results of similar size and magnitude to those reported from American samples; Yilmaz & Saribay, 2016). In an effort to more broadly assess the dual process model of religious belief, Gervais et al. (2017) deployed the CRT and a measure of belief in god in 13 culturally and religiously diverse samples (e.g., Buddhists in Singapore, Hindus in Mauritius, Muslims in the United Arab Emirates, and secularized nations such as the Czech Republic). In aggregate, this effort produced a reliable but very small estimate of the effect of analytical thinking on belief in god in the direction predicted by the dual process model of belief (i.e., an average 2-point decrease on a 1-100 scale of belief in god with higher analytical thinking). While providing some cross-cultural support for the dual process model of religious belief, the observed effect of CRT was also found to be stronger in more religious countries \u00E2\u0080\u0093 an interaction that the dual process model itself cannot directly explain. Most interestingly, however, is some initial evidence that in the United Kingdom the effect was reversed, with analytical thinking predicting greater belief in god. From this, Gervais et al. (2017) propose a third account of this relationship, what I call the counter-normative rationality model. This model proposes that the content of our intuitions are not just the output of evolved cognitive systems but also (at least in part) culturally-learned norms (Henrich, 2015). And thus, part of overriding one\u00E2\u0080\u0099s intuitions is to question prevailing cultural norms. And thus, the effect so far observed of analytical thinking on religious beliefs may be an expression of questioning the prevailing norm of religiosity in majority-religiously affiliated cultures. When one\u00E2\u0080\u0099s cultural context is highly secularized \u00E2\u0080\u0093 questioning the norm might (somewhat ironically) then predict increased religiosity \u00E2\u0080\u0093 which is what Gervais et al. (2017)\u00E2\u0080\u0099s 8 initial data from the United Kingdom might suggest. More broadly, analytical thinking has been previously demonstrated to lead individuals to question norms. For example, in cooperative dilemmas \u00E2\u0080\u0093 analytical thinkers are the most likely to override normative intuitions for fairness and find ways to benefit themselves through defection (Rand, 2016). In Chapter 2 of this dissertation, the predictions of these three models regarding how intuition comes to be related to religious belief are explicitly compared and contrasted. In Chapter 2, these models are tested against the evidence from samples of undergraduate students from UBC, and broader samples of the Canadian, American and Indian (i.e., mostly Hindu) populations. No previous research has tested these three models simultaneously in the same study. Given the recent failure to replicate the causal (i.e., experimental) effect of induced analytical thinking on the erosion belief in god (Sanchez, Sundermeier, Gray, & Calin-Jageman, 2017 replicating Study 2, Gervais & Norenzayan, 2012), there is all the more reason to better understand more precisely the ways in which analytical thinking comes to be correlated with beliefs. Mentalizing & belief The question of how intuition comes to be related to belief is an important one that has generated a growing body of research \u00E2\u0080\u0093 but this work has said less about the specific intuitions that contribute to and sustain religious beliefs. On this front, however, the cognitive scientists of religion have long theorized and, more recently, empirically focused in on the importance of a small set of social intuitions implicated in perceiving and reasoning about minds. Gods, ghosts, spirits and other supernatural agents are commonly believed to be strikingly human-like. Even when gods take on ostensibly non-human forms they are often still conceptualized as intentional agents with human-like mental states (Purzycki, 2013). The pervasiveness of belief in mentalized supernatural agents is a puzzle that has generated extensive 9 empirical interest in the cognitive sciences of religion. This approach to studying religious beliefs theorizes that observed supernatural forms are expressions of evolved cognitive intuitions (Atran & Norenzayan, 2004; Bloom, 2006; Boyer, 2001; Kelemen, 2004; McCauley, 2011). In the case of mentalized gods (i.e., supernatural agents with human-like mental states), one key hypothesis holds that believers intuitively reason about supernatural agents as if they had human-like mental states because humans are so prone to making sense of their everyday social environments by mentalizing (i.e., reasoning about the mental states of other people; Gervais, 2013; Guthrie et al., 1980). One version of this hypothesis, the broad anthropomorphism hypothesis, has a long intellectual history (e.g., Barrett, 2004; Feuerbach, 1957; Guthrie et al., 1980; Hume, 1779). In this view, human capacities for mentalizing encourage generalized anthropomorphism \u00E2\u0080\u0093 the attribution of human-like mental characteristics to all types of non-human things and beings, infusing agentic qualities to the natural world (e.g., Atran & Medin, 2008; Guthrie et al., 1980). The persistent experience of perceiving human-like qualities in the natural world gives rise to the belief that there are human-like entities everywhere in the natural world. This generalized anthropomorphism is argued to facilitate animistic beliefs, which when reflected upon may lead individuals to conceptualize much of the seemingly unexplainable events in their environments as resulting from the wills and whims of unobservable supernatural minds. In this view, belief in gods and other supernatural agents are simply further and perhaps passive output from this cognitive system that attributes human-like agency to our environments. However, Willard & Norenzayan (2013) empirically tested this longstanding hypothesis in American and Canadian samples and found that the generalized anthropomorphism of nature, technology and animals predicted paranormal beliefs such as extrasensory perception and astrology, but was unrelated to belief in God. 10 In contrast to the broad anthropomorphism hypothesis, the narrow anthropomorphism hypothesis holds that believers actively deploy their capacities for mentalizing to specifically attribute mental states to and reason about the minds of specific supernatural agents. And indeed, recent work demonstrates that reasoning about supernatural minds engages everyday social-cognitive capacities for mentalizing. For example, thinking about or praying to God activates brain regions implicated in mental state reasoning (Kapogiannis et al., 2009; Schjoedt, St\u00C3\u00B8dkilde-J\u00C3\u00B8rgensen, Geertz, & Roepstorff, 2009). Furthermore, this hypothesis holds that mentalizing supports belief in part because the projection of human-like mental states to supernatural agents renders the supernatural intuitively compelling. It follows that deficits in mentalizing, as found in the autism spectrum, are associated with reductions in belief in God. This hypothesis has found empirical support, holding constant demographic variables (Caldwell-Harris, Murphy, Velazquez, & McNamara, 2011; Gray, Jenkins, Heberlein, & Wegner, 2011; Lindeman & Lipsanen, 2016a; Norenzayan, Gervais, & Trzesniewski, 2012). However, the replicability, magnitude, and even direction of the relationship between individual differences in mentalizing and religious belief is being actively debated (e.g., Jack, Friedman, Boyatzis, & Taylor, 2016; Lindeman, Svedholm-H\u00C3\u00A4kkinen, & Lipsanen, 2015; Reddish, Tok, & Kundt, 2016; Vonk & Pitzen, 2016; Wlodarski & Pearce, 2016), calling for further research and refinements. A third theoretical model - the anthropocentric knowledge account - holds that given one\u00E2\u0080\u0099s belief in God, individuals may be increasingly motivated to apply what they know about human mental states to fill in the gaps of the informational vacuum presented by the problem of inferring the traits of a believed, but unseen agent. In this account, people come to believe in God, driven partly by mentalizing tendencies and other factors (such as cultural learning). In turn, belief in God may well lead to the attribution of more human-like mental qualities to God. Thus, rather than a 11 mentalized God being more believable, this account reverses the causal arrow and says a believed God is more mentalized. This is consistent with the principle of accessible anthropocentric knowledge, leading people to treat familiar human minds as a rich inferential base to make sense of unfamiliar nonhuman minds (Epley, Waytz, & Cacioppo, 2007; Hume, 1757). In Chapter 3 of this dissertation, data is presented that reopens this debate by moving beyond a description of the relationship between mentalizing and belief and returning to the question of how mentalizing facilitates belief in supernatural agents in a direct empirical test of the broad, narrow and anthropocentric hypotheses. That is, in addition to providing new evidence of the replicability of the correlation between mentalizing and belief \u00E2\u0080\u0093 this chapter tests three theoretical accounts of this relationship. Specifically, the three proposed models are tested as mediation models and compared in four distinct samples: University of British Columbia undergraduate students (Study 1), American Christians (Study 1 and 2B), Indian Hindus (Study 1), and American Hindus (Study 2A). A cross-cultural examination of mentalized god concepts and beliefs Despite the vast religious diversity across cultures, the existing psychological literature asking nearly any questions about \u00E2\u0080\u0098religion\u00E2\u0080\u0099 have largely focused on Western samples and almost exclusively on Christians. Indeed, the extant literature on religion is far from as theologically-diverse as the world (Norenzayan, 2016). Of course, there have been a number of notable exceptions (Gervais et al., 2017; Purzycki, 2013, 2016; Purzycki et al., 2016; Saribay & Yilmaz, 2017, 2017). The WEIRD problem in psychology is nothing new (Henrich et al., 2010) \u00E2\u0080\u0093 but to broadly theorize and test predictions about \u00E2\u0080\u0098religion\u00E2\u0080\u0099 in any meaningful sense researchers must sample from outside the typically-studied Western populations. Indeed, Western populations occupy a very narrow band of the world\u00E2\u0080\u0099s socio-demographic diversity and an equally narrow 12 bandwidth of the world\u00E2\u0080\u0099s religious diversity. And thus, it remains an open question as to how some core findings in the psychology of religion would hold up outside of the typically examined samples. In Chapter 4 of this dissertation, a systematic cross-cultural examination of the prevalence of mentalized god concepts in a large sample of participants from 14 diverse societies was conducted. Unlike many existing studies of religious beliefs, this sample spans the various scales of human social complexity (from hunter-gatherer groups to fully-market integrated urban samples) and its religious traditions are remarkably diverse, including animism, shamanism, polytheism, and monotheism. Data from these samples are employed to in a first cross-cultural test of the prevalence, antecedents and correlates of mentalized deity concepts. Summary of key questions This dissertation provides tests of three models of when and how analytical thinking comes to predict belief, three models of how intuitions for reasoning about minds come to predict belief, and an examination of the prevalence and correlates of mentalized god concepts in a cross-cultural sample. In Chapter 2, the focus will be on replicating predicted effects of the dual-process model of religious belief and providing first tests of the alternative accounts \u00E2\u0080\u0093 i.e., the expressive rationality and counter-normative rationality models. In two studies, Chapter 3 tests three accounts of how mentalizing comes to be implicated in belief: (1) the broad anthropomorphism hypothesis, the (2) narrow anthropomorphism hypothesis, and (3) the anthropocentric knowledge hypothesis. In Chapter 4, a large cross-cultural sample is examined for evidence of mentalized god concepts, their antecedents and relationship to belief in culturally diverse samples. Broadly speaking, this dissertation provides novel direct tests of longstanding theory in the cognitive science of religion 13 of religion while demonstrating the importance of taking a culturally, and religiously diverse sampling approach. 14 Chapter 2 Intuition and Belief Introduction Religion is ubiquitous. To explain the cross-cultural prevalence of religious beliefs, researchers have long directed their theorizing towards our evolved psychology as the potential source of religious-like thinking. If religious-like cognitions result from evolved (and potentially, sometimes assumed to be, universal) psychological processes, then the persistence of religion in human-life across cultures and throughout history can be, at least in part, accounted for. And indeed, a growing body of evidence suggests that religious-like cognitions are \u00E2\u0080\u0098natural\u00E2\u0080\u0099 (e.g., Bloom, 2006; Kelemen, 2004; McCauley, 2011). That is, there is something about the evolved design of the human mind that makes us prone to religious-thinking (and especially so given relevant cultural input throughout the lifespan; e.g., observing others\u00E2\u0080\u0099 commitment to supernatural agents; Gervais & Najle, 2015; Gervais, Willard, Norenzayan, & Henrich, 2011; Willard & Cingl, 2017; Willard & Norenzayan, 2013). Central to this view \u00E2\u0080\u0093 the dual process model of belief \u00E2\u0080\u0093 is the growing body of evidence for a small but robust correlation between broad tendencies for overriding one\u00E2\u0080\u0099s intuitions and decreased (broadly defined) religious beliefs (Pennycook, Ross, et al., 2016). Although robust (at least in North American samples), the correlation between analytical thinking and decreased religiosity says very little about why this relationship exists in the first place. Fruitful but somewhat adjacent research programs have highlighted the kinds of evolved intuitions that give rise to religious-like cognitions (e.g., teleological thinking; Kelemen, 2004) that do indeed require cognitive effort to override (Kelemen, Rottman, & Seston, 2013). And thus, the assumption is often made that the observed negative relationship between broad tendencies for analytical thinking and religiosity reflects tendencies for inhibiting specific intuitions (like teleological 15 thinking) that give rise more directly to religious-like thinking. However, the extent to which these are indeed complementary findings remains to be determined by future research. Moreover, the extant literature ignores alternative accounts of (1) in what ways analytical thinking is employed in everyday reasoning, and (2) alternative (i.e., cultural) sources of human intuitions. It is in addressing these latter two points that the current study stands to contribute to the cognitive sciences of religion. Does analytical thinking always lead to the inhibition of intuitions? A core tenet of the dual process model of belief is that analytical thinking is employed in overriding one\u00E2\u0080\u0099s intuitions. If religious-like cognitions are indeed intuitive, then the beliefs of individuals who deploy their analytical thinking should converge on not believing (by virtue of inhibiting the intuitions that give rise to these beliefs). However, recent evidence has demonstrated that analytical thinking is often deployed in bolstering one\u00E2\u0080\u0099s otherwise-held convictions and not necessarily inhibiting intuitively-supported ones. That is, analytical thinkers are sometimes likely to use their effortful reasoning to justify their pre-existing beliefs, even despite evidence to the contrary (perhaps by being better able to discount it; Kahan, 2017). Kahan & Stanovich (2016) argue that this is especially likely when the belief in question is social identity-congruent; in which case analytic thinking can be deployed in an identity-protective manner. This expressive rationality model of belief would thus predict that analytical thinking should (a) increase the endorsement of religious belief for those belonging to social groups in which religiosity is an important identifying feature (e.g., politically conservative individuals) but (b) decrease the endorsement of religious belief in individuals from non-religious social contexts (e.g., politically liberal individuals). If supported, this model questions the extent to which the observed negative relationship between 16 analytical thinking and religious belief is owed solely to the inhibition of intuitions that give rise to religious-like cognitions. What are the sources of human intuitions? Another core tenet of the dual process model of belief is that the intuitions being suppressed by analytical thinking are the output of evolved cognitive systems. However, the extant theoretical models of the intuitive source of religious cognitions may be missing an important piece of the puzzle by not considering how religiosity is culturally-learned. Humans have evolved to be particularly capable of learning cultural information from others, including normative beliefs and practices (Henrich, 2015). Indeed, young children are \u00E2\u0080\u0098promiscuous normativists\u00E2\u0080\u0099 and can infer strict norms (and begin to enforce them) from even single observations of others\u00E2\u0080\u0099 behaviours (Schmidt, Butler, Heinz, & Tomasello, 2016). The efficiency with which we learn norms from others enables us to successfully navigate our varied cultural and ecological environments throughout the lifespan. Moreover, learning from those around us (who have at least to some extent managed to navigate these environments themselves) increases our chances for success (Henrich, 2015). Importantly for the current discussion, culturally learned norms can come to shape our intuitions. For example, in the West where normative fairness implies an equal division of resources (Henrich et al., 2010), the guiding intuitions at play in resource-sharing economic games under time pressure (i.e., evoking intuitions) is to divvy up the potential winnings equally between players (Rand, 2016). Analytical thinkers, on the other hand, who take the time to question this norm are those most likely to defect. In other parts of the world (e.g., India), where decisions in these same games normatively differ (i.e., much less is typically shared between players; Raihani, Mace, & Lamba, 2013), the intuitive response elicited under time pressure is to defect (Nishi, 17 Christakis, & Rand, 2017). What this all suggests for the use of analytical thinking in everyday reasoning is that analytical thinking is not only employed to inhibit evolved intuitions but also internalized cultural norms. In the case of religion, persistent cultural exposure to others\u00E2\u0080\u0099 religious beliefs and practices (i.e., perceiving normative religiosity) is known to contribute immensely to religious adherence in both degree (i.e., the strength of one\u00E2\u0080\u0099s beliefs; Willard & Cingl, 2017) and content (e.g., why most Americans do not believe in Zeus; Gervais et al., 2011). And thus, the relationship between analytical thinking and decreased religiosity may not be owed to an inhibition of evolved-intuitions that give rise to religious-like cognitions, but rather a tendency to question prevailing religious cultural norms. This counter-normative rationality model would thus predict that the observed negative correlation between analytical thinking and religious belief should be largest in normatively-religious places. In support of this, Gervais et al. (2017) recently demonstrated that the correlation is moderated by country-level religiosity such that in less normatively religious nations the correlation is diminished and even, surprisingly, reverses in some highly secularized places (e.g., the United Kingdom). If supported further, this model also calls into question the extent to which the observed negative relationship between analytical thinking and religious belief is owed solely to the inhibition of evolved intuitions that give rise to religious-like cognitions. Testing the three models The growing record of a robust negative correlation between analytical thinking and religious belief is not enough to account for the psychological origins of this relationship. In this study, first the replicability of the basic correlation between belief in god and analytical thinking (a test of the dual process model of belief) is assessed. Next, I examined how well this model 18 extends to other types of religious beliefs (namely, belief in karma, belief in witchcraft, and the belief that religion is necessary for morality). Then, the expressive rationality model of belief is tested by examining the interaction between analytical thinking and political orientation in predicting belief in god and the belief that religion is necessary for morality. As both belief in god and the belief that religion is necessary for morality (i.e., you need to believe in god to be a good person) are often considered a hallmark of North American conservatism (e.g., Haidt, 2012) \u00E2\u0080\u0093 the expressive rationality model predicts that analytical thinking will promote these beliefs amongst politically conservative individuals. Lastly, I tested predictions of the counter-normative rationality model by examining whether analytical thinking positively predicted belief in counter-normative types of beliefs (i.e., beliefs in karma and beliefs in witchcraft). More specifically, this analysis tests the prediction that if the counter-normative rationality model is supported then (1) in India (where karmic belief is normative) analytical thinking should be negatively related to karma beliefs. And (2), that in the North American samples (where karmic belief is \u00E2\u0080\u0093 at least less \u00E2\u0080\u0093 culturally supported in religious traditions), analytical thinking should be positively related to karma beliefs. Witchcraft beliefs, however, are presumed to be less normative in both samples and thus should be positively related to analytical thinking in both samples. Methods Sample A total 5284 participants from seven datasets were combined to test the three focal models (see Table 2.1 for sample details). Participants were recruited from the University of British Columbia Psychology Department\u00E2\u0080\u0099s Human Subject Pool; a national sample of the Canadian population, two samples of Indians (Amazon\u00E2\u0080\u0099s Mechanical Turk and a broader sample recruited 19 by an online market research company), and a broad sample of Americans (recruited on Amazon\u00E2\u0080\u0099s Mechanical Turk). In Table 2.2, the presence and absence of the focal measures in each data set is presented across the data sets. Summary statistics of all focal measures are presented in Table 2.3 and correlations in Table 2.4. Measures Measures of belief Across the combined datasets, indicators of belief in god were measured on different response scales (5-point and 7-point scales). To be able to compare belief across datasets in analyses, indicators of belief in god were first standardized by the mean and standard deviations within each dataset. In four of the included datasets, a measure of the believed necessity of religion for being moral (i.e., a \u00E2\u0080\u0098good\u00E2\u0080\u0099 person) was included. This 6-item scale asks participants to rate the extent to which they agree with items such as, \u00E2\u0080\u009CGenerally speaking, people need religion to be morally good\u00E2\u0080\u009D, and \u00E2\u0080\u009CAn individual who does not believe in god cannot lead a moral life\u00E2\u0080\u009D (Cronbach\u00E2\u0080\u0099s Alpha = .96). This measure was included to test whether the dual process model of belief extends to expressions of religious beliefs (rather than just beliefs in specific supernatural agents). Moreover, this globally wide-spread set of beliefs is known to be particularly divisive between North American conservatives and liberals (Pew Research Center, 2014), making it a suitable measure for testing predictions of the expressive rationality model of belief. Belief in karma was assessed using either the 16-item or 4-item version of a karmic belief scale (White, 2017; White, Norenzayan, & Schaller, 2018). This scale assesses belief in karma with items such as, \u00E2\u0080\u009CKarma is a force that influences the events that happen in my life\u00E2\u0080\u009D, and \u00E2\u0080\u009CWhen people experience good fortune, they have brought it upon themselves by behaviour in a past life\u00E2\u0080\u009D. 20 Witchcraft beliefs were assessed using a 7-item scale (e.g., \u00E2\u0080\u009CPeople can harm others with supernatural power, e.g., by cursing or casting spells on people\u00E2\u0080\u009D, and \u00E2\u0080\u009CIf other people have had bad thoughts towards you, it can make you sick\u00E2\u0080\u009D). The belief in karma and belief in witchcraft scales are employed here to test competing predictions made by the dual process model of belief (i.e., that all types of supernatural beliefs should be negatively related to analytical thinking) and the counter-normative rationality model (i.e., that analytical thinking should be positively related to non-normative belief sets). Measures of analytical thinking The Cognitive Reflection Test (Frederick, 2005) is a three-item measure of one\u00E2\u0080\u0099s willingness and general tendencies for inhibiting intuitive responses and thinking more deeply (i.e., reflecting). This test asks participants three questions to which there is an intuitively compelling (but wrong) answer (e.g., \u00E2\u0080\u009CIf a bat and a ball cost $1.10, and the bat costs 1.00$ more than the ball, how much does the ball cost?\u00E2\u0080\u009D). Individuals who tend not to reflect often give the answer \u00E2\u0080\u009C10 cents\u00E2\u0080\u009D (the modal response). Individuals who do make the effort to reflect are more likely to arrive at the correct answer, \u00E2\u0080\u009C5 cents\u00E2\u0080\u009D. Participant\u00E2\u0080\u0099s responses are coded for correctness and summed. This test has been broadly used in testing the contributions of analytical thinking to religiosity (e.g., Pennycook, Ross, Koehler, & Fugelsang, 2016). Across the datasets employed here, this measure was fairly reliable for being a 3-item test (a = .75). The Faith in Intuition subscale of the Rational Experiential Inventory (Pacini & Epstein, 1999) is a 20-item self-report measure that asks participants to indicate their agreement with a series of statements reflecting an explicit preference for not overthinking and trust in one\u00E2\u0080\u0099s intuitions (e.g., \u00E2\u0080\u009CI like to rely on my intuitive impressions\u00E2\u0080\u009D, and \u00E2\u0080\u009CI believe in trusting my hunches\u00E2\u0080\u009D, I hardly ever go wrong when I listen to my deepest gut feelings to find an answer\u00E2\u0080\u009D). This scale has 21 also been employed in the examination of the contributions of analytical/intuitive thinking styles and religiosity (e.g., Pennycook et al., 2016). Across the datasets employed here, this measure was very reliable (a = .96). Demographics Reported participant demographics included age, sex (1 = male), and political orientation (1 = very liberal, 7 = very conservative). 22 Table 2.1 Sample characteristics of the data sets Notes: 1Political orientation was assessed on a 1 = very liberal to 7 = very conservative response scale Source of Data Code Sample N Male (%) Age [M (SD)] Political Conservatism1 [M (SD)] White, Norenzayan, & Schaller (2018) \u00E2\u0080\u0093 Dataset 1 WN1 Canada - National 1000 49 46.7 (15.2) 3.61 (1.71) Baimel, Li, & Norenzayan (unpublished) BL1 Undergraduate Students 816 21.1 2.4 (3.13) 3.13 (1.23) White, Norenzayan, & Schaller (2018) \u00E2\u0080\u0093 Dataset 1 WN1 India - National 1000 49.1 38.6 (13.5) 3.57 (1.81) White, Norenzayan, & Schaller (2018) \u00E2\u0080\u0093 Dataset 2 WN2 India - MTurk 319 70.3 32.7 (9.53) 3.90 (1.64) Baimel, Norenzayan & Sarkissian (unpublished) \u00E2\u0080\u0093 Dataset 1 BN1 USA - MTurk 428 50.5 36.5 (12.2) 3.32 (1.66) Baimel, Norenzayan & Sarkissian (unpublished) \u00E2\u0080\u0093 Dataset 2 BN2 USA - MTurk 291 53.3 34.7 (11.8) 3.50 (1.75) Baimel, Li & Norenzayan (unpublished) BL2 USA - MTurk 792 38.3 38.3 (13.3) 3.51 (1.85) White, Norenzayan & Schaller (2018) \u00E2\u0080\u0093 Dataset 2 WN2 USA - MTurk 417 38.5 36.7 (12.2) 3.53 (1.76) White, Norenzayan, & Schaller (2018) \u00E2\u0080\u0093 Dataset 3 WN3 USA - MTurk 215 39 35.6 (1.9) 3.44 (1.80) Total 5284 43.4 36.3 (14.5) 3.49 (1.70) 23 Table 2.2 Included measures in each dataset Code Source of Data Sample Belief in God CRT REI KARMA WITCH MORAL WN1 White, Norenzayan, & Schaller (2018)\u00E2\u0080\u0093 Dataset 1 Canada 1 \u00E2\u0080\u0093 7 Scale (1 item) \u00E2\u009C\u0093 \u00E2\u009C\u0093 BL1 Baimel, Li, & Norenzayan (unpublished) Canada \u00E2\u0080\u0093 UBC 1 \u00E2\u0080\u0093 7 Scale (1 item) \u00E2\u009C\u0093 \u00E2\u009C\u0093 WN1 White, Norenzayan, & Schaller (2018)\u00E2\u0080\u0093 Dataset 1 India 1 \u00E2\u0080\u0093 7 Scale (1 item) \u00E2\u009C\u0093 \u00E2\u009C\u0093 WN2 White, Norenzayan, & Schaller (2018)\u00E2\u0080\u0093 Dataset 2 India 1 \u00E2\u0080\u0093 5 Scale (3 items) \u00E2\u009C\u0093 \u00E2\u009C\u0093 \u00E2\u009C\u0093 BN2 Baimel, Norenzayan & Sarkissian (unpublished) \u00E2\u0080\u0093 Dataset 1 USA 1 \u00E2\u0080\u0093 7 Scale (1 item) \u00E2\u009C\u0093 \u00E2\u009C\u0093 \u00E2\u009C\u0093 BN2 Baimel, Norenzayan & Sarkissian (unpublished) \u00E2\u0080\u0093 Dataset 2 USA 1 \u00E2\u0080\u0093 7 Scale (1 item) \u00E2\u009C\u0093 \u00E2\u009C\u0093 \u00E2\u009C\u0093 BL1 Baimel, Li & Norenzayan (unpublished) USA 1 \u00E2\u0080\u0093 7 Scale (1 item) \u00E2\u009C\u0093 \u00E2\u009C\u0093 WN2 White, Norenzayan, & Schaller (2018)\u00E2\u0080\u0093 Dataset 2 USA 1 \u00E2\u0080\u0093 5 Scale (3 items) \u00E2\u009C\u0093 \u00E2\u009C\u0093 \u00E2\u009C\u0093 WN3 White, Norenzayan, & Schaller (2018)\u00E2\u0080\u0093 Dataset 3 USA 1 \u00E2\u0080\u0093 5 Scale (3 items) \u00E2\u009C\u0093 \u00E2\u009C\u0093 \u00E2\u009C\u0093 \u00E2\u009C\u0093 Notes: CRT = Cognitive Reflections Test; REI = Faith in Intuition subscale from the Rational Experiential Inventory; KARMA = Belief in Karma Scale; WITCH = Belief in Witchcraft; MORAL = Belief that religion is necessary for morality 24 Table 2.3 Summary statistics of focal measures by sample (means and standard deviations) Source Sample Belief CRT INT KARMA WITCH MORAL WN1 Canada 4.48 (2.27) --- 3.57 (0.62) 2.71 (0.82) --- --- BL1 Students 3.41 (2.28) 1.53 (1.21) --- --- --- 1.91 (0.98) WN1 India 6.10 (1.51) --- 3.63 (0.54) 3.69 (0.72) --- --- WN2 India 3.97 (0.85) --- 4.79 (0.70) 3.63 (0.83) 3.22 (0.79) --- BN1 USA 3.46 (2.38) 1.51 (1.25) 4.31 (1.01) --- --- 2.17 (1.28) BN2 USA 3.38 (2.45) 1.53 (1.19) 4.33 (1.03) --- --- 2.19 (1.43) BL2 USA 4.00 (2.53) 1.30 (1.23) --- --- --- 2.29 (1.51) WN2 USA 3.58 (1.33) --- 5.16 (1.03) 2.73 (0.99) 2.32 (0.98) --- WN3 USA 3.43 (1.43) 1.47 (1.20) 4.94 (1.17) 2.71 (1.13) 2.12 (0.90) --- Notes: CRT = Cognitive Reflections Test; REI = Faith in Intuition subscale from the Rational Experiential Inventory. KARMA = Belief in Karma scale; WITCH = Belief in witchcraft scale; MORAL = Belief that religion is necessary for morality. 25 Table 2.4 Correlations [95% confidence intervals] by sample of focal variables Country Belief in God CRT Intuition Karma Witchcraft Morality-Religion CRT Canada --- India --- Students -.07 [-.14, .00] USA -.19 [-.24, -.15] Faith in Intuition Canada .07 [.00, .13] --- India .18 [.13, .23] --- Students --- --- USA .18 [.13, .23] -.22 [-.28, -.16] Belief in Karma Canada .30 [.24, .35] --- .27 [.21, .33] India .39 [.34, .43] --- .28 [.23, .33] Students --- --- --- USA .33 [.25, .39] -.27 [-.39, -.14] .28 [.21, .35] Belief in Witchcraft Canada --- --- --- --- India .05 [-.07, .16] --- .11 [-.01, .22] .54 [.46, .62] Students --- --- --- --- USA .33 [.26, .40] -.31 [-.43, -.19] .21 [.13, .28] .54 [.48, .59] Religion is necessary for morality Canada --- --- --- --- --- India --- --- --- --- --- Students .44 [.39, .50] -.11 [-.18, -.04] --- --- --- USA .51 [.47, .55] -.17 [-.22, -.12] .05 [-.02, .13] --- --- Conservatism Canada .24 [.18, .30] --- .01 [-.06, .07] .03 [-.04, .09] --- --- India .13 [.07, .18] --- -.06 [-.11, -.00] .15 [.10, .20] .34 [.23, .44] --- Students .20 [.13, .27] .12 [.05, .19] --- --- --- .28 [.21, .34] USA .41 [.37, .44] -.06 [-.11, -.01] .04 [-.01, .10] .01 [-.07, .08] .06 [-.02, .14] .38 [.33, .42] 26 Results Dual process model of belief First, I tested the core prediction of the dual process model of belief by examining the correlations between religious belief and two measures of thinking style. In aggregate, belief in god was negatively correlated with higher scores on the Cognitive Reflections Test: r (2516) = -.15 [-.19, -.12]. However, this correlation was notably weaker in the undergraduate student sample (r (798) = -.07 [-.14, .00]) than among American M-Turkers (r (1716) = -.19 [-.24, -.15]). Similarly, belief in god was positively correlated in aggregate with self-reported faith in intuition: r (3596) = .14 [.11, .17]. And, this correlation was also found to similarly vary between sites (USA: r (1344) = .18 [.13, .23]; India: r (1258) = .19 [.13, .24]; Canada: r (990) = .07 [.00, .13]). To examine these sample differences in contributions of analytical thinking to belief in god controlling for demographics (age, sex, and political conservativism), a series of mixed-effect linear regressions were conducted for the CRT and the faith in intuition scales. In these analyses, a random intercept for source of the data was included to model potential variability in belief across the examined datasets. Model summaries for the CRT are presented in Table 2.5. In Model 1, the fixed effect of CRT was included and provides an estimate for the aggregate effect of CRT across samples. In Model 2, the moderating role of sample was included (Reference group = UBC students; 1 = USA), and reiterates the observations from the correlations \u00E2\u0080\u0093 namely that the negative association between analytical thinking and belief was stronger in the American sample (b = -.16 [-.28, -.03]) than in the student sample (b = -.06 [-.11, .00]). And, in Model 3, fixed effects for participant demographics (age, sex, and political conservativism) were added. The addition of these controls did not affect the estimated contributions of CRT to belief in either sample. Importantly for the current discussion, the effect of CRT was robust to controlling for 27 political conservativism. These results replicate previous demonstrations of the effects of CRT on belief in god. Indeed, the magnitude of this effect is within the confidence bounds of the meta-analytic estimate (i.e., r = -.18 [-.21, -.16]) reported by Pennycook et al., (2016). The observed reduction of this effect amongst undergraduate students at UBC, however, indicates that the magnitude of the contributions of analytical thinking to belief may be variable across populations (these correlations were more in line with the overall estimate of the relationship reported by Gervais et al., 2017). As currently posited, the dual-process model of belief is not suitably specified to account for this variability across samples. 28 Table 2.5 Predicting belief in god from cognitive reflections test score Next, I examined the sample differences in the relationship between faith in intuition and belief in god while controlling for demographics as above. Note that here, the dual process model of belief predicts a positive cross-sample relationship between faith in intuition and belief. The contributions of faith in intuition to belief were assessed in a series of mixed effect linear models (see Table 2.6). In Model 1, the fixed effect of intuition was added, providing an estimate of the aggregate contribution of intuition to belief. This relationship (b = .15 [.12, .18]) is of a similar magnitude to that which was observed for the CRT and belief in god (but as predicted, reversed in-sign), and the meta-analytic effect reported by Pennycook et al., (2016). In Model 2, the moderating contributions of country (reference group = Canada) were assessed. As above, the relationship was stronger in the Indian (b = .19 [.05, .33]) and American samples (b = .18 [.05, .32]) than the Canadian sample (b = .06 [.00, .12]). In Model 3, fixed effects for age, sex (1= Model 1 Model 2 Model 3 Predictors b .95CI b .95CI b .95CI (Intercept) 0.18 0.12 \u00E2\u0080\u0093 0.25 0.08 -0.03 \u00E2\u0080\u0093 0.19 0.16 0.05 \u00E2\u0080\u0093 0.26 CRT -0.13 -0.16 \u00E2\u0080\u0093 -0.09 -0.06 -0.11 \u00E2\u0080\u0093 0.00 -0.07 -0.12 \u00E2\u0080\u0093 -0.02 Sample difference in belief (0 = UBC; 1 = USA) 0.14 0.01 \u00E2\u0080\u0093 0.28 0.14 0.01 \u00E2\u0080\u0093 0.26 CRT * Sample - USA -0.10 -0.17 \u00E2\u0080\u0093 -0.03 -0.06 -0.12 \u00E2\u0080\u0093 0.01 Age (years, std.) 0.08 0.04 \u00E2\u0080\u0093 0.12 Sex (1 = Male) -0.25 -0.33 \u00E2\u0080\u0093 -0.17 Conservativism (std.) 0.34 0.30 \u00E2\u0080\u0093 0.38 Observations 2463 2463 2463 AIC 6949.32 6954.08 6612.30 29 male), and political conservativism were included. The relationship of intuition to belief in god was robust to the addition of these controls across the three samples. Taken together with the CRT analyses, these results provide twice-fold support for the dual process model of belief across two complementary measures of thinking style. However, the magnitude of this relationship was variable between samples for both measures. And, the dual process model has little to say about the sources of this observed sample variability. Table 2.6 Predicting belief in god from faith in intuition Model 1 Model 2 Model 3 Predictors b .95CI b .95CI b .95CI (Intercept) 0.00 -0.03 \u00E2\u0080\u0093 0.03 -0.00 -0.06 \u00E2\u0080\u0093 0.06 0.12 0.05 \u00E2\u0080\u0093 0.19 Faith in Intuition (std.) 0.15 0.12 \u00E2\u0080\u0093 0.18 0.06 -0.00 \u00E2\u0080\u0093 0.12 0.05 -0.01 \u00E2\u0080\u0093 0.11 Sample (India) (Ref = Canada) -0.00 -0.08 \u00E2\u0080\u0093 0.08 0.01 -0.07 \u00E2\u0080\u0093 0.09 Sample (USA) (Ref = Canada) 0.00 -0.08 \u00E2\u0080\u0093 0.09 -0.01 -0.08 \u00E2\u0080\u0093 0.07 Intuition * India 0.13 0.05 \u00E2\u0080\u0093 0.21 0.15 0.07 \u00E2\u0080\u0093 0.23 Intuition * USA 0.12 0.04 \u00E2\u0080\u0093 0.20 0.10 0.03 \u00E2\u0080\u0093 0.18 Age (years, std.) 0.13 0.10 \u00E2\u0080\u0093 0.16 Sex (1 = Male) -0.24 -0.31 \u00E2\u0080\u0093 -0.18 Conservativism (std.) 0.26 0.23 \u00E2\u0080\u0093 0.29 Observations 3513 3513 3513 AIC 9910.390 9925.687 9564.089 30 Next, I tested whether the dual process model of belief extended to other types of supernatural beliefs. Interestingly, both the CRT and faith in intuition measures were (1) more strongly correlated with belief in karma and belief in witchcraft across samples than either measure was with belief in god and (2) all in the directions predicted by the dual process model of belief (i.e., negatively correlated with the CRT and positively correlated with faith in intuition; see Table 2.4). These results provide support for the dual process model of belief, extended to belief in karma and witchcraft in diverse samples. Expressive rationality model of belief The results thus far demonstrate that the relationship between analytical thinking (or faith in intuition) and belief in god is robust to controlling for political orientation. But, the expressive rationality model hypothesizes this relationship is moderated by political orientation such that analytical thinking is positively related to belief amongst conservatives. To test this for both the CRT and the faith in intuition measures, two additional models were assessed that included the fixed effect interaction of political orientation (1 = very liberal; 7 = very conservative) with the sample by analytical thinking measure interaction (Table 2.7 \u00E2\u0080\u0093 CRT; Table 2.8\u00E2\u0080\u0093 faith in intuition scale). Put differently, these additional models examined the interaction between thinking style and political orientation within each sample. The relationship between CRT and belief in god was moderated by political orientation in the UBC student sample; but not in the American sample (see Figure 2.1). Amongst more liberal students, CRT was (as would be predicted both by the dual process model of belief and the expressive rationality model) negatively related to belief in god (b = -.13 [-.20, -.05]). This relationship, however, was reduced at mean levels of political conservativism (b = -.06 [-.11, -.01]); and amongst more conservative students, CRT was unrelated to belief in god (b = .01 [-.06, 31 .09]). This latter result demonstrating only weak support for the expressive rationality model; but demonstrating that the dual process model does not hold at all levels of political conservatism. For Americans, no interaction was observed (see Figure 2.1). Table 2.7 Testing the expressive rationality model \u00E2\u0080\u0093 cognitive reflections test Predictors b .95CI (Intercept) 0.13 0.02 \u00E2\u0080\u0093 0.23 CRT -0.06 -0.11 \u00E2\u0080\u0093 -0.01 Sample differences in belief (0 = UBC; 1 = USA) 0.12 0.02 \u00E2\u0080\u0093 0.22 Conservativism 0.17 0.04 \u00E2\u0080\u0093 0.29 Age (years, std.) 0.08 0.04 \u00E2\u0080\u0093 0.11 Sex (1 = Male) -0.25 -0.33 \u00E2\u0080\u0093 -0.17 CRT * USA 0.07 0.02 \u00E2\u0080\u0093 0.12 USA * Conservativism -0.07 -0.13 \u00E2\u0080\u0093 -0.00 CRT * Conservativism 0.24 0.12 \u00E2\u0080\u0093 0.37 CRT * Conservativism * USA -0.05 -0.11 \u00E2\u0080\u0093 0.01 Observations 2463 AIC 6609.849 32 Figure 2.1 The moderating effect of political orientation on the relationship between CRT and belief in god Notes: Shaded regions are 95% confidence intervals. Predictions were made from the model presented in Table 2.7. Turning to the faith in intuition scale as the predictor, I tested the expressive rationality model as above in three samples (Canadians, Indians and Americans; see Table 2.8). The predicted reversal of the relationship between intuition and belief in god amongst conservatives was not supported by the data from any of the three samples. However, in all samples, the relationship between intuition and belief in god was reduced at increasing levels of political conservatism (see Figure 2.2). Simple slope analyses indicate that amongst more liberal Canadians, increased faith in intuition was positively related to belief in god (b = .09 [.01, .17]) \u00E2\u0080\u0093 as would be predicted by both the expressive rationality model and the dual process model of belief. This relationship was reduced at mean levels of political conservativism (b = .05 [-.00, .11]). And, reduced further amongst conservative Canadians (b = .01 [-.06, .09]). This pattern of results (i.e., decreasing strength of the association with higher levels of political conservatism) was found, but more UBC USA0 1 2 3 0 1 2 3\u00E2\u0088\u00920.40.00.4Cognitive Reflections Test (# of Correct Responses)Predicted belief in god (std.)POLITICS+ LiberalMean+ Conservative 33 weakly, in both the Indian (+ liberal = .24 [.18, .31]; Mean conservatism = .18 [.13, .23]; + conservative = .12 [.03, .20]) and American samples (+ liberal = .19 [.13, .26]; Mean conservatism = .14 [.10, .20]; + conservative = .10 [.03, .17]). As with the CRT results, this only provides weak evidence for the expressive rationality model as the predicted reversal was not supported by the data in any sample. However, these results demonstrate that the dual process models\u00E2\u0080\u0099 predictions are less supported amongst conservatives. And thus, these results suggest that the dual process model may not apply in all contexts or that it is constrained by other factors. 34 Table 2.8 Testing the expressive rationality model \u00E2\u0080\u0093 faith in intuition scale Belief in god (standardized) b .95CI (Intercept) 0.12 0.05 \u00E2\u0080\u0093 0.18 Faith in Intuition (std.) 0.05 -0.00 \u00E2\u0080\u0093 0.11 Sample differences in belief \u00E2\u0080\u0093 India (Ref = Canada) 0.24 0.18 \u00E2\u0080\u0093 0.30 Sample differences in belief \u00E2\u0080\u0093 USA (Ref = Canada) 0.00 -0.07 \u00E2\u0080\u0093 0.08 Conservativism (std.) -0.01 -0.08 \u00E2\u0080\u0093 0.07 Age (years, std.) 0.13 0.09 \u00E2\u0080\u0093 0.16 Sex (1 = Male) -0.24 -0.30 \u00E2\u0080\u0093 -0.18 Intuition * Sample (India) -0.04 -0.09 \u00E2\u0080\u0093 0.02 Intuition * Sample (USA) 0.13 0.05 \u00E2\u0080\u0093 0.20 Intuition * Conservativism 0.09 0.02 \u00E2\u0080\u0093 0.17 Sample (India) * Conservativism -0.08 -0.16 \u00E2\u0080\u0093 -0.01 Sample (USA) * Conservativism 0.15 0.07 \u00E2\u0080\u0093 0.23 Intuition * Sample (India) *Conservativism -0.03 -0.10 \u00E2\u0080\u0093 0.05 Intuition * Sample (USA) *Conservativism -0.01 -0.08 \u00E2\u0080\u0093 0.06 Observations 3513 AIC 9546.883 35 Figure 2.2 The moderating effect of political orientation on the relationship between faith in intuition and belief in god Notes: Shaded regions are 95% confidence intervals. Predictions were made from the mode presented in Table 2.8. Next, I examined how well the expressive rationality model of belief accounted for beliefs that religion is necessary for morality by examining the interaction of political orientation and analytical thinking in predicting these beliefs. The belief that religion is necessary for morality is known to be divisive between liberals and conservatives in the American cultural context (Haidt, 2012). Moreover, although the overall proportion of individuals reporting that religion is necessary to be a \u00E2\u0080\u0098good person\u00E2\u0080\u0099 is decreasing in the United States, Republican endorsement of this belief is at new heights, while Liberal rejection of this belief at all-time lows (Pew Research Center, 2017). In a series of models, however, I found no evidence of an interaction between political orientation and CRT in predicting the belief that religion is necessary for morality in either the student or American samples (see Table S2.1). Canada India USA\u00E2\u0088\u00926 \u00E2\u0088\u00924 \u00E2\u0088\u00922 0 2 4 \u00E2\u0088\u00926 \u00E2\u0088\u00924 \u00E2\u0088\u00922 0 2 4 \u00E2\u0088\u00926 \u00E2\u0088\u00924 \u00E2\u0088\u00922 0 2 4\u00E2\u0088\u00922\u00E2\u0088\u0092101Faith in Intuition (std.)Predicted belief in god (std.)POLITICS+ LiberalMean+ Conservative 36 Counter-normative rationality model of belief Earlier, the prediction of the counter-normative rationality model that analytical thinking style would be positively associated with endorsement of non-normative supernatural beliefs was not supported. Indeed, analytical thinking was negatively associated with belief in karma in all samples, regardless of how normative it was in the particular sample. This was also the case for belief in witchcraft (assumed to be non-normative in all samples; and see low mean levels of support across samples in Table 2.3). And thus, these predictions of the counter-normative rationality model of belief were unsupported by the data. Discussion In support of the dual process model of belief, the results presented here replicate previously reported results that analytical thinking is negatively correlated, although modestly, with belief in god. Moreover, it was demonstrated that tendencies for analytical thinking (as measured by the CRT) and placing faith in one\u00E2\u0080\u0099s intuition were similarly related to belief. Taken together, this provides further evidence that the effect of thinking style on belief is robust to two different cognitive measures. Indeed, while the CRT does reliably measure tendencies for overriding one\u00E2\u0080\u0099s intuitions, it has somewhat surprisingly been demonstrated to be a relatively poor indicator of individual differences in reliance on intuitions (Pennycook, Cheyne, Koehler, & Fugelsang, 2016). And thus, the growing body of work that employs the CRT in examining the relationship between analytical thinking that and belief in god is better understood as documenting the negative effect of analytical thinking on belief, and not necessarily the positive effect of intuition. Here, the consistently observed positive effect of self-reported faith in intuition speaks to the inverse relationship \u00E2\u0080\u0093 that a reliance on one\u00E2\u0080\u0099s intuitions is also related to belief. Moreover, 37 these associations extended to other types of beliefs (i.e., karma and witchcraft) \u00E2\u0080\u0093 in further support for the dual process model of belief. The dual process model of belief, however, has little to say that can account for the observed variability between samples in the magnitude of the effect of analytical thinking and intuition on belief in god. Indeed, the basic relationship was markedly reduced in undergraduate students, and in a Canadian national sample as compared to amongst Americans and Indians. In their assessment of the relationship between CRT and belief in god cross-culturally, Gervais et al. (2017) provide some evidence that the size of the effect is indeed moderated by population-level religiosity (with the effect being reduced in less religious countries). Here, the Indian and American samples were both much more religious than the Canadian and student samples \u00E2\u0080\u0093 and thus supports this interpretation. However, this still cannot account for when and in what ways analytical thinking and intuition were found to be related to belief in the student and national Canadian samples. In all samples, the negative effect of analytical thinking and the positive association of faith in intuition on belief in god was greatest for more politically-liberal leaning participants. Moreover, this is particularly striking given that conservatives are reported to rely more heavily on intuitive thinking styles, and are generally more religious (Deppe et al., 2015; Haidt, 2012; Nail, McGregor, Drinkwater, Steele, & Thompson, 2009; Pew Research Center, 2017). If the dual process model is correct then, this would suggest that if there was going to be a difference in effects amongst conservative and liberal individuals \u00E2\u0080\u0093 it would follow more clearly that the effects be larger (not smaller or reversed) amongst conservatives. The expressive rationality model that analytical thinking is employed not to override intuitions but rather to engage in identity-confirming cognitive reasoning (Kahan & Stanovich, 38 2016) can, however, account for some aspects of the current data. As an important caveat, this model\u00E2\u0080\u0099s predicted reversal in the direction of the association was not supported by the data. However, in all samples the association was reduced (and sometimes to zero) amongst conservatives. An all-together different model, the counter-normative rationality model was unsupported by the data. In this model, analytical thinking is deployed in questioning prevailing cultural norms \u00E2\u0080\u0093 including normative religiosity and related beliefs. As the datasets employed here were constrained in their reach of population-level variability in religiosity, the hypothesis was tested by examining the relationships between thinking style and non-normative belief sets (belief in karma and belief in witchcraft). Results indicate that analytical thinking style was negatively related to counter-normative beliefs across samples (in support of the dual process model of belief extended to other types of religious beliefs). Taken together, it is clear that none of these models can account for all of the observed data. And importantly, none of them (as they are currently posited) seem fully equipped to deal with how intuition and/or analytical thinking may or may not be implicated in \u00E2\u0080\u0098religion\u00E2\u0080\u0099 in a variety of different cultural contexts. And although it is the counter-normative rationality model that is the least consistently supported \u00E2\u0080\u0093 it is the dual process model of belief that requires the most re-calibration. The dual process model of belief as it is currently formulated cannot account for variation in when and in what ways intuition comes to support and analytical thinking comes to erode religious beliefs. And thus, observations \u00E2\u0080\u0093 like those made here - that the relationship between thinking styles and belief is moderated by other factors such as political orientation can help elucidate the boundary conditions of this model\u00E2\u0080\u0099s account of religious belief. Limitations and future directions 39 A clear way forward in unpacking the contributions of intuitive thinking styles to belief is to continue broadening the scope of these types of investigations across cultures and more broadly across contexts within cultures. The recent cross-cultural effort by (Gervais et al., 2017) is an important step in the right direction. However, this work looked only at the effect between belief in god and analytical thinking as measured by the cognitive reflections test. Although they found a small negative effect of analytical thinking on belief, the use of the CRT leaves open the equally important question as to whether and in what ways intuition comes to support belief. The observation that analytical thinking erodes belief is relevant to the discussion \u00E2\u0080\u0093 but it remains a distinct line of research from how beliefs are and come to be intuitively supported. Moreover, the entire literature on the contributions of analytical thinking (including the work presented here) is deeply limited by its inherent assumptions that reductions in conventional beliefs (i.e., in god) equates to reduction in all supernatural beliefs. For example, recent years have seen a striking rise in belief in astrology, horoscopes, and the Tarot amongst North American youth (Beck, 2018; Pew Research Center, 2009). And thus, while the literature might suggest that analytical thinking erodes supernatural thinking, broadly speaking, most of the existing literature has focused on belief in god, and thus might benefit from more careful hesitation in extending too far beyond this specific belief. Future work should then, broaden the scope of the content of examined supernatural beliefs. 40 Chapter 3 Mentalizing and Belief Introduction The search for the psychological foundations of religion has long been focused on the contributions of mentalizing (i.e., how we think about the minds of others) to supernatural agent beliefs (e.g., Boyer, 2001; Guthrie et al., 1980). And indeed, evidence has accumulated in support of the hypothesis that belief in supernatural agents (like god) and how individuals come to interact with supernatural minds requires at some basic level the capacity to reason about the mental states of others and that individual differences in mentalizing account for some of the variability in beliefs (Willard & Norenzayan, 2013, 2017; Caldwell-Harris, Murphy, Velazquez, & McNamara, 2011; Gray, Jenkins, Heberlein, & Wegner, 2011; Kapogiannis et al., 2009; Kelemen, 2004; Lindeman, Svedholm-H\u00C3\u00A4kkinen, & Lipsanen, 2015; Norenzayan, Gervais, & Trzesniewski, 2012; Schjoedt, St\u00C3\u00B8dkilde-J\u00C3\u00B8rgensen, Geertz, & Roepstorff, 2009). However, the literature has provided little with regard to direct tests of the potential mechanisms by which individual differences in capacities for thinking about human minds come to be implicated in belief in supernatural agents. Moreover, recent studies (e.g., Maij et al., 2017) have called into question the importance of mentalizing\u00E2\u0080\u0099s contributions to belief in god once controlling for cultural factors (i.e., religious upbringing). In three studies, the work presented here (1) tests the replicability of the relationship between mentalizing and belief in diverse samples (students at a Canadian university, Americans, Indians, American Christians, and American Hindus); (2) tests three explanatory accounts of the mechanism by which mentalizing comes to support belief; and (3) further specifies and tests the boundary conditions of this relationship. That is, this work examines how and when mentalizing comes to be implicated in which kinds of supernatural beliefs. 41 Extending the cross-cultural reach of the cognitive sciences of religions The extant research on the contributions of mentalizing to belief has largely sampled from Christian populations. The work presented here provides a first test of the contributions of individual differences in mentalizing to belief in a sample of Indians (Study 1; mostly Hindu) and American Hindus (Study 2A); religious contexts that are typically understudied in the cognitive sciences of religion. Unlike predominant forms of North American Christianity, Hinduism acknowledges the existence of an extensive pantheon of supernatural deities. This work stands as the first direct contrast between Christian monotheists and Hindu polytheists in an assessment of how and in what ways mentalizing is implicated in belief. Notwithstanding their distinct cultural evolutionary trajectories on different continents in different millennia, Hinduism and Christianity share devotion to supernatural agents as a central tenet of their respective traditions. And thus, tests of how and when mentalizing becomes implicated in this devotion in supernatural agents in the context of Hinduism is a particularly interesting test case of the generalizability of this core finding in the cognitive science of religion. How does mentalizing come to support belief in god? How is it that individual differences in mentalizing (i.e., variability in the willingness or tendencies with which people thinks about other human minds) come to be related to belief in supernatural agents? In one account \u00E2\u0080\u0093 referred to here as the broad anthropomorphism hypothesis \u00E2\u0080\u0093 individuals who are more willing or prone to reasoning about human mental states are more likely to broadly anthropomorphize (i.e., attribute human mental states to) the world writ-large. This view holds that belief in supernatural agents stem, in part, from the persistent experience of \u00E2\u0080\u0098seeing\u00E2\u0080\u0099 human-like mental states as the explanatory cause of otherwise unexplained events in one\u00E2\u0080\u0099s general environment (e.g., Guthrie et al., 1980). The persistent experience of perceiving 42 human-like minds in one\u00E2\u0080\u0099s environment, overtime, coming to support explicit beliefs that there are human-like minds in one\u00E2\u0080\u0099s environment. And thus, this view predicts that mentalizing supports belief in god by virtue of supporting broad anthropomorphic tendencies. In a direct test of the broad anthropomorphism hypothesis, Willard & Norenzayan (2013) found that generalized anthropomorphic tendencies although weakly related to individual differences in mentalizing, were unrelated to belief in god (in a sample of students at a Canadian university and a broad sample of American adults). However, anthropomorphic tendencies did mediate the relationship between mentalizing and paranormal beliefs. Taken together, these results could suggest (1) that broad anthropomorphism is not a mechanism by which mentalizing comes to support belief in god or (2) that broad anthropomorphism is specifically unrelated to belief in god amongst Christians (where doctrine explicitly prohibits anthropomorphizing the natural world). And indeed, Willard & Norenzayan (2013) found that greater exposure to Christianity (as indexed by living in predominantly Christian locales) suppressed reported anthropomorphic tendencies, and thus potentially limited the observable contributions of anthropomorphism to belief in god. In the current studies, the potential mediating role of anthropomorphism on the relationship between mentalizing and belief in god is tested in both predominantly Christian and Hindu samples. Unlike in Christianity, anthropomorphism is central to Hindu beliefs that the divine exists in all forms of life as well in the depictions of many of its deities. And thus, the Hindu samples provide an opportunity to test whether anthropomorphism is largely unrelated to belief in god, or rather that the relationship is culturally suppressed amongst Christians. An alternative account \u00E2\u0080\u0093 the narrow anthropomorphism hypothesis \u00E2\u0080\u0093 holds that individual differences in mentalizing come to support belief in god not by encouraging broad anthropomorphism, but by encouraging the specific attribution of mental states to the supernatural 43 agent in question. In the case of god then, mentalizing is actively deployed to think about god\u00E2\u0080\u0099s mind in intuitively-supported human-like ways. Although this account has not been explicitly tested, evidence does suggest, for instance, that praying to god activates brain regions associated with mental state reasoning (Schjoedt et al., 2009). And thus, the narrow anthropomorphism hypothesis (tested here in five samples across three studies) holds that the observed relationship between mentalizing and belief in god will be mediated (and better accounted for) by the extent to which individuals attribute god with human-like mental states. If supported, this account suggests that it is not individual differences in mentalizing per se that supports belief, but rather whether mentalizing is actively employed in thinking about specific supernatural minds. Moreover, this account predicts that the narrow anthropomorphism of god\u00E2\u0080\u0099s mind supports belief in god by virtue of making god\u00E2\u0080\u0099s mind more intuitively grasped. As a first test of this hypothesis, in Studies 2A and 2B, the relationship between tendencies for relying on one\u00E2\u0080\u0099s intuitions and narrowly anthropomorphizing god\u00E2\u0080\u0099s mind is examined. A third account \u00E2\u0080\u0093 the anthropocentric knowledge hypothesis \u00E2\u0080\u0093 holds that the reason individuals mentalize their gods is not in support of belief, but rather because of it. That is, an otherwise acquired belief in god presents itself with the problem of deciphering the largely unknowable contents of a supernatural mind. Believers, then, are motivated to apply what they know about human minds to make sense of god\u00E2\u0080\u0099s otherwise unknowable mind. This follows from the principle of anthropocentric knowledge that individuals use human minds as a rich inferential base from which to understand non-human minds; and increasingly so the more different the mind in question seems to be (Epley et al., 2007). And thus, this model predicts that belief will mediate the relationship between mentalizing and mental state attribution to god. Although subtle, the differences between the narrow anthropomorphism and anthropocentric knowledge accounts are 44 important for explanatory accounts of how mentalizing comes to be implicated in belief. Indeed, while the narrow anthropomorphism account suggests that it is human psychology that shapes and supports belief; the anthropocentric knowledge account suggests that it is culturally acquired beliefs that evoke specific mental representations. In five samples, the current studies test the predictions of the broad anthropomorphism hypothesis, the narrow anthropomorphism hypothesis and the anthropocentric knowledge hypothesis by assessing the associated mediation models (see Figure 3.1). In addition to providing first direct tests of how mentalizing comes to be related to belief, the current studies provide a first test of the relationship between mentalizing and belief in a non-predominantly Christian sample (Indians and American Hindus). 45 Figure 3.1 The proposed mediation models When and what kinds of beliefs are supported by mentalizing? In Study 1, the three proposed models of how mentalizing comes to be related to belief are assessed in a sample of students at the University of British Columbia, Americans and Indians (recruited from M-Turk). In Studies 2A and 2B (both pre-registered), these models are tested again in targeted samples of American Christians and American Hindus. Additionally, these latter two studies included additional measures to (1) compare the contributions of mentalizing to belief with that of religious upbringing; (2) further specify the boundary conditions of the relationship by examining what kinds of beliefs regarding what kinds of supernatural agents are related to mentalizing. To assess the relative contributions of cognition (i.e., mentalizing and related cognitive biases) and culture (i.e., religious upbringing) to belief in god, Willard & Cingl (2017) tested models of both in samples from religious Slovakia and largely-secular Czech Republic. Although these countries share a lengthy cultural history (i.e., they used to be one nation), demographic data 46 indicates that they are, in the present-day, vastly different in terms of their levels of religiosity. In their models, cultural exposure to religion accounted for (1) twice the variability in belief in god than mentalizing (and associated cognitive measures) in both countries, and (2) a substantial proportion of the difference in belief between these populations. Similarly, Maij et al. (2017) provided evidence that in large samples of Dutch (predominantly secular) participants, mentalizing was a much less robust predictor of religiosity than religious upbringing and that this was also true in a broad sample of American (predominantly religious) participants. However, in the American sample mentalizing had a similar (and small) sized effect to that reported elsewhere (e.g., Willard & Norenzayan, 2013). Taken together, these results provide some mixed evidence for what has long been predicted in the cognitive sciences of religion (e.g., Gervais, Willard, Norenzayan, & Henrich, 2011) \u00E2\u0080\u0093 namely that these results are not inconsistent with at least some cognitive theories of religion that argue that cognitive tendencies are necessary but insufficient to account for religious belief. Moreover, these results also suggest that cultural exposure to religion may be a necessary cause of (at the very least) belief in the Christian God. Importantly, these results do not preclude that mentalizing and religious upbringing can have differing contributions to belief in different populations. And interestingly, neither Maij et al. (2017) nor Willard & Cingl (2017) provide details regarding whether an interaction was observed between mentalizing and religious upbringing. Given this, the non-robust effects of mentalizing on belief observed in the Dutch sample might be explained by the low levels of cultural exposure to religion in the majority of the sample. And thus, it remains an open question as to whether the contribution of mentalizing to belief is greater in more normatively religious contexts (i.e., where cultural exposure to religion is more likely). In Studies 2A and 2B, the relative contributions of mentalizing and religious 47 upbringing are compared and contrasted in a sample of American Christians and American Hindus. Additionally, a first test of the interaction between mentalizing and religious upbringing is conducted in both samples. The small effects of mentalizing (or the absence altogether) on belief reported by Maij et al. (2017) could have also stemmed from their examination of \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 as a composite indicator of items that was not specific to belief (e.g., \u00E2\u0080\u009Cto what extent do you consider yourself a religious person\u00E2\u0080\u009D) and religious practices (e.g., attendance at religious services). On this point, the cognitive sciences of religion have explicitly posited that mentalizing should contribute specifically to belief (e.g., Boyer, 2001; Gervais, 2013; Guthrie et al., 1980; Norenzayan et al., 2012; Willard & Norenzayan, 2013, 2017). And thus, in Study 1 I focused on examining the contributions of mentalizing to belief in god. In Studies 2A and 2B, I examined the relative contributions of mentalizing to belief in god, broader indicators of religiosity and spirituality, practices (prayer, service attendance, ritual performance), and importantly \u00E2\u0080\u0093 participants\u00E2\u0080\u0099 beliefs in a personal god. A personal god is one that is believed to be invested in the lives of believers, providing the necessary impetus for believers to think about god\u00E2\u0080\u0099s mental states (Norenzayan et al., 2012). And thus, I hypothesized that mentalizing should be most strongly related to belief in a personal god. To further specify the boundary conditions of the relationship between mentalizing and belief, Studies 2A and 2B also examined the contributions of mentalizing to belief in an abstract god in the American Christian and Hindu samples. Abstract god concepts are prevalent in both Christianity (e.g., God as the \u00E2\u0080\u0098greatest conceivable existent\u00E2\u0080\u0099) and Hinduism (e.g., Brahman as the transcendent reality). However, as abstract gods are devoid of human-like mental states, these types of beliefs should not be supported by mentalizing. But, does this mean that mentalizing comes to support belief in all mindful supernatural targets? 48 I hypothesized that mentalizing should not support belief in all \u00E2\u0080\u0098mindful\u00E2\u0080\u0099 supernatural targets but specifically belief in moralized supernatural targets (i.e., those attributed with knowledge of one\u00E2\u0080\u0099s thoughts and behaviours, as well as capacities for punishing bad and rewarding good behaviours). That is, mentalizing should come to support belief when mentalizing (i.e., thinking and reasoning about the contents of) a supernatural mind (or failing to do so) is believed to have consequences for the believer. To address this question, I asked the American Christian and American Hindu samples to report on their beliefs in other supernatural agents/forces (ghosts, angels/devas, demons/asuras, and karma), the extent to which they attributed mental states to these agents, as well as the extent to which they attributed these supernatural targets with moral knowledge. Specifically, this hypothesis predicts that moral knowledge attributions should moderate the relationship between mentalizing and belief. Summary of research questions In brief, the three studies presented here (1) test the replicability of the relationship between individual differences in mentalizing in diverse samples; (2) assess three theoretical models of how mentalizing comes to be implicated in belief; and (3) delineate the boundary conditions of the relationship. Studies 2A and 2B were preregistered on the Open Science Framework \u00E2\u0080\u0093 all research questions were pre-registered except for one (noted below; Hindu sample - https://osf.io/9qdra/; Christian sample - https://osf.io/u2yph/). Specifically, the work presented here asks: 1. Does the relationship between individual differences in mentalizing and belief in god vary between different religious populations? (Study 1, 2A and 2B) 2. Which model of how mentalizing comes to be related to belief is best supported by the data? (Study 1, 2A and 2B) a. Does generalized anthropomorphism mediate the relationship? 49 b. Does mental state attribution to god mediate the relationship? c. Does belief mediate the relationship between individual differences in mentalizing and mental state attribution to god? 3. What are the boundary conditions of the relationship between mentalizing and belief? (Study 2A and 2B) a. Do individual differences in mentalizing support belief in god, controlling for religious upbringing? b. Do individual differences in mentalizing and religious upbringing interact in support of belief in god? (not pre-registered) c. Do individual differences in mentalizing specifically predict beliefs (as compared to religious practices and other indicators of religiosity)? d. Do individual differences in mentalizing support all types of supernatural beliefs? i. Do individual differences in mentalizing predict abstract god beliefs? ii. Do individual differences in mentalizing predict belief in moralized supernatural agents? Study 1: Mentalizing and Belief in Three Samples In Study 1, the replicability of the relationship between mentalizing and belief in god was assessed in three samples \u00E2\u0080\u0093 Canadian students, and American and Indian adults (not students). Furthermore, three theoretical models of how mentalizing comes to be implicated in belief in god were examined: (1) the broad anthropomorphism hypothesis, (2) the narrow anthropomorphism hypothesis, and (3) the anthropocentric knowledge hypothesis. That is, this study provides tests of whether the relationship between mentalizing and belief in god is mediated by (1) generalized anthropomorphic tendencies or (2) by mental state attribution to god; and (3) a test of whether it 50 is belief that mediates the relationship between individual differences in mentalizing and mental state attribution to god. Methods Six hundred and sixty-one participants were sampled from three populations: undergraduate students at a Canadian university, and Americans and Indians recruited on Amazon\u00E2\u0080\u0099s Mechanical Turk (M-Turk) in March 2013. Sample characteristics by population are presented in Table 3.1. The Canadians received course credit for their participation in this study, the American and Indian M-Turkers received .50 USD$. The Canadians and Americans were predominantly Christian, Catholic or Agnostic/Atheist. The Indian sample was mostly Hindu. Table 3.1 Sample characteristics of the three samples Canadian American Indian N 209 223 229 Females 69% 48% 39% Age (years) 19.69 (2.27) 35.19 (13.72) 29.43 (8.48) Formal ed. (years) ---1 13.87 (4.43) 15.75 (4.83) Religious affiliation (n) Christian 41 59 26 Catholic 32 21 5 Protestant 0 12 3 Jewish 2 4 0 Buddhist 20 4 4 Muslim 11 1 30 Sikh 8 1 3 Hindu 0 4 154 Agnostic/ Atheist/None 86 84 2 Other 9 33 2 Notes. 1Undergraduate students were not asked how many years of formal education they have completed. Values for age and formal education are means and standard deviations (in parentheses). 51 Measures Individual differences in mentalizing Individual differences in mentalizing were assessed with 16 items that were adapted from the Empathy Quotient (EQ; Baron-Cohen & Wheelwright, 2004). Participants self-reported on their tendencies for considering the mental states of others (8-items, e.g., \u00E2\u0080\u009CI can usually tell if someone else wants to enter a conversation\u00E2\u0080\u009D), and their self-reported mentalizing abilities (8-item, e.g., \u00E2\u0080\u009CHow easy is it for you to tell if someone else wants to enter a conversation?\u00E2\u0080\u009D). This scale was found to be reasonably reliable in the Canadian (a = 81) and American samples (a = .86); and less so in the Indian sample (a = .75). The EQ features as a recurrent measure of individual differences in mentalizing in the cognitive sciences of religion (e.g., Willard & Norenzayan, 2013). Moreover, this scale is reported to reliably discriminate between individuals diagnosed along the autism spectrum (Baron-Cohen & Wheelwright, 2004) which is associated with decreased capacities for and/or willingness to mentalize. Although a self-report scale, this measure captures relevant variability in participants\u00E2\u0080\u0099 capacities and/or willingness to reason about the mental states of other people; and making it a suitable focal variable for the current research. Mental state attribution to god To assess the extent to which participants attribute mental states to god, participants were asked to, \u00E2\u0080\u009CThink about God\u00E2\u0080\u009D and to rate the extent to which God has the capacities for different mental states: Agency subscale \u00E2\u0080\u0093 making plans, remembering, knowing things; Experience subscale - fear, feeling happy, feeling hunger (1 = \u00E2\u0080\u0098not at all\u00E2\u0080\u0099, 5 = \u00E2\u0080\u0098very much\u00E2\u0080\u0099). These mental states were selected from previous work exploring the types and extent of mental state attribution to a variety of targets, including God, who has previously been reported to be high in agency and lower (than humans \u00E2\u0080\u0093 i.e., typical minds) in experience (Gray, Gray, & Wegner, 2007). Responses 52 on these items were averaged to create an index of mental state attribution to God (Canadian sample: a = .89; American sample: a = .91; Indian sample: a = .85). This measure serves as a direct indicator of the active mentalization of god\u00E2\u0080\u0099s mind hypothesized by the narrow anthropomorphism account to mediate the relationship between individual differences in mentalizing and belief. Generalized anthropomorphism In the Indian and American samples (but not the undergraduate student sample), participants completed the individual differences in anthropomorphism questionnaire (IDAQ; Waytz, Cacioppo, & Epley, 2014). This measure assesses the extent to which participants endorse the attribution of mental states (e.g., consciousness, intentions) to animals, nature and technology and serves as an index of generalized anthropomorphism (American sample: a = .91; Indian sample: a = .91). The broad anthropomorphism hypothesis predicts that it is broad tendencies for ascribing human-like mental states to a diversity of non-human targets that mediates the relationship between individual differences in mentalizing and belief; and employing the IDAQ is a reliable and direct way to test this hypothesis. Belief in god In all samples, belief in god was assessed by the item: \u00E2\u0080\u009CCompared to most people, how much do you believe in God?\u00E2\u0080\u009D (1 = Not at all, 5 = A lot more than most). Although an imperfect measure of belief in god, this item should capture whether or not participants believe in god, and the perceived strength of their conviction. In Studies 2A and 2B, more direct and other types of belief in god were examined. These measures were embedded as part of a larger survey of participant\u00E2\u0080\u0099s religious beliefs and practices. The order of presentation of all measures in the survey were randomized, with 53 demographics placed at the end. All participants completed the survey in English. An additional 27 American and 51 Indian M-Turk workers were removed from the samples prior to analyses for failing at least one of the three attention checks embedded in the questionnaire to ensure data quality (Crump, McDonnell, & Gureckis, 2013). Summary statistics of the measures of mentalizing, mental state attribution to God, and belief in God for each sample are shown in Table 3.2. Table 3.2 Means and standard deviations of focal variables by sample Measure Canadian American Indian Mentalizing 3.05 (.40) 2.98 (.44) 2.81 (.37) Mental state attribution to God 3.31 (.94) 3.25 (1.36) 3.76 (.95) Belief in God 3.01 (1.30) 2.62 (1.40) 3.98 (1.11) Results First, the broad anthropomorphism hypothesis that mentalizing encourages generalized anthropomorphism, which supports belief in God was tested. Although the Indian sample (M = 5.97, SD = 2.04) scored higher than the American sample on generalized anthropomorphism (M = 3.91, SD = 1.84), d = .94, .95CI = [.74, 1.13], there was no correlation between anthropomorphism and belief in God in either sample: Indian sample, r(227) = .03, .95CI = [-.10, .16]; American sample, r(219) = .03, .95CI = [-.10, .16]. This null result replicates Willard & Norenzayan's (2013) test of this relationship and renders the proposed mediation model of mentalizing on belief through generalized anthropomorphism untenable. Further, it was observed that individual differences in mentalizing and anthropomorphism were negatively correlated in both the American (r(220) = -.13, .95CI = [-.26, .00]) and Indian samples (r(226) = -.28, .95CI = [-.39, -.15]). As evidence that broad anthropomorphism and the specific anthropomorphism of god\u00E2\u0080\u0099s mind are distinct processes, there was no relationship between generalized anthropomorphism and mental state attribution to 54 God: Indian sample, r(225) = .07, .95CI = [-.06, .20]; American sample, r(220) = .05, .95CI = [-.08, .19]. Even in the Indian sample where anthropomorphism of nature, animals and technology was more strongly endorsed, these tendencies did not predict belief in God. Next, the narrow anthropomorphism hypothesis was tested by examining whether the extent to which individuals attributed mental states to God mediated the relationship between mentalizing and belief in God. Mediation was assessed in a series of mixed effect models that included a random intercept for sample, and fixed effect controls for age and sex (see Table 3.3 and Figure 3.2). All variables were standardized within sample, and thus regression coefficients should be interpreted in terms of standardized deviation changes from within-sample means (Enders & Tofighi, 2007). Following Preacher & Heyes (2004), the magnitude of the indirect effect of mentalizing in predicting belief in God through mental state attribution to God was calculated in a 9999-replicate bootstrap model. The results of which indicated a non-zero indirect effect of mentalizing on belief in God (Indirect effect = .06, .95CI = [.02, .11]). Including mental state attribution to God in the model reduced the estimated direct effect of mentalizing on belief in God by 28%; and increased overall R2 by 12%. This suggests robust indirect effects across three samples of mentalizing on belief in God as mediated by explicit mental state attributions to God \u00E2\u0080\u0093 and support for the narrow anthropomorphism hypothesis. 55 Table 3.3 Study 1 mediation models of the narrow anthropomorphism hypothesis Model 1 DV = Belief in God Model 2 DV = Mental state attribution to God Model 3 DV = Belief in God Predictors b .95CI b .95CI b .95CI (Intercept) -.13 -.24 \u00E2\u0080\u0093 -.02 -.06 -.17 \u00E2\u0080\u0093 .05 -.11 -.21 \u00E2\u0080\u0093 -.01 Mentalizing .18 .10 \u00E2\u0080\u0093 .25 .14 .06 \u00E2\u0080\u0093 .22 .13 .06 \u00E2\u0080\u0093 .20 Age (years) .04 -.04 \u00E2\u0080\u0093 .11 .07 -.01 \u00E2\u0080\u0093 .15 .01 -.06 \u00E2\u0080\u0093 .08 Female (1 = Yes) .25 .10 \u00E2\u0080\u0093 .41 .11 -.04 \u00E2\u0080\u0093 .27 .22 .07 \u00E2\u0080\u0093 .36 Mental state attribution .35 .28 \u00E2\u0080\u0093 .42 Observations 644 644 644 AIC 1814.77 1833.21 1735.17 Conditional R2 .06 .03 .18 Notes: All models included a random-intercept for sample. Figure 3.2 Narrow anthropomorphism hypothesis mediation model Notes: All paths include participant age, sex and were modelled with a random-intercept for sample. Values in brackets are 95% confidence intervals. c\u00E2\u0080\u0099 is the estimated mediated effect size of mentalizing on belief in god. 56 Next, the third theoretical model - the anthropocentric knowledge hypothesis \u00E2\u0080\u0093 was evaluated (Table 3.4 and Figure 3.3). This account predicts that mental state attribution to god is motivated by belief, not in support of it. In aggregate, the results of this mediation analysis indicated a non-zero indirect effect of mentalizing on mental state attribution to God (Indirect effect = .07, .95CI = [.04, .11]) and an overall 43% decrease in the direct effect. The anthropocentric knowledge model had the virtue of producing more complete mediation (43% vs. 27%) than the narrow anthropomorphism model. But the difference in variance explained in mental state attribution to God in the anthropocentric knowledge model (R2adj = .15) was negligible in comparison to the variance explained in belief in God in the narrow anthropomorphism model (R2adj = .18). Thus, the current data similarly supports the narrow anthropomorphism and anthropocentric knowledge accounts. 57 Table 3.4 Study 1 mediation models of the anthropocentric knowledge hypothesis Model 1 DV = Mental state attribution to God Model 2 DV = Belief in God Model 3 DV = Mental state attribution to God Predictors b .95CI b .95CI b .95CI (Intercept) -.06 -.17 \u00E2\u0080\u0093 .05 -.13 -.24 \u00E2\u0080\u0093 -.02 -.01 -.11 \u00E2\u0080\u0093 .09 Mentalizing .14 .06 \u00E2\u0080\u0093 .22 .18 .10 \u00E2\u0080\u0093 .25 .08 .00 \u00E2\u0080\u0093 .15 Age (years) .07 -.01 \u00E2\u0080\u0093 .15 .04 -.04 \u00E2\u0080\u0093 .11 .06 -.02 \u00E2\u0080\u0093 .13 Female (1 = yes) .11 -.04 \u00E2\u0080\u0093 .27 .25 .10 \u00E2\u0080\u0093 .41 .02 -.13 \u00E2\u0080\u0093 .17 Belief in God .36 .29 \u00E2\u0080\u0093 .43 Observations 644 644 644 Conditional R2 .031 .059 .153 AIC 1833.21 1814.77 1753.54 Figure 3.3 Anthropocentric knowledge hypothesis mediation model Notes: All paths include participant age, sex and were modelled with a random-intercept for sample. Values in brackets are 95% confidence intervals. c\u00E2\u0080\u0099 is the estimated mediated effect size of mentalizing on belief in god. 58 Discussion In this study, individual differences in mentalizing predicted belief in God in samples of Canadians, Americans, and Indians. This is a direct replication of previous work, with an effect magnitude comparable to that of earlier research (\u00CE\u00B2 = .17; Norenzayan et al., 2012; and see also, Willard & Norenzayan, 2013, Banerjee & Bloom, 2015; Lindeman & Lipsanen, 2016; Lindeman, Svedholm-H\u00C3\u00A4kkinen, & Lipsanen, 2015). These results, however, provide new evidence for the contribution of mentalizing to religious belief outside of a predominantly Christian religious context \u00E2\u0080\u0093 a necessary step in testing the generalizability of key findings in the cognitive science of religion (Norenzayan, 2016). In addition to replicating previous work, this study demonstrated that self-reported capacities and tendencies for considering the mental states of other people predicted the extent to which participants attributed mental states to God. That is, mentalizing is not just implicated in any sorts of supernatural beliefs, but specifically about beliefs about god\u00E2\u0080\u0099s mind and its capacities. Although small in size, the observed relationships between mentalizing and mental state attribution to god reflect the small but important contribution of human psychology in shaping mental models of supernatural agents. Lastly, this work specifically demonstrates that the attribution of mental states to God predicts greater belief. In other words, thinking that God has human-like mental capacities, and is therefore more readily conceptualized through mentalizing, may indeed make God more easily believed \u00E2\u0080\u0093 as is predicted by the narrow anthropomorphism hypothesis. These are, however, correlational findings, and thus these models should not be interpreted as indication of causality. As an alternative account, however, reversing the mediation paths in the tested models suggested that \u00E2\u0080\u0093 in support of the anthropocentric knowledge hypothesis - when one readily 59 mentalizes to understand other agents (as do high EQ-scoring mentalizers), a more ardent belief in God leads to greater attribution of mental states to God \u00E2\u0080\u0093 making a more believed God more mentalized. Nevertheless, further work is necessary to help distinguish between these models with more confidence. Importantly, in all samples, the effect of mental state attribution to God was notably larger than the direct effect of mentalizing in predicting belief in God. This is precisely what cognitive scientists of religion have theorized (e.g., McCauley, 2011; Bering, 2011; Norenzayan, 2013) \u00E2\u0080\u0093 but so far had remained untested. That is, the growing literature documenting the relationship between individual differences in mentalizing and belief in god is capturing (but not explicitly so) the tendency of believers to be those who engage in active mentalistic consideration of god\u00E2\u0080\u0099s mind. Although the specific anthropomorphism of god\u00E2\u0080\u0099s mind was found to support belief, generalized anthropomorphism of non-human targets was unrelated to belief in God, replicating previous research (Willard & Norenzayan, 2013). However, this does not imply that the former plays no role in other religious and spiritual beliefs and experiences. The generalized tendency to anthropomorphize the world is correlated with paranormal beliefs such as belief in astrology and extrasensory perception, animistic beliefs such as forest dwelling ancestor spirits, and spiritual experiences involving magical ideation (Atran & Medin, 2008; Willard & Norenzayan, 2013, 2017). Thus, there is more work to be done to further specify the role of anthropomorphic tendencies \u00E2\u0080\u0093 broad and narrow - in various religious and spiritual beliefs and experiences. 60 Studies 2A and 2B: Mentalizing and belief in American Hindus and Christians Studies 2A and 2B were designed as pre-registered and high-powered tests of the replicability of the relationship between mentalizing and belief and focal meditation models presented in Study 1 in two additional samples \u00E2\u0080\u0093 American Hindus and American Christians. In following with the results of Study 1, I specifically hypothesized that broad anthropomorphism would not mediate the relationship between mentalizing and belief in either sample but made no specific hypotheses regarding which of narrow anthropomorphism or anthropocentric knowledge accounts would be better supported in these samples. To get a better estimate of the focal relationships of these proposed models, I employed more reliable measures of individual differences in mentalizing (i.e., the full-length \u00E2\u0080\u0098empathy quotient\u00E2\u0080\u0099; Wakabayashi et al., 2006) and mental state attribution to god (i.e.., a longer list of mental state attribution items taken from Gray et al., 2007). Additionally, rather than focusing on \u00E2\u0080\u0098belief in god\u00E2\u0080\u0099 broadly construed, these studies additionally examined the relationships between mentalizing and beliefs in a personal god \u00E2\u0080\u0093 i.e., a god that is concerned with believers as individuals. It is this type of belief in god that is hypothesized to be most related to believers\u00E2\u0080\u0099 capacities for mentalizing (Norenzayan et al., 2012). Mentalizing \u00E2\u0080\u0093 considering the intentions, desires, and emotional states of another person has long been implicated in relationship success (e.g., Davis & Oathout, 1987; Hooker, Verosky, Germine, Knight, & D\u00E2\u0080\u0099Esposito, 2008). In the case of god(s), the active deployment of one\u00E2\u0080\u0099s capacities for mentalizing in reasoning about god\u00E2\u0080\u0099s mind should then help foster a sense of personal connection with god (Epley, Akalis, Waytz, & Cacioppo, 2008) \u00E2\u0080\u0093 a potentially otherwise unknowable supernatural agent. 61 In these studies, additional measures were included that allow for pre-registered direct tests of the boundary conditions of the relationship between mentalizing and belief. Specifically, these studies tested the following questions: 1. Do individual differences in mentalizing support belief in god, controlling for religious upbringing? 2. Do individual differences in mentalizing and religious upbringing interact in support of belief in god? (not pre-registered) 3. Do individual differences in mentalizing specifically predict beliefs (as compared to religious practices and other indicators of religiosity)? 4. Do individual differences in mentalizing predict all types of supernatural beliefs? a. Do individual differences in mentalizing predict abstract god beliefs? b. Do individual differences in mentalizing predict belief in moralized supernatural agents? The relative contributions of cognition and culture to beliefs Recent evidence (e.g., Maij et al., 2017) has questioned whether mentalizing can be said to contribute anything to \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 after controlling for cultural exposure to religion - typically measured using a self-report scale of parental religiosity (Lanman & Buhrmester, 2016). Specifically, Maij et al. (2017) demonstrated that in a large sample of Dutch participants \u00E2\u0080\u0093 the contributions of individual differences in mentalizing to a broad indicator of \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 were zero after controlling for the extent to which participants grew up in religious households. This work, however, doesn\u00E2\u0080\u0099t necessarily demonstrate that religious upbringing is a better predictor of belief than mentalizing in all cultural contexts. Indeed, this might suggest that low cultural exposure to religion early in life (as was the case amongst mostly secular Dutch participants) suppresses the 62 potential contributions of mentalizing to belief. In the American Hindu and Christian samples employed in the following studies, cultural exposure to religion is expected to be greater than in Maij et al.\u00E2\u0080\u0099s (2017) Dutch sample. And thus, I hypothesized that mentalizing will have a small but robust effect on belief in god after controlling for religious upbringing in these two samples. In an exploratory analysis, I also performed a test of the interaction between mentalizing and religious upbringing in predicting belief to test the hypothesis that low cultural exposure suppresses the contributions of mentalizing to belief. What parts of \u00E2\u0080\u0098religion\u00E2\u0080\u0099 should mentalizing predict? The cognitive sciences of religion have long theorized that the contributions of mentalizing to \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 should be specific to religious beliefs (e.g., Boyer, 2001). However, empirical tests of the contributions of mentalizing to \u00E2\u0080\u0098religion\u00E2\u0080\u0099 are sometimes assessed with broad indicators of religiosity rather than specific contents of beliefs (e.g., Maij et al., 2017). Here, the contributions of mentalizing to belief in god, belief in a personal god, religious participation and generalized religiosity/spiritualty are compared and contrasted. I hypothesized that mentalizing should most strongly and positively predict belief in a personal god in both samples. To further test the specificity of the relationship between mentalizing and belief in a personal god, these next studies also examined whether mentalizing is related to non-mentalistic god concepts (i.e., belief in an abstract god). Beliefs in abstract god(s) are prevalent in both Hindu and Christian religious contexts, but I hypothesized that these types of mindless supernatural agent beliefs (when endorsed) should not be supported by mentalizing. To test whether mentalizing and generalized anthropomorphic tendencies come to support belief in a variety of supernatural targets, these studies also included measures of belief in other supernatural forces (ghosts, angels/devas, demons/asuras, and karma). If mentalizing and anthropomorphism do indeed differentially predict 63 belief in these different agents \u00E2\u0080\u0093 it remains an open question as to what accounts for these differences. In these studies, the hypothesis that mentalizing comes to predict belief in moralized supernatural agents was tested by examining the interactions of mentalizing and anthropomorphism with attributions of moral knowledge. Methods Pre-registration The sampling, materials, and explicit hypotheses (signalled in the text above and below by \u00E2\u0080\u009CI hypothesized\u00E2\u0080\u00A6\u00E2\u0080\u009D) were pre-registered on the Open Science Framework prior to data collection for both samples and are available at the following links: Hindu sample (https://osf.io/9qdra/), Christian sample (https://osf.io/u2yph/). Data collection took place over several weeks in January 2018 for the Hindu sample, and two days in July 2018 for the American Christian sample. In these pre-registrations, the analysis plans were detailed only for the focal mediation models. For other research questions, verbal hypotheses were pre-registered but details about what analysis would be used to test these hypotheses were not. For full disclosure, not all pre-registered hypotheses or variables are presented here. Sample 752 American Hindu and 778 American Christian participants were recruited to participate in Studies 2A and 2B respectively. The Hindu participants were recruited by Qualtrics Survey Panels and were compensated 5 USD$ for completing the 15-minute survey. The Christian participants were recruited via Amazon\u00E2\u0080\u0099s Mechanical Turk and were compensated 1.50 USD$. Data quality checks were implemented in both samples. For the Hindu sample, Qualtrics automatically excluded from the final dataset (prior to delivering the data to me) all participants who completed the study in less than a third of the mean completion time, did not respond 64 appropriately to an embedded attention check (i.e., \u00E2\u0080\u009CIf you are paying attention to this question choose strongly disagree\u00E2\u0080\u009D), or provided non-sensical or gibberish responses (e.g., bot-generated Latin phrases) to an open-ended response question. On MTurk, data quality was checked with the use of two attention checks embedded in the survey. An additional 136 Christians completed the survey but were excluded from and prior to all analyses for responding inappropriately to one or both of these checks. The American Christians mostly identified as Catholic (n = 306) or Protestant (n = 261), but also Evangelical (n = 95), Mormon (n = 18), Episcopalian (n = 14), Baptist (n = 10), Jehovah\u00E2\u0080\u0099s Witness (n = 9), non-denominational Christian (n = 32), and other Christian (n = 24); denominational affiliation was missing from 2 participants. For the Hindu sample, Qualtrics pre-screened and excluded participants based on their religious affiliation, and when asked specifically to report their religious affiliation in the survey 83% identified as \u00E2\u0080\u0098Hindu\u00E2\u0080\u0099; the remaining participants reported mixed affiliations (e.g., Hindu-Jain, Hindu-Buddhist, Hindu-Hare Krishna). Further demographic details are presented in Table 3.5. Table 3.5 Participant demographics in studies 2A and 2B Study 2A: American Hindus Study 2B: American Christians M (SD) or % Range NAs M (SD) or % Range NAs Age (years) 37.40 (12.82) 18 - 93 1 36.65 (12.03) 18 -84 4 Female 58.77% --- 0 53.29% --- 3 Born in the USA? 32.31%1 --- 0 95.87% --- 3 If not, years lived in USA? 15.24 (12.54) 1 - 55 8 21 (15.27) 2-61 4 Income 4.2 (1.77) 1 - 8 0 3.16 (1.40) 1 - 8 4 Formal education (years) 14.65 (6.35) 0 - 36 2 13.79 (4.98) 0 - 25 6 # of Children .96 (.94) 0 - 4 0 1.18 (1.41) 0 - 14 5 Political Conservatism 3.49 (1.51) 1 - 7 0 4.16 (1.73) 1 - 7 4 Notes:187% of participants not born in the USA were born in India. Income was assessed on an 8-point scale (0 USD \u00E2\u0080\u0093 19 000 USD; > 220 000 USD). Political conservatism was assessed on a 7-point scale (1 = very liberal; 7 = very conservative). 65 Measures Participants in both samples were presented with the following scales and items in the order presented here. Anthropomorphism As in Study 1, generalized tendencies for attributing human mental states to non-human targets was assessed with the individual difference in anthropomorphism questionnaire [Hindu a = .90, Christian a = .94] (Waytz et al., 2014). Individual differences in mentalizing Individual differences in mentalizing were assessed using the Short-Form Empathy Quotient (Wakabayashi et al., 2006). This 22-item validated scale was designed to broadly capture individual differences in capacities for reasoning about other minds with items such as, \u00E2\u0080\u009CI can pick up quickly if someone says one thing but means another\u00E2\u0080\u009D, and \u00E2\u0080\u009CI am good at predicting how someone will feel\u00E2\u0080\u009D. Participants were asked to rate their agreement with the items on a 4-point (strongly disagree to strongly agree scale) and responses indicating high mentalizing were scored with a 2, weak mentalizing was scored as 1, and all other responses received no points (in following with Wakabayashi et al., 2006). Scored values were then averaged across all items to create an index of participant\u00E2\u0080\u0099s mentalizing. Responses indicated good reliability in both samples: Hindu a =.86, Christian a = .89. Mental state attribution to god In Study 1, mental state attribution to god was assessed using only six items (half of which assessed attributions of \u00E2\u0080\u0098agency\u00E2\u0080\u0099 (e.g., \u00E2\u0080\u009CGod longs for and hopes for things\u00E2\u0080\u009D, \u00E2\u0080\u009CGod makes plans and works towards goals\u00E2\u0080\u009D), and half attributions of \u00E2\u0080\u0098experience\u00E2\u0080\u0099 (e.g., \u00E2\u0080\u009CGod experiences 66 embarrassment\u00E2\u0080\u009D, \u00E2\u0080\u009CGod experiences pride\u00E2\u0080\u009D). In Studies 2A and 2B, the full 18-item list of mental state attributions from Gray, Gray, & Wegner (2007) was employed to more reliably capture the ways in which individuals attribute mental states to god. Participants rated the extent of their agreement with statements conveying mental state attribution to god on a 7-point (strongly disagree to strongly agree) scale and responses were averaged into a single overall score (Hindu a = .94; Christian a = .90) and along the two dimensions of mind perception: agency (Hindu a = .92; Christian a = .90) and experience (Hindu a = .90; Christian a = .80). Belief in god and a personal god Study 1 examined the relationship between individual differences in mentalizing and belief in god (measured broadly). In Studies 2A and 2B, in addition to a generalized indicator of belief in god (i.e., \u00E2\u0080\u009CI believe in god\u00E2\u0080\u009D; 1 = strongly disagree, 7 = strongly agree), I also included a more specific measure of belief in a personal god. Belief in a personal god was assessed with 10 items from the spiritual well-being scale (Paloutzian & Kaufman, 1991) that focus on individuals perception of their experienced personal connection to god (e.g., \u00E2\u0080\u009CI have a personally meaningful relationship with God\u00E2\u0080\u009D, and \u00E2\u0080\u009CI believe that God is concerned about my problems\u00E2\u0080\u009D). Participants rated the extent to which they agreed with the provided statements on a 7-point scale (strongly disagree to strongly agree). Responses indicated high reliability in both samples (Hindu a = .90; Christian a = .94). This measure was included to test hypotheses regarding whether mentalizing is most strongly related to belief in personal rather than other forms of supernatural agent beliefs. Belief in an abstract god Participants were asked to rate the extent to which they agreed with 6 items depicting belief in an abstract god (e.g., \u00E2\u0080\u009CI believe in a God that is abstract\u00E2\u0080\u009D, \u00E2\u0080\u009CI believe in a God that is impersonal\u00E2\u0080\u009D, \u00E2\u0080\u009CI believe in a God that is attributeless\u00E2\u0080\u009D) on a 7-point scale (strongly disagree to strongly agree). 67 These items were selected from a larger pool of items that in a pre-test sample of American Christians demonstrated good internal consistency; and were distinguished from other types of abstract beliefs about god\u00E2\u0080\u0099s transcendental qualities. Responses indicated good reliability in both samples (Hindu a = .88; Christian a = .89) and were averaged to create an index of beliefs in an abstract god. This measure was included to test hypotheses regarding the specificity of mentalizing\u00E2\u0080\u0099s contributions to personal god beliefs (and not just any type of god belief). Belief in and attributed qualities of other supernatural targets Participants were then asked to respond to a series of items assessing their belief in (\u00E2\u0080\u009CI believe in [target]\u00E2\u0080\u009D) and the extent to which they attributed moral knowledge to (\u00E2\u0080\u009CI believe that [target] knows about my [thoughts/actions], \u00E2\u0080\u009CI believe that [target] [punishes bad/rewards good] behaviors\u00E2\u0080\u009D) four other supernatural targets. In both samples, participants were asked these questions about ghosts and karma. In the Hindu sample, participants were also asked about devas and asuras (\u00E2\u0080\u0098helpful\u00E2\u0080\u0099 and \u00E2\u0080\u0098hindering\u00E2\u0080\u0099 Hindu-relevant supernatural agents). In the Christian sample, participants were instead asked about angels and demons (\u00E2\u0080\u0098helpful\u00E2\u0080\u0099 and \u00E2\u0080\u0098hindering\u00E2\u0080\u0099 Christian-relevant supernatural agents). Analytical thinking Analytical thinking was assessed using the three-item cognitive reflections test (Frederick, 2005). Participants responses to the three questions were given a score of 1 for correct answers, and 0 for all other responses. Scores were then summed. Reliability of this three-item test was lower in the Hindu sample (a = .69) than in the Christian sample (a = .80). This measure was included to test whether \u00E2\u0080\u0093 as is predicted by the narrow anthropomorphism hypothesis \u00E2\u0080\u0093 the attribution of mental states to god is supported by an intuitive thinking style. Religious upbringing 68 As an indicator of cultural exposure to \u00E2\u0080\u0098religion\u00E2\u0080\u0099, participants responded to a 7-item measure of religious upbringing (Lanman & Buhrmester, 2016). Participants were asked to report the extent to which (1 \u00E2\u0080\u0093 to no extent at all; 7 \u00E2\u0080\u0093 to an extreme extent), for example, their \u00E2\u0080\u009Ccaregiver(s) attended religious services or meetings\u00E2\u0080\u009D, or their \u00E2\u0080\u009Ccaregiver(s) acted as good religious role models\u00E2\u0080\u009D. Average scores were calculated as an index of cultural exposure to religion, and responses demonstrated high reliability in both samples (Hindu a = .91, Christian a = .92). When controlling for cultural exposure to religion as measured by this scale of religious upbringing, the contributions of mentalizing have been reported to be negligible or small (e.g., Maij et al., 2017; Willard & Cingl, 2017). Here, I examine the relative contributions of mentalizing and religious upbringing (using the same measures) in two different cultural contexts. Other indicators of \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 Participants were asked to respond to a number of indicators of religiosity including: generalized religiosity/generalized spirituality (\u00E2\u0080\u009CHow religious/spiritual are you\u00E2\u0080\u009D, 7-point not at all to very scale), religious service attendance (never to more than once a week, 8-point scale), and prayer frequency (never to more than once a week, 8-point scale). These items were included to test hypotheses regarding the specificity of the contributions of mentalizing to belief (and not, or at least more than, any and all indicators of religiosity). Demographics Lastly, participants provided demographic information about their age, sex, ethnicity, whether they were born in the United States (if not, where they were born and how long they have they lived in the USA), income, number of children, years of formal education, and political orientation (1 = very liberal, 7 = very conservative). 69 Results Does the relationship between individual differences in mentalizing and belief in a personal god vary between different religious populations? It was hypothesized that individual differences in mentalizing would be similarly related to belief in god in both the American Christian and American Hindu samples. In support of this hypothesis, mentalizing was found to be comparably correlated with belief in a personal god in the Hindu (r (750) = .25 [.18, .31]) and Christian samples (r (776) = .31 [.25, .37]). Given the stable positive correlations in both samples, I next tested the planned mediation models (full mediation summary tables are presented in the appendices, Table S3.1). Does generalized anthropomorphism mediate the relationship between mentalizing and belief in a personal god? In following the results of Study 1, it was hypothesized that generalized anthropomorphism would not predict belief in a personal god and thus, would not mediate the relationship between mentalizing and belief. Put simply, I hypothesized that no support for the broad anthropomorphism hypothesis would be observed in either of these additional samples. Importantly, the American Hindu sample provides a further opportunity to test whether the lack of an observed relationship between anthropomorphism and belief in god results from the suppression of anthropomorphism in Christians; or rather that broad anthropomorphic tendencies do not support belief in god. Contrary to my prediction (but still not in support of the broad anthropomorphism hypothesis), generalized anthropomorphism was found to be negatively correlated with belief in a personal god in Christians (r (776) = -.19 [-.26, -.12]). This latter result may again reflect the active suppression of broad anthropomorphism in Christianity. In line with my prediction, anthropomorphism was unrelated to belief in a personal god in the Hindu sample (r (750) = -.02 [-.09, .05]). And thus, it 70 appears that even in a religious context that promotes broad anthropomorphism, these tendencies do not come to predict belief in a personal god. Moreover, in both samples, anthropomorphism was unrelated or even weakly negatively related to individual differences in mentalizing (Hindu sample: r (750) = -.04 [-.11, .03]; Christian sample: r (776) = -.07 [-.14, .00]). In support of my prediction and replicating Study 1 results, these correlations make the mediation model proposed by the broad anthropomorphism hypothesis untenable. Does mental state attribution to god mediate the relationship between mentalizing and belief in a personal god? To directly test the narrow anthropomorphism hypothesis, the mediating role of mental state attribution to god on the relationship between individual differences in mentalizing and belief in god was assessed (summarized in Figure 3.4). As hypothesized, the bootstrapped indirect effect in both samples was found to be positive, but small (Hindu sample: .03 [.00, .06]; Christian sample: .07 [.04, .11]). The estimated direct effect of mentalizing on belief was reduced by 13.6% in the Hindu sample and 26% in the Christian sample with the addition of mental state attribution to god in the paths. The difference between samples is perhaps attributable in-part to a smaller contribution of mentalizing to mental state attribution to god in the Hindu sample. The estimated standardized effect of mental state attribution to belief in a personal god, interestingly, was strikingly similar in both samples. The final model accounted for 16% of the variance in belief in a personal god in the Hindu sample, and 30% in the Christian sample. Thus, as was observed in Study 1 \u00E2\u0080\u0093 the data supports the model in which mentalizing comes to be related to belief in a personal god in part because individuals deploy their capacities for reasoning about human minds to actively and specifically mentalize god\u00E2\u0080\u0099s mind. The attribution of mental states to god, in turn, supports belief in a personal god. 71 Figure 3.4 Testing the narrow anthropomorphism hypothesis in two samples Does belief in a personal god mediate the relationship between mentalizing and mental state attribution to god? The anthropocentric knowledge hypothesis predicts that mental state attribution to god results from an otherwise acquired belief in a personal god rather than in support of it. As in Study 1, this hypothesis was tested by assessing the mediating role of belief in a personal god in the relationship between individual differences in mentalizing and mental state attribution to god. No specific hypothesis was made a-priori about this model. As in Study 1, mediation analyses (summarized in Figure 3.5) revealed a positive indirect effect in both samples (Hindu sample = 72 .09 [.00, .13]; Christian sample = .13 [.09, .17]). In the Hindu sample, the direct effect of mentalizing on mental state attribution to god was reduced, on average, to zero \u00E2\u0080\u0093 and in the Christian sample by 72.22%. The total model accounted for 9% of the variance in belief in a personal god in the Hindu sample, and 21% in the Christian sample. Thus, the data from both of these samples also supports the hypothesis that belief in a personal god leads individuals to deploy their mentalizing in support of anthropomorphic god beliefs. Figure 3.5 Testing the anthropocentric knowledge hypothesis in two samples 73 Selecting between the narrow anthropomorphism and broad anthropocentric knowledge hypotheses The rearrangement of the predicted paths of the narrow anthropomorphism and anthropocentric knowledge accounts does not change the underlying shared covariance matrix of the included variables. And thus, there is little to be done statistically to select between these two mediation models in the current mediational framework (Thoemmes, 2015). However, additional exploratory analyses (i.e., the following analyses were not pre-registered) can help infer the limits of these theoretical models in accounting for the data. The anthropocentric knowledge hypothesis posits that believers apply models of human minds to fill in the gaps of their understanding of god\u00E2\u0080\u0099s mind given their otherwise acquired belief in god. Under this model, there would be little reason to expect that the relationship between mentalizing and belief in a personal god would be differentially mediated by the types of mental states attributed to god. As (1) human minds are both high in agency and experience (Gray et al., 2007), and (2) mentalizing is deployed to reason about both types of human mental states (Baron-Cohen & Wheelwright, 2004; Gray et al., 2011) \u00E2\u0080\u0093 it follows that attributing god with both of these types of mental states should make god\u00E2\u0080\u0099s mind more human-like, and thus even more knowable than the attribution of either dimension alone. In these samples, god was found to be attributed with more agency than experience, but was not denied experience (Figure 3.6, Panel A; replicating past work Gray et al., 2007)). And, although these facets are strongly correlated in both samples (Hindu r (750) = .71 [.68, .75]; Christian r (776) = .59 [.55, .64]; Figure 3.6; Panel B); their contributions to belief in a personal god and their relationships with mentalizing were found to markedly differ. 74 Figure 3.6 Distribution and correlations of agency and experience attributions As presented in Figure 3.7, the relationship between mentalizing and belief in a personal god is specifically mediated by attributions of agency and not experience to god \u00E2\u0080\u0093 in both samples. Indeed, the indirect effect of mentalizing on belief in a personal god through mental state attribution to god is suppressed by the treatment of mental state attribution to god as a single dimension. Compared to the previously presented tests of the narrow anthropomorphism hypothesis (see Figure 3.4) \u00E2\u0080\u0093 indirect effect estimates increased by 333% in the Hindu sample (total model R2 = .39; D = + .25) and 185% in the Christian sample (model R2 = .40; D = +.14). In these models, mentalizing is specifically related to the attribution of agency \u00E2\u0080\u0093 which in turn more strongly predicts belief in a personal god. Unrelated to mentalizing, the attribution of experience to god negatively predicted belief in a personal god. Thus, mentalizing is not deployed to make god\u00E2\u0080\u0099s mind more human in the ways that would be predicted by the anthropocentric knowledge account \u00E2\u0080\u0093 but rather, specifically more agentic. And, while demonstrating a limit to the anthropocentric knowledge account, these models also provide further statistical support for the narrow anthropomorphism hypothesis in both samples (i.e., the indirect effects are largest HINDU CHRISTIAN2 4 6 2 4 60.00.10.20.30.4Mental State AttributionDensityAGENCY EXPERIENCEA2462 4 6AGENCYEXPERIENCEHINDUCHRISTIANB 75 through the agency pathway; see Figure 3.7). The question remains, however, as to why agency-attributions are specifically implicated in the relationship between mentalizing and belief. Figure 3.7 Agency and experience models Are narrowly anthropomorphic conceptions of god supported by intuitive thinking? The narrow anthropomorphism hypothesis posits that mentalizing supports belief by virtue of making god\u00E2\u0080\u0099s mind more intuitively-grasped. And thus, I hypothesized that mental state attribution to god would be negatively correlated with analytical thinking as measured by the Cognitive Reflections Test. To test this hypothesis, the contributions of scores on the CRT to mental state attribution to god were assessed in a linear regression models in each sample (see Table 3.6). A negative, but small, effect of analytical thinking on mental state attribution to god was observed in the Hindu sample \u00E2\u0080\u0093 as was hypothesized. However, analytical thinking was unrelated to mental state attribution in Christians. Moreover, analytical thinking was very weakly related to belief in a personal god in either sample (Hindu r (750) = -.03 [-.10, .04]; Christian r (776) = -.05 [-.12, .02]). And thus, while the mediation models mostly support the narrow 76 anthropomorphism hypothesis and to a lesser extent the anthropocentric knowledge account, no support for intuitive thinking style as a potential mechanism was observed in these samples. Table 3.6 The contributions of analytical thinking to anthropomorphic god beliefs Mental state attribution to god Hindu Christian Predictors b CI b CI (Intercept) .04 -.09 \u00E2\u0080\u0093 .18 -.07 -.20 \u00E2\u0080\u0093 .06 Analytical thinking -.10 -.16 \u00E2\u0080\u0093 -.03 -.00 -.06 \u00E2\u0080\u0093 .06 Age (years) -.16 -.23 \u00E2\u0080\u0093 -.09 .01 -.07 \u00E2\u0080\u0093 .08 Female (1 = yes) .06 -.09 \u00E2\u0080\u0093 .21 .13 -.01 \u00E2\u0080\u0093 .28 Formal education (years) -.07 -.15 \u00E2\u0080\u0093 .00 -.06 -.13 \u00E2\u0080\u0093 .02 R2 / adjusted R2 .056 / .051 .008 / .003 Do individual differences in mentalizing support belief in god after controlling for religious upbringing? It was hypothesized that mentalizing would have a small but robust effect on belief in a personal god after controlling for cultural exposure to religion (i.e., religious upbringing). To test this, the contributions of mentalizing and religious upbringing (controlling for demographics \u00E2\u0080\u0093 i.e., age, sex, and years of formal education) to belief in a personal god were examined in a linear regression model in both samples (see Table 3.7). As predicted, mentalizing positively predicted belief in a personal god after controlling for religious upbringing in the Hindu sample, and as predicted the relationship between mentalizing and belief was smaller than that of religious upbringing. However, mentalizing was a slightly stronger predictor of belief in a personal god in 77 the American Christian sample (contrary to my prediction and recent evidence taken to suggest that religious upbringing should be more related to belief; Maij et al., 2017). Speaking to the unique contributions of these cognitive and cultural contributors to belief in a personal god \u00E2\u0080\u0093 no clear interaction between these measures was observed in either sample (Hindu sample interaction = -.00 [-.06, .06]; Christian sample interaction = .04 [-.02, .10]). These results indicate that at least in the religious contexts examined here \u00E2\u0080\u0093 mentalizing and religious upbringing independently contribute to belief in a personal god. Table 3.7 Linear regression models of cultural and cognitive predictors of belief in a personal god Belief in a personal god Hindu Sample Christian Sample b .95CI b .95CI (Intercept) -.15 -.25 \u00E2\u0080\u0093 -.04 -.13 -.23 \u00E2\u0080\u0093 -.03 Religious upbringing .26 .19 \u00E2\u0080\u0093 .33 .19 .13 \u00E2\u0080\u0093 .26 Mentalizing .17 .10 \u00E2\u0080\u0093 .24 .25 .18 \u00E2\u0080\u0093 .32 Age (years) .05 -.02 \u00E2\u0080\u0093 .12 .19 .12 \u00E2\u0080\u0093 .26 Female (1 = Yes) .25 .11 \u00E2\u0080\u0093 .39 .24 .10 \u00E2\u0080\u0093 .37 Formal education (years) .00 -.07 \u00E2\u0080\u0093 .07 -.01 -.07 \u00E2\u0080\u0093 .06 R2 / adjusted R2 .140 / .135 .180 / .175 Does mentalizing specifically predict beliefs in a personal god? It was hypothesized that individual differences in mentalizing should most strongly and positively predict belief in a personal god. To test this hypothesis, the bivariate correlations of mentalizing and forms of religiosity were examined (see Figure 3.8). In two distinct religious traditions, the hypothesis that mentalizing would most strongly predict belief in a personal god 78 was supported. For Christians, the relationship between mentalizing and belief in a personal god was the largest (followed by a single item indicator of belief in god, prayer frequency, and spirituality), whereas the correlations with generalized religiosity and ritual practices were centered around zero. A similar pattern of results was observed in the Hindu sample, with the addition of an unpredicted relationship between mentalizing and spirituality. These results support the notion that mentalizing is most implicated in religious beliefs (and not \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099). Figure 3.8 Bivariate correlations [95% confidence intervals] of mentalizing and religiosity indicators by samples Does mentalizing support all types of supernatural beliefs? The previous analyses demonstrate that mentalizing is most strongly related to religious beliefs. Importantly, however, this does not mean that mentalizing comes to predict all types of supernatural beliefs nor even all types of beliefs about supernatural entities. To test this, I predicted that although prominent in both Christian and Hindu traditions, mentalizing would not predict belief in abstract (i.e., mindless) gods. In a linear regression, mentalizing was found to be negatively, but weakly, related to belief in an abstract god (see Table 3.8). Although, on average, Service AttendanceRitual PerformanceReligiositySpiritualityPrayer FrequencyBelief \u00E2\u0088\u0092 God (1 Item)Belief \u00E2\u0088\u0092 Personal God0.0 0.1 0.2 0.3Correlation with mentalizing [EQ]Measure HINDUCHRISTIAN 79 abstract god concepts were endorsed less than personalized ones in both samples (Hindu d = -.55 [-.66, -.45]; Christian d = -1.36 [-1.47, -1.25] \u00E2\u0080\u0093 neither sample entirely denied belief in abstract god concepts (Hindu M = 4.32, SD = 1.50; Christian M = 3.38, SD = 1.68; on a 7-point scale). Thus, these results suggest that mentalizing is not related to belief in any type of god \u00E2\u0080\u0093 but more specifically personal ones. And interestingly, greater mentalizing might constrain belief in non-personalized god concepts.Table 3.8 Predicting belief in an abstract god Belief in an abstract god Hindu Sample Christian Sample b CI b CI (Intercept) 0.07 -0.05 \u00E2\u0080\u0093 0.18 0.15 0.05 \u00E2\u0080\u0093 0.25 Mentalizing -0.09 -0.16 \u00E2\u0080\u0093 -0.01 -0.17 -0.23 \u00E2\u0080\u0093 -0.10 Age (years) -0.06 -0.14 \u00E2\u0080\u0093 0.01 -0.17 -0.24 \u00E2\u0080\u0093 -0.10 Female (1 = Yes) -0.11 -0.26 \u00E2\u0080\u0093 0.04 -0.28 -0.42 \u00E2\u0080\u0093 -0.14 Formal education (years) -0.00 -0.08 \u00E2\u0080\u0093 0.07 -0.06 -0.13 \u00E2\u0080\u0093 0.01 R2 / adjusted R2 . 0.015 / 0.009 0.106 / 0.101 To further test hypotheses regarding what types of supernatural agent beliefs are supported by mentalizing, I examined the relationship between mentalizing, broad anthropomorphism and belief in four other supernatural targets (karma, devas/angels, asuras/demons, and ghosts; see Figure 3.9). 80 Figure 3.9 Correlations between mentalizing, anthropomorphism and belief in varied supernatural targets Notes: EQ = Individual differences in mentalizing (empathy quotient); IDAQ = individual differences in generalized anthropomorphism questionnaire In these samples, variability in both direction and magnitude was observed in the relationships between mentalizing, generalized anthropomorphism and belief in supernatural targets other than god (see Figure 3.9). Of particular interest to the current discussion is (1) determining whether the variability in contributions of mentalizing and anthropomorphism within samples can be accounted for by differences in the believed features of these supernatural targets, and (2) whether differences between samples can be accounted for by sample differences in these same features. In brief, these analyses were conducted to begin accounting for why mentalizing and anthropomorphism are differentially related to belief in certain supernatural targets. To address these questions, I first examined whether attributions of moral knowledge (i.e., whether these targets know about one\u00E2\u0080\u0099s thoughts/behaviours and congruently punish and reward HINDU CHRISTIAN\u00E2\u0088\u00920.1 0.0 0.1 0.2 0.3 \u00E2\u0088\u00920.1 0.0 0.1 0.2 0.3GHOSTSASURAS/DEMONSDEVAS/ANGELSKARMACorrelation [95% CI]Belief (1 Item)EQ IDAQ 81 good and bad thoughts/behaviours) moderated the contributions of mentalizing and anthropomorphism to belief in these other agents within samples. Then, I more closely examined the clearest observed departure between samples \u00E2\u0080\u0093 the relationships between mentalizing, anthropomorphism and belief in karma (as is observable in Figure 3.9) \u00E2\u0080\u0093 and tested whether sample differences in the attribution of moral knowledge to karma mediated the observed sample differences in the relationships between mentalizing, anthropomorphism and belief in karma. That is, in two different ways \u00E2\u0080\u0093 I tested the hypothesis that attributions of moral knowledge moderate the contributions of mentalizing and anthropomorphism to belief. As the narrow anthropomorphism hypothesis suggests, mentalizing is employed to actively reason about supernatural mental states; but it is attributions of moral knowledge that motivate thinking about the contents of a supernatural agent\u00E2\u0080\u0099s mind in the first place. If supernatural minds did not contain relevant information for believers (i.e., like whether or not they are going to be eternally punished for their behaviours), there is little reason for believers to consider their mental states. To test for the moderating role of moral knowledge attributions within samples, a mixed-effect linear regression model with a random-intercept for supernatural target (i.e., karma, devas/angels, asuras/demons, and ghosts) predicting belief in these targets was conducted for each sample. To do this, responses to the belief items and moral knowledge attribution index were stacked by deity. Included as fixed effects in the models were mentalizing, anthropomorphism, age, sex, and years of formal education (see Table 3.9, Model 1). These models indicate that mentalizing (Hindu sample = .08 [.05, .11]; Christian sample = .07 [.03, .10]) and anthropomorphism (Hindu sample = .12 [.09, .15]; Christian sample = .16 [.13, .20]) contribute positively, but weakly, to reported belief in these other targets in both samples. And contrary to 82 previously reported models predicting belief in god \u00E2\u0080\u0093 generalized anthropomorphism was a better predictor of belief in these other agents, in both samples. Next, the main effects of moral knowledge attribution and its interaction with mentalizing and with anthropomorphism were added as additional fixed effects (see Table 3.9, Model 2). From this, it is observed that moral knowledge attributions have a similarly sizeable positive main effect on belief in both samples (Hindu sample = .64 [.60, .67]; Christian sample = .67 [.63, .71]). Furthermore, a small moderating effect of moral knowledge attribution on mentalizing was observed in the American Christian sample (b = .04 [.01, .07]), but not in the Hindu sample (b = -.01 [-.03, .02]). In the Christian sample, mentalizing only weakly predicted belief when moral knowledge attributions were low (b = .06 [.03, .10]). When moral knowledge attributions were high, mentalizing was more strongly, although still weakly, related to belief (b = .14 [.09, .20]). And thus, this provides some preliminary evidence that at least amongst Christians mentalizing comes to be more related to belief in relation to the extent to which supernatural targets are attributed with moral knowledge. Interestingly, in the Christian sample the relationship between anthropomorphism and belief decreased in size the more targets were attributed with moral knowledge. Indeed, at low levels of moral knowledge attribution, anthropomorphism was largely unrelated to belief (b = -.04 [-.09, .02]). And, at high levels of moral knowledge attribution, anthropomorphism was more negatively related to belief (b = -.14 [-.18, -.09]). This latter result was also observed in the Hindu sample, but the effect was weaker (b = -.06 [-.10, -.02]). And at low levels of moral knowledge attribution, anthropomorphism more positively, but again weakly, predicted belief in the Hindu sample (b = .05 [.01, .09]). These interactions are depicted in Figures 3.10 and 3.11. 83 Table 3.9 The moderating effects of moral knowledge on the contributions of mentalizing and anthropomorphism to belief Hindu Sample Christian Sample Model 1 Model 2 Model 1 Model 2 Predictors b CI b CI b CI b CI (Intercept) -.18 -.76 \u00E2\u0080\u0093 .40 -.04 -.25 \u00E2\u0080\u0093 .17 -.22 -.60 \u00E2\u0080\u0093 .16 -.04 -.26 \u00E2\u0080\u0093 .19 Mentalizing .08 .05 \u00E2\u0080\u0093 .11 .09 .06 \u00E2\u0080\u0093 .12 .07 .03 \u00E2\u0080\u0093 .10 .10 .07 \u00E2\u0080\u0093 .13 Anthropomorphism .12 .09 \u00E2\u0080\u0093 .15 -.01 -.03 \u00E2\u0080\u0093 .02 .16 .13 \u00E2\u0080\u0093 .20 -.09 -.12 \u00E2\u0080\u0093 -.05 Age (years) -.02 -.05 \u00E2\u0080\u0093 .02 .01 -.02 \u00E2\u0080\u0093 .04 -.01 -.04 \u00E2\u0080\u0093 .03 .04 .01 \u00E2\u0080\u0093 .07 Female (1 = yes) .07 .00 \u00E2\u0080\u0093 .13 .05 -.00 \u00E2\u0080\u0093 .11 .12 .05 \u00E2\u0080\u0093 .19 .14 .08 \u00E2\u0080\u0093 .20 Formal education (years) -.05 -.08 \u00E2\u0080\u0093 -.01 -.01 -.04 \u00E2\u0080\u0093 .02 -.05 -.09 \u00E2\u0080\u0093 -.02 -.04 -.07 \u00E2\u0080\u0093 -.01 Moral Knowledge .64 .60 \u00E2\u0080\u0093 .67 .67 .63 \u00E2\u0080\u0093 .71 Mentalizing * Moral Knowledge -.01 -.03 \u00E2\u0080\u0093 .02 .04 .01 \u00E2\u0080\u0093 .07 Anthropomorphism * Moral Knowledge -.06 -.08 \u00E2\u0080\u0093 -.03 -.05 -.08 \u00E2\u0080\u0093 -.02 Observations 2994 2994 3083 3083 Marginal R2 / Conditional R2 .024 / .332 .415 / .460 .036 / .171 .316 / .364 84 Figure 3.10 Interactions of mentalizing and anthropomorphism with moral knowledge attributions \u00E2\u0080\u0093 Hindu sample Figure 3.11 Interactions of mentalizing and anthropomorphism with moral knowledge attributions \u00E2\u0080\u0093 Christian sample \u00E2\u0088\u00921.0\u00E2\u0088\u00920.50.00.51.0\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2 3MentalizingBelief (1 Item)\u00E2\u0088\u00921.0\u00E2\u0088\u00920.50.00.51.0\u00E2\u0088\u00922 0 2AnthropomorphismBelief (1 Item) Moral Knowledge\u00E2\u0088\u00921 SDMean+1 SD\u00E2\u0088\u00921.0\u00E2\u0088\u00920.50.00.51.0\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2MentalizingBelief (1 Item)\u00E2\u0088\u00921.0\u00E2\u0088\u00920.50.00.51.0\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2 3AnthropomorphismBelief (1 Item) Moral Knowledge\u00E2\u0088\u00921 SDMean+1 SD 85 These models suggest that moral knowledge attributions do indeed account for some of the variability in the extent to which mentalizing and anthropomorphism are related to belief in diverse supernatural targets. However, the observed patterns were not entirely consistent between samples. Next, I examined whether moral knowledge attributions mediated the more obvious sample differences in the contributions of mentalizing and anthropomorphism to belief in karma. That is, I tested the hypothesis that the difference in how anthropomorphism and mentalizing come to be related to belief between samples will be mediated by sample differences in moral knowledge attribution. Hindu participants reported more belief in karma (d = .90 [.80, 1.01]) and attributed karma with more moralized knowledge (d = .91 [.81, 1.02]) than Christians (see Figure 3.12). Importantly, however, the Christian sample \u00E2\u0080\u0093 on average \u00E2\u0080\u0093 did not strongly deny belief in karma, nor moral knowledge attributions to karma. Figure 3.12 Mean belief in and moral knowledge attribution to karma by sample As would be expected given the correlations above, the estimated relationship between mentalizing and belief in karma was found to differ between samples in a linear regression model on the collapsed sample with controls for age, sex, and formal education (see Model 1, Table 3.10. 0246Belief Moral KnowledgeKarmaMean HINDUCHRISTIAN 86 In Hindus, belief in karma was positively predicted by mentalizing (b = .13 [.07, .19] and weakly negatively by anthropomorphism (b = -.05 [-.12, -.01). In Christians, the reverse pattern was observed \u00E2\u0080\u0093 a weak positive effect of mentalizing (b = .06 [-.09, .17]) and a clear positive effect of anthropomorphism (b = .33 [.17, .48]) on belief in karma. In Model 2 (see Table 3.10), moral knowledge attributions and its interactions with mentalizing and anthropomorphism were added. The inclusion of which mediated the already small sample difference in the relationship between mentalizing and belief in karma and greatly reduced the estimated cultural difference in the relationship between anthropomorphism and belief in karma by 82% (see difference in bolded values between Models 1 and Model 2 in Table 3.10). Although moral knowledge did not specifically moderate either the effect of mentalizing or anthropomorphism in the Hindu sample, both effects were moderated in the expected directions in the Christian sample (see Figure 3.13). What this indicates is that for American Christians who attribute moral knowledge to karma, the relationship between individual differences in mentalizing and belief in karma is more comparable to that of the average relationship between mentalizing and belief in karma in American Hindus. More broadly, these results support the hypothesis that mentalizing \u00E2\u0080\u0093 specifically - becomes increasingly relevant for belief when a supernatural agent/force is attributed with moralized knowledge. 87 Table 3.10 Predicting belief in karma across samples Model 1 Model 2 Predicting belief in karma b CI b CI (Intercept) .33 .25 \u00E2\u0080\u0093 .41 .21 .14 \u00E2\u0080\u0093 .28 Difference in belief in karma in Christians compared to Hindus -.82 -.90 \u00E2\u0080\u0093 -.73 -.28 -.36 \u00E2\u0080\u0093 -.20 Mentalizing (in Hindu sample) .13 .07 \u00E2\u0080\u0093 .19 .12 .06 \u00E2\u0080\u0093 .17 Difference in the effect of mentalizing in the Christian sample -.07 -.16 \u00E2\u0080\u0093 .02 -.03 -.11 \u00E2\u0080\u0093 .04 Anthropomorphism (in Hindu sample) -.05 -.12 \u00E2\u0080\u0093 .01 -.10 -.15 \u00E2\u0080\u0093 -.04 Difference in the effect of anthropomorphism in the Christian from that of the Hindu sample .38 .29 \u00E2\u0080\u0093 .47 .07 -.00 \u00E2\u0080\u0093 .15 Moral knowledge (in Hindu sample) .36 .29 \u00E2\u0080\u0093 .42 Difference in the effect of moral knowledge in the Christian sample from that of the Hindu sample .45 .36 \u00E2\u0080\u0093 .53 Mentalizing * Moral knowledge (in Hindu Sample) -.03 -.09 \u00E2\u0080\u0093 .02 Difference in the Mentalizing * Moral knowledge interaction in the Christian sample from that of the Hindu sample .08 .01 \u00E2\u0080\u0093 .16 Anthropomorphism * Moral Knowledge (in Hindu sample) .02 -.05 \u00E2\u0080\u0093 .08 Difference in the Anthropomorphism * Moral knowledge interaction in the Christian sample from that of the Hindu sample -.14 -.22 \u00E2\u0080\u0093 -.06 Age (years) -.00 -.05 \u00E2\u0080\u0093 .04 .05 .01 \u00E2\u0080\u0093 .08 Female (1 = yes) .15 .06 \u00E2\u0080\u0093 .24 .09 .02 \u00E2\u0080\u0093 .16 Formal education (years) -.01 -.06 \u00E2\u0080\u0093 .03 -.00 -.04 \u00E2\u0080\u0093 .03 R2 / adjusted R2 .240 / .236 .562 / .558 88 Figure 3.13 The moderating effects of moral knowledge attributions on the contributions of mentalizing and anthropomorphism to belief in karma Moral knowledge and agency attributions to personal gods Taken together, the results thus far indicate that individual differences in mentalizing are (1) most strongly related to belief in a personal god (and not other indicators of religiosity, or other types of gods \u00E2\u0080\u0093 i.e., abstract ones) and (2) that the contribution of mentalizing to belief in a personal god are robust to additional controls for religious upbringing (at least in these samples of American HINDU CHRISTIAN\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2 3 \u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2 3\u00E2\u0088\u00921.0\u00E2\u0088\u00920.50.00.51.01.5MentalizingBelief in KarmaHINDU CHRISTIAN\u00E2\u0088\u00922 0 2 \u00E2\u0088\u00922 0 2\u00E2\u0088\u00921.0\u00E2\u0088\u00920.50.00.51.01.5AnthropomorphismBelief in KarmaMoral Knowledge \u00E2\u0088\u00921 SD Mean +1 SD 89 Christians and American Hindus). Moreover, (3) mentalizing does not come to predict belief by virtue of supporting broad anthropomorphic tendencies but rather comes to support belief by means of supporting active reasoning about god\u00E2\u0080\u0099s agency. More broadly, (4) preliminary evidence suggests that mentalizing is increasingly related to belief (and anthropomorphism decreasingly so) the more supernatural targets are attributed with moral knowledge. Given that mentalizing is narrowly deployed to imbue and reason about the agentic mental qualities of god, I next examined whether reasoning about god\u00E2\u0080\u0099s agentic qualities (i.e., god\u00E2\u0080\u0099s capacities to remember things, and make plans) at least in part supports inferences about god\u00E2\u0080\u0099s moral knowledge \u00E2\u0080\u0093 which in having consequences for believers provides the impetus for believing in and maintaining a personal relationship with god. That is, in one final set of exploratory analyses, I examined whether the relationship between attributing god with agency and belief in a personal god is itself mediated by moral knowledge attributions to god. The proposed mediation models are summarized in Figure 3.14 and are strikingly similar in both samples (full mediation models are presented in the appendices, Table S3.2 and Table S3.3). Indeed, all estimated paths and the estimated indirect effects were very closely matched in the two samples. In support of this post-hoc hypothesizing, moral knowledge attributions accounted for 61% and 55% of the relationship between agency attributions and belief in god in the Hindu and Christian samples respectively. 90 Figure 3.14 The mediating role of moral knowledge attributions on the relationship between agency and belief in a personal god 91 General Discussion In three studies and five samples, mentalizing was found to positively predict belief in god to similar degrees. More specifically, belief in a personal god was found to be most strongly predicted by mentalizing. Moreover, the contributions of mentalizing to belief in a personal god were robust to additional controls for cultural exposure to religion (i.e., religious upbringing). Importantly, these cognitive and cultural predictors of belief did not interact; suggesting that they are unique effects. Given some examples of observed null effects between mentalizing and belief (e.g., Jack et al., 2016; Maij et al., 2017), a closer examination of the differences between these published findings and the current studies is mandated. Statistically, the reporting in studies where a null-effect of mentalizing is observed often conflate lack of statistical significance with the absence of an effect (even when confidence intervals for the estimated effect are just shy of excluding zero and extend clearly in the hypothesized direction). Methodologically, belief is often conflated with religious practice (and specifically church attendance) in forming an index of generalized \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 \u00E2\u0080\u0093 even though theory in the cognitive science of religion explicitly says nothing about how mentalizing should account for religious practices \u00E2\u0080\u0093 these theories are about supernatural agent beliefs. This latter decision potentially suppresses the magnitude of the underlying relationship between mentalizing and belief in these datasets. Furthermore and in spite of the null-effects of mentalizing presented by Maij et al. (2017) specifically, these authors do raise the very important point that the contributions of mentalizing should be considered in light of other factors known to contribute to \u00E2\u0080\u0098religiosity\u00E2\u0080\u0099 (Gervais et al., 2011). While amongst the Dutch cultural exposure to religion may be the best explanatory predictor of religiosity, the current data suggest that this is not the case in all religious contexts. 92 While the results presented here provide an example of when mentalizing does continue to predict belief even after controlling for cultural exposure to religion \u00E2\u0080\u0093 the observed variability in the current studies calls for further investigations of when and how \u00E2\u0080\u0098culture\u00E2\u0080\u0099 and \u00E2\u0080\u0098cognition\u00E2\u0080\u0099 come to shape religious beliefs. Accounting for the contributions of mentalizing to belief How then does mentalizing come to foster beliefs that one has a personal connection with god? In three studies, three theoretical models of how mentalizing comes to be implicated in belief in god were assessed and are discussed in turn: (1) the broad anthropomorphism hypothesis; (2) the narrow anthropomorphism hypothesis, and (3) the anthropocentric knowledge hypothesis. To reiterate, the broad anthropomorphism hypothesis holds that mentalizing promotes belief because capacities for reasoning about human minds are overextended and applied to the world writ-large. Under this model, the frequent perception of human-like mental states detached from humans promotes belief in the existence of disembodied minds. And thus, I examined whether generalized anthropomorphism (i.e., the tendency to attribute mental states to non-human targets) mediated the observed relationship between mentalizing and belief. In five samples, no support for this mediation model was found. Indeed, mentalizing was found to be largely unrelated to generalized anthropomorphism. And thus while the process of attributing mental states to non-human targets may require some base-level capacity for reasoning about other minds (Severson & Lemm, 2016; Waytz et al., 2014) \u00E2\u0080\u0093 any further relationship seems to be negligible, at least in the current samples. This latter result emphasizes a disconnection between mental state reasoning and mental state perception and reflects a more nuanced understanding of mentalizing and related cognitions as a less monolithic system than the extant literature might suggest (as has been argued elsewhere, e.g., Lindeman & Lipsanen, 2016b; Schaafsma, Pfaff, Spunt, & Adolphs, 2015). 93 Furthermore, anthropomorphism was found to either be unrelated to belief in god \u00E2\u0080\u0093 or negatively related to belief in god amongst Christians. This latter effect further supports the conclusions of Willard & Norenzayan (2013) that Christianity may actively suppress generalized anthropomorphic tendencies amongst believers. Importantly, however, anthropomorphism did not predict belief in god in the Hindu sample either \u00E2\u0080\u0093 despite explicit and rampant anthropomorphism in the Hindu religious tradition. And thus, the lack of support for the broad anthropomorphism hypothesis cannot be explained away by the explicit cultural suppression of anthropomorphism in some religious contexts (i.e., Christianity \u00E2\u0080\u0093 where this effect has been previously examined). Importantly, just because broad anthropomorphism does not predict belief in a personal god, does not mean that anthropomorphism is necessarily unrelated to other types of supernatural targets. The narrow anthropomorphism hypothesis \u00E2\u0080\u0093 however \u00E2\u0080\u0093 was well supported as the relationship between mentalizing and belief in god was found to be mediated (to mildly varying extents) by the specific anthropomorphism of god\u00E2\u0080\u0099s mind in all samples. This model holds that mentalizing is deployed not to anthropomorphize the world writ-large, but specifically god\u00E2\u0080\u0099s mind. In one version of this hypothesis, the specific anthropomorphism of god\u00E2\u0080\u0099s mind makes god\u00E2\u0080\u0099s mind more intuitively grasped, thus in turn promoting belief by making it easier (i.e., more intuitive) to believe in god. However, in Studies 2A and 2B, I examined whether analytical thinking (as measured by the cognitive reflections test) negatively predicted attributing mental states to god \u00E2\u0080\u0093 and observed only a weak correlation (albeit in the hypothesized direction) amongst Hindu participants. Therefore, despite finding clear support for this model\u00E2\u0080\u0099s account of the relationships between mentalizing and belief across samples \u00E2\u0080\u0093 additional theorizing and interpretation of additional analyses is necessary to account for why this model was supported. 94 In what may seem (at first) like a hit to the narrow anthropomorphism account, the anthropocentric knowledge hypothesis was also well supported across all samples. In this account, it is not mental state attribution to god that promotes belief but rather when faced with an otherwise acquired belief in god, individuals are motivated to deploy what they know about human minds in order to make sense of this otherwise unknowable supernatural mind. This rationale follows from theory suggesting that anthropomorphism (not just to God) is increasingly deployed when there is an impetus to make sense of a non-human agent (Epley et al., 2007). This model, however, fails to account for the specific ways in which god\u00E2\u0080\u0099s mind is anthropomorphized. Indeed, additional analyses of the data in Studies 2A and 2B demonstrated that it is not the attribution of all human mental states that supported belief in god; but specifically, agency-related mental states. Although god was not denied experience in either sample, attributions of experience were negatively related to belief in both. The anthropocentric knowledge account as is currently posited cannot account for why experience \u00E2\u0080\u0093 a perfectly common and important dimension of human mental states \u00E2\u0080\u0093 does not predict belief. Past research also points to some limitations of the anthropocentric knowledge account. For example, in a test of the development of nonhuman agent concepts, Barrett, Richert, & Driesenga (2001) tested whether children anthropocentrically attribute false-beliefs equally to all types of social agents. Their results demonstrate that even 3-year old American children tend to differentially attribute God, animals with supernatural powers, and other humans with false-beliefs. And thus, even for young children (who presumably \u00E2\u0080\u009Cknow\u00E2\u0080\u009D God less, and are also perhaps more motivated to apply what they know about the human mind to make sense of God\u00E2\u0080\u0099s mind) \u00E2\u0080\u0093 the anthropocentric knowledge account is not supported. 95 However, the question remains as to why agency attributions, specifically, come to mediate the relationship of mentalizing on belief in god. In an additional set of analyses, it was observed that attributions of moral knowledge to god mediated the relationship between agency attributions and belief in a personal god in both samples. This tells us something important about how mentalizing comes to be implicated in beliefs about one\u00E2\u0080\u0099s personal relationship with god. Specifically, these additional analyses demonstrate that agentic conceptualizations of god come to mediate the relationship between mentalizing and belief because agency-attributions facilitate reasoning about what god knows and cares about (i.e., god\u00E2\u0080\u0099s moral knowledge). The belief that god knows about one\u00E2\u0080\u0099s thoughts/actions and cares enough to act on this knowledge in morally congruent ways (rewarding good/punishing bad) in turn (1) provides its own impetus for maintaining a personal relationship with god while (2) stressing the importance of inferring god\u00E2\u0080\u0099s mental states because it is god\u00E2\u0080\u0099s agentic capacities (i.e., making plans, remembering things) that can have consequences for the believer. What this all suggests for the narrow anthropomorphism hypothesis is that mentalizing is not implicated in maintaining a personal relationship with god by virtue of making god more intuitively grasped \u00E2\u0080\u0093 but rather because expending more cognitive effort to reason more deeply about the mental states of god is a means to figure out what god knows, cares about, and how god is going to act on this knowledge \u00E2\u0080\u0093 which in not so many words, might be what it means to maintain a personal relationship with god in the first place. And thus, individuals who are better able to reason about human mental states are better able to engage more deeply with their beliefs about the mental states of god, and better able to maintain their personal connection to a god who is believed to reciprocally care about them. 96 The hypothesis that mentalizing is deployed to engage with supernatural minds because doing so allows for inferences about what supernatural agents know and care about was further supported by additional analyses demonstrating that moral knowledge attributions moderated the relationships between mentalizing, generalized anthropomorphism and belief in other supernatural targets. That is, mentalizing was found to be increasingly relevant to belief the more agents were attributed with moral knowledge. And interestingly, the reverse was true \u00E2\u0080\u0093 quite consistently in both samples \u00E2\u0080\u0093 in the case of generalized anthropomorphism. In the specific case of belief in karma \u00E2\u0080\u0093 the cultural differences observed in the contributions of mentalizing (positive in the Hindu sample, weak in the Christian sample) and anthropomorphism (negative in the Hindu sample, positive in the Christians sample) were mediated by cultural differences in moral knowledge attribution to karma. The effect of anthropomorphism on belief in karma in Christians was greatly reduced amongst participants who attributed karma with moral knowledge. And, for Christians who attributed moral knowledge to karma \u00E2\u0080\u0093 the contributions of mentalizing to their belief in karma were comparable to that of the average contribution of mentalizing to belief in karma amongst Hindus (who attributed more moral knowledge, on average, to karma). Taken together, these results suggest that effortful mental state reasoning predicts belief when doing so has consequences (i.e., the supernatural agent is believed to know things and can act on this knowledge; and thus, figuring out that knowledge becomes relevant). Otherwise, when agents are believed but not mentalized in a similar sense, generalized anthropomorphic tendencies do indeed predict belief. More broadly, this might explain why anthropomorphism does not but mentalizing does predict belief in god \u00E2\u0080\u0093 i.e., because god is highly moralized. Limitations and Future Directions 97 The work presented in this chapter speaks to the replicability of the observed relationship between mentalizing and belief in five samples. Moreover, this work delineates the boundary conditions of the effects of mentalizing on belief and moves beyond the detection of the effect in attempts to account for the process by which mentalizing comes to support beliefs. Although this work stands to contribute substantially to the understanding of how mentalizing comes to be related to belief in the cognitive sciences of religion \u00E2\u0080\u0093 this work is limited by the extent to which causality can be inferred from this type of cross-sectional correlational data. Moreover, there is a mounting dissatisfaction with mediation analyses as applied to cross-sectional correlational data as has been conducted here (e.g., see (Grice, Cohn, Ramsey, & Chaney, 2015; Kline, 2015; Thoemmes, 2015). These critics have targeted the methods for assessing mediation delineated in (Baron & Kenny, 1986) that with over 75 000 citations, has been an integral part of the statistical tool-kit of social scientists for three decades. The harshest of these critics posit that without experimental manipulation of the independent and mediating variables \u00E2\u0080\u0093 mediation analyses should not be conducted (in following with Spencer, Zanna, & Fong, 2005). In presenting the current analyses, a number of the criticisms have been taken on board. Specifically, I did not select between models using reverse causation models but rather used additional theoretically-justified models to test the limits of these mediation accounts. Furthermore, statistical significance of direct and mediated paths in the models was not used to determine the extent of the mediation. Indeed, the presence of a mediating effect was identified solely by the presence of an indirect effect and its associated bootstrapped confidence interval \u00E2\u0080\u0093 in following with the advice laid out by these critics. Nonetheless, more experimental work in the cognitive science of religion is certainly necessary. 98 When it comes to disentangling the contributions of \u00E2\u0080\u0098culture\u00E2\u0080\u0099 and \u00E2\u0080\u0098cognition\u00E2\u0080\u0099 to religious beliefs \u00E2\u0080\u0093 the world is abounding with natural experiments ripe for examination. For example, an assessment of the relative cultural and cognitive predictors in religious converts (between religions, or from non-religious to religious) as compared to those who have inherited a life of belief from their family may prove particularly fruitful in unpacking the cultural and cognitive contributions to beliefs. And although this work has taken a slightly cross-cultural approach by testing its hypotheses in distinct religious contexts \u00E2\u0080\u0093 there remains an immense and mostly-untapped pool of religious diversity in the world. The question then of when and in what ways mentalizing comes to predict belief in supernatural agents would truly benefit from a vastly more cross-cultural approach. Indeed, this approach can only benefit the cognitive science of religion. In the current work, specific conclusions about the contributions of mentalizing to belief in relation to moral knowledge attribution would not have been possible without the cross-cultural comparisons of Hindu and Christian beliefs in karma. Furthermore, we have come to know increasingly more about the extant variability in what god(s) and other supernatural agents care about in different religious traditions (Purzycki & Sosis, 2011; Purzycki, 2013a, 2013b; Purzycki et al., 2016). These works have laid the foundations for what could be a richly informative systematic cross-cultural investigation into how mentalizing\u00E2\u0080\u0099s relationship to belief covaries with god(s) concerns and knowledge. 99 Chapter 4 A cross-cultural examination of mentalized deity concepts and belief Introduction Longstanding theory in the cognitive sciences of religion holds that supernatural agents are represented in the minds of those that believe in them by means of ordinary social cognitive processes for mentalizing (e.g., Atran & Norenzayan, 2004; Boyer, 2001; Guthrie et al., 1980; see also Chapter 3 of this dissertation). And indeed, despite diversity in the forms supernatural agents take across cultures, examinations of deities do suggest that they are typically believed to have human-like mental states (Purzycki & Sosis, 2011; Purzycki, 2013a). However, to date there exists no systematic cross-cultural examination of mentalistic deity beliefs. To begin filling this gap, the current study employs an immensely diverse sample to assess the frequency with which and when individuals mentalize (i.e., think about the minds of) their deities. With notable exceptions, research about religion in the psychological sciences has long equated \u00E2\u0080\u0098religion\u00E2\u0080\u0099 with Christianity - and particularly the North American and European varieties, ignoring even the burgeoning Christendom in the South America, Africa and many parts of Asia (Pew Research Center, 2015). Although this WEIRD problem (Henrich et al., 2010) is pervasive in the psychological sciences \u00E2\u0080\u0093 it is particularly striking in the study of religion given that we already know there to be immense diversity in religious beliefs and practices across (and even within) cultures, denominations, temples, churches, etc. (e.g., Armstrong, 1994; Bellah, 2017; Luhrmann, 2012; Purzycki, 2013; Rappaport, 1999; and for a discussion of religious diversity see Norenzayan, 2016). Existing research that aims to assess religiosity across cultures has overwhelmingly focused on world religions in nation states, country-level relationships and rarely examines diversity in the content of religious beliefs (e.g., Norris & Inglehart, 2011). The current study provides a systematic examination of the reported religious beliefs of individuals living in 100 small- to large-scale societies and adhering to diverse religious traditions. No single contribution to the literature can realistically solve this WEIRD problem, but the work presented in this chapter stems from a large-scale effort designed to take it more seriously \u00E2\u0080\u0093 and thus stands to contribute to the generalizability of the growing account of how mentalizing is implicated in religious beliefs on a more global scale. Building on the evidence presented in Chapter 3 of this dissertation, this study provides cross-cultural tests of (1) the extent to which mentalizing deities contributes to belief and (2) the extent to which deities are mentalized the more they are moralized. In Chapter 3 of this dissertation, it was demonstrated that although individual differences in mentalizing predict the extent to which individuals believe in personal and moralized supernatural agents \u00E2\u0080\u0093 it is active reasoning about a deity\u00E2\u0080\u0099s mental states that more strongly supports belief. The current study cannot test the extent to which individual differences in mentalizing contribute to belief, but it does provide a test of the second half of this proposed account. That is, to what extent does mentalizing the minds of gods support belief across cultures? The strengths of the dataset employed here stem not just from its cross-cultural reach, but also its examination of two types of deities within each population. Specifically, these data speak to beliefs and conceptualizations of deities that differ in the extent to which they are attributed with concerns for interpersonal social behaviours. Within and across cultures, there exists substantial variability in the believed concerns of supernatural (Purzycki, 2013a; Purzycki & McNamara, 2016). In the current study, the observed variation in deities\u00E2\u0080\u0099 concerns is exploited to build on the evidence presented in Chapter 3 of this dissertation regarding what kinds of deities are mentalized. Specifically, this study tests the hypothesis that cross-culturally deities are increasingly mentalized the more they are moralized. That is, if deities are believed to be largely unconcerned with one\u00E2\u0080\u0099s behavior \u00E2\u0080\u0093 there is little reason for believers to actively consider the 101 mental states of gods and other supernatural agents (as there are little consequences for not doing so). The dataset from which focal variables examined in this chapter were selected is the product of the Evolution of Religion and Morality Project (Purzycki et al., 2016a, 2016b; Purzycki, Henrich, et al., 2018). This project\u00E2\u0080\u0099s primary goals were to cross-culturally assess the contributions of religious beliefs to moral decision-making (i.e., not cheating; and anonymous giving in economic games). This project generated a wealth of additional data from post-experiment interviews with participants. Here, this data is used to test the generalizability of important questions in the psychology of religion: (1) To what extent are deities mentalized across cultures? (2) To what extent does conceptualizing deities as mentalistic contribute to belief across cultures? (3) To what extent does the attribution of moral knowledge to deities contribute to their mentalization? Methods Sample The working data for this project was culled from the larger dataset resulting from the Evolution of Religion and Morality project (Purzycki et al., 2016b, 2016a). Across two waves of data collection, this project sampled 2228 participants from 15 diverse populations. Note, however, that analyses presented here exclude all data collected from the Hadza as responses to focal items were not measured in the same way as at the other sites (resulting in 14 sites). Demographic summaries by site are presented in Table S4.1 (see appendices). Measures 102 Deity selection Target deities were determined during pre-test interviews at each site (deities are listed by site in Table 4.1). During these interviews, participants at each site were asked to free-list up to five deities, to rank these listed deities in order of their importance; as well as to rate the extent of these deities\u00E2\u0080\u0099 knowledge, and how punitive/rewarding they are believed to be. From these ratings, a \u00E2\u0080\u0098moralistic\u00E2\u0080\u0099 deity (i.e., one that was high in these qualities) and a \u00E2\u0080\u0098local\u00E2\u0080\u0099 deity (i.e., one that remained salient across participants but was rated lower in knowledge, power, moral concern) were selected at each site. Extensive post-test analyses of the selected deities and their believed attributes suggest that participants did indeed distinguish between these deities along the planned dimensions (for more details, see Purzycki et al., 2016, 2018). At the majority of the sites, the salient moralistic deity was the Christian God. At predominantly Hindu sites (i.e., Lovu, Mauritius, and Mysore), Shiva was selected. At the Inland Tanna, and Tyva sites, the determined site-specific deities were determined to be Haine, Kalpapan, and Buddha Burgan. The identities of the local deities were more varied (see Table 4.1). At the Huatasani and Kananga sites, participants were largely unfamiliar with the local deities identified during pre-tests (Apus and Kadima, respectively) and thus, some participants were asked about different deities (Christian saints and local spirits) at test. At the Lovu and Samburu sites, no local deity was identified \u00E2\u0080\u0093 and thus relevant items discussed below about a \u00E2\u0080\u0098local\u00E2\u0080\u0099 deity for these sites were not asked. 103 Table 4.1 Selected deities and belief items summaries by site Culture Moralistic Deity Belief (n) Mentalistic Belief M (SD) Local Deity Belief (n) Mentalistic Belief M (SD) Yes No Yes No Cachoeira Christian God --- --- .94 (.15) Ogum --- --- . 41(.38) Coastal Tanna Christian God 124 3 .80 (.22) Garden spirit 69 56 .25 (.29) Huatasani Christian God 91 1 .78 (.27) Apus/saints 69 21 .31 (.29) Inland Tanna Kalpapan 9 0 .76 (.33) Garden spirit 9 0 .75 (.32) Kananga Christian God 197 3 .91 (.16) Kadima/local spirit 200 0 .07 (.16) Lovu Shiva --- --- .36 (.17) --- --- --- --- Marajo Christian God --- --- .91 (.17) St. Mary --- --- .65 (.37) Mauritius Shiva 145 4 .69 (.27) A spirit 70 79 .21 (.26) Mysore Shiva 154 11 .64 (.29) Chamundeschwari 153 12 .63 (.31) Samburu Christian God 40 0 .86 (.21) --- --- --- --- Sursurunga Christian God 162 0 .77 (.17) Forest spirit 144 17 .03 (.08) Turkana Christian God 235 1 .89 (.16) Ancestor spirit 94 142 .17 (.22) Tyva Republic Buddha --- --- .62 (.30) Spirit masters --- --- .42 (.29) Yasawa Christian God --- --- .78 (.19) Ancestor spirits --- --- .16 (.05) Total 1157 23 .78 (.25) 605 530 .29 (.34) 104 Belief At Wave II sites, participants were explicitly asked whether they believed (1 = yes) or did not believe (0 = no) in the targeted local and moralistic deity. Note, that, this item was not asked to participants the Cachoeria site. Mentalistic deity concepts In both waves, participants were asked, \u00E2\u0080\u009Chow often do you think about [moralistic/local deity]?\u00E2\u0080\u009D, and \u00E2\u0080\u009Chow frequently do you worry about [moralistic/local deity] thinks about you?\u00E2\u0080\u009D. These items were variably correlated across sites (Moralistic deity items: r = -.16 to .61; Local deity items: r = -.19 to .69). In both cases, the negative correlations between these items were amongst Yasawans. These latter two items will serve as an imperfect index of the extent to which participants think about these deities mentalistically (i.e., thinking about what another agent thinks about you requires some capacity for reasoning about mental statess). In Wave I, responses to these frequency items were coded on the following scale (Scale 1: 0 = very rarely/never; 1 = a few times a year; 2 = a few times per month; 3 = a few times per week; 4 = every day or multiple times per day) except at the Lovu, Mauritius and Yasawa sites where the following response scale was employed (Scale 2: 0 = never; 1 = sometimes, 2 = frequently or often, 3 = always or all the time). In Wave II, response scale 1 was used for the \u00E2\u0080\u0098think\u00E2\u0080\u0099 items and scale 2 was used for the \u00E2\u0080\u0098worry\u00E2\u0080\u0099 item except at the Sursurunga and Cachoeira sites were scale 1 was employed for both items. The use of these varying response scales across waves and sites is problematic for comparison, and thus all responses were rescaled by dividing participants responses by the maximum possible scale value they could have chosen. These rescaled scores were then dichotomized to indicate no/low scores (=0; responses < site-specific mean) or high scores (=1; responses > site-specific mean). Summary statistics by site and deity are presented in Table 4.1. 105 Moral knowledge At all sites, participants responded to four items that taken together serve as an index of the extent to which the moralistic and local deities are attributed with capacities for knowledge (i.e., they are believed to know things) and punishment (i.e., they are believed to wield influence over individuals lives). Specifically, participants were asked (1 = yes; 0 = no): 1. Can [moralistic/local deity] see into people\u00E2\u0080\u0099s hearts or know their thoughts and feelings? 2. Can [moralistic/local deity] see what people are doing if they are far away, in [distant town or city familiar to participants]? 3. Does [moralistic/local deity] ever punish people for their behaviour? 4. Can [moralistic/local deity] influence what happens to people after they die? Participants dichotomous responses to these four items were summed to create an index of moral knowledge attributions for each deity (Across site reliability: Moralistic deity a ~ .60; Local deity a ~ .72). Mean moral knowledge attributions at each site are presented in Table 4.2. 106 Table 4.2 Means and 95% confidence intervals for the moral knowledge index by deity and site Moral Knowledge Index Moralistic Deity Local Deity Cachoeira 2.38 [2.21, 2.54] 1.27 [1.10, 1.42] Coastal Tanna 3.19 [3.03, 3.35] 1.21 [1.03, 1.40] Hutatasani 3.11 [2.89, 3.33] 1.71 [1.45, 1.98] Inland Tanna 1.95 [1.63, 2.26] 1.91 [1.61, 2.22] Kananga 3.57 [3.46, 3.68] .71 [.56, .85] Lovu 3.63 [3.49, 3.77] ---1 Mauritius 2.99 [2.86, 3.11] 1.80 [1.62, 1.97] Mysore 3.03 [2.84, 3.22] 2.99 [2.81, 3.16] Maraj\u00C3\u00B3 3.31 [3.07, 3.56] 2.17 [1.90, 2.55] Samburu 3.68 [3.53, 3.82] ---1 Sursurunga 3.96 [3.91, 4.01] 1.37 [1.21, 1.53] Turkana 3.81 [3.71, 3.91] .60 [.49, .71] Tyva Republic 3.12 [2.89, 3.36] 2.98 [2.74, 3.21] Yasawa 3.01 [2.88, 3.14] .93 [.88, .99] Total 3.05 [3.00, 3.10] 1.32 [1.26, 1.38] Notes: 1Local deity information was not collected at these sites. Analytical decisions and methods For all results that follow, participants that were missing data on any of the included variables in the analysis were excluded (list-wise deletion). Unless otherwise noted, all non-categorical predictors were standardized within site. This allows for simpler interpretations of model intercepts when the interest is in interpreting the person-level effects of the variables (Enders & Tofighi, 2007). All analyses were conducted in R (R Core Team, 2017). Mixed effect regression models were assessed with the lme4 package (Bates, M\u00C3\u00A4chler, Bolker, & Walker, 2015). Summary tables and plots were generated with sjPlot (L\u00C3\u00BCdecke, 2018). Results Mentalistic deity beliefs across cultures Mentalistic deity beliefs (i.e., thinking and worrying about what a deity thinks) tended to be more strongly and consistently endorsed for moralistic deities than the local deities, especially 107 at Christian sites (see Figure 4.1; Christian sites are identified with a cross). Moreover, local deity mentalistic beliefs were markedly reduced in the Christian sites compared to mentalistic beliefs about local deities at the other sites where the deities were less distinguished (with the exception of Mauritius). At the non-Christian sites (e.g., Inland Tanna, Mysore, Tyva Republic), the endorsement of mentalistic deity beliefs was more diffuse for both deities. 108 Figure 4.1 Density plots of mentalistic belief (mean \u00E2\u0080\u0098think\u00E2\u0080\u0099 and \u00E2\u0080\u0098worry\u00E2\u0080\u0099) composite by site and deity Notes: Christian sites are identified with a cross. Local deity beliefs were not assessed at the Lovu and Samburu sites. Flatter density curves indicate more diffuse responses (i.e., greater variability). \u00E2\u0080\u00A0\u00E2\u0080\u00A0\u00E2\u0080\u00A0\u00E2\u0080\u00A0\u00E2\u0080\u00A0\u00E2\u0080\u00A0\u00E2\u0080\u00A0\u00E2\u0080\u00A0 \u00E2\u0080\u00A0Tyva Republic YasawaMysore Samburu Sursurunga TurkanaKananga Lovu Marajo MauritiusCachoeira Co.Tanna Huatasani In.Tanna0.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.000.00 0.25 0.50 0.75 1.00012345012345012345012345Mentalistic BeliefDensityDEITYMoralisticLocalMentalistic Belief Composite 109 To explicitly estimate the differences in beliefs between deities and traditions, mentalistic beliefs (0= no/low; 1 = high) on the stacked responses to the dichotomized \u00E2\u0080\u0098think\u00E2\u0080\u0099 and \u00E2\u0080\u0098worry\u00E2\u0080\u0099 items were modelled in a mixed-effect logistic regression with a random intercept for site and item (think/worry; see Table 4.3). In Model 1, the estimated effect of deity (1 = moralistic) reiterates what is clearly observable from the distributions of responses. That is, the odds of moralistic deities being mentalized were 3.43 to 4.21 times that of the local deities. In Model 2, the fixed effect of tradition (1= Christian; 0 = Other) and its interaction with deity were added. And in Model 3, the random effect of deity by site was also included \u00E2\u0080\u0093 which demonstrates a clearer difference between the effects of moralistic deities in Christian and non-Christian sites. These latter two models demonstrating the rather clear effects of the presence of Christianity across sites. That is, the observed difference between the extent to which moralistic and local deities are mentalized is greatly reduced in non-Christian sites (see Figure 4.2). Table 4.3 Testing for differences in mentalistic beliefs across deities and traditions Mentalistic deity beliefs (0 = no/low; 1 = high) Model 1 Model 2 Model 3 OR CI OR CI OR CI (Intercept) .61 .41 \u00E2\u0080\u0093 .89 .76 .44 \u00E2\u0080\u0093 1.31 .89 .39 \u00E2\u0080\u0093 2.01 Deity (1 = Moralistic) 3.80 3.43 \u00E2\u0080\u0093 4.21 1.77 1.50 \u00E2\u0080\u0093 2.10 1.39 .68 \u00E2\u0080\u0093 2.84 Tradition (1 = Christian) .73 .38 \u00E2\u0080\u0093 1.41 .63 .23 \u00E2\u0080\u0093 1.69 Deity * Christian 3.24 2.61 \u00E2\u0080\u0093 4.01 3.78 1.58 \u00E2\u0080\u0093 9.08 Observations 7209 7209 7209 Random Effects SITE + ITEM SITE + ITEM Deity by SITE + ITEM AIC 8956.61 8844.71 8723.26 110 Figure 4.2 Exponentiated mean estimates of tradition differences in mentalistic beliefs by deity Notes: Odds ratios are exponentiated mean estimates for each group predicted from parameters of Model 3. Error bars are 95% confidence intervals. The x-axis is on a logarithmic scale. Odds ratios > 1 indicate increased mentalistic beliefs, odds ratios <1 indicate decreased mentalistic beliefs. Predicting belief When asked to report (yes/no) whether they believed in the moralistic deity only 23 participants across the entire dataset explicitly said \u00E2\u0080\u0098no\u00E2\u0080\u0099. For the local deities, responses were more variable (see Table 4.1; although responses at Kananga and Inland Tanna site were also all \u00E2\u0080\u0098yes\u00E2\u0080\u0099 at these sites for the local deities). The cross-site consensus of belief in moralistic deities makes it unreasonable to assess what contributes to these beliefs. Thus, the contributions of mentalistic beliefs (i.e., the sum of dichotomized thinking and worrying about what a deity thinks about you) to explicit belief was assessed only for the local deities. Explicit belief in the local deities (1 = yes; 0 = no) was modelled using a mixed-effect logistic regression model with a random intercept for site. In this model, mentalizing the local NoYes0.5 1.0 2.0 3.0Odds Ratio [95% CI]Christian site? DeityMoralisticLocal 111 deities predicted a whopping 6.48 to 12.35 times increase in the odds of explicit belief (Intercept OR = .38 [.04, 3.89]). The contributions of mentalizing (i.e., thinking and worrying) the local deity in each site are presented in Figure 4.3. Overall, explicit belief was more likely at higher levels of reporting mentalistic beliefs (note, the ceiling and floor effects at the Kananga and Sursurunga sites). 112 Figure 4.3 Predicted probabilities of belief in local deities by site as predicted by mentalistic beliefs Notes: Black lines are predicted probabilities by site, gray lines (same on every plot) are the average baseline model predictions of mentalistic belief to belief in local deities. X axis values are the possible values of summed mentalistic belief. 0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Coastal Tanna0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Huatasani0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Inland Tanna0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Kananga0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Mauritius0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Mysore0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Sursurunga0.00.40.8Mentalistic BeliefProb. of Belief0 1 2Turkana 113 When are deities mentalized? It was hypothesized (in following with the results of Studies 3A and 3B in Chapter 3), that mentalistic deity beliefs should be more consistently endorsed when deities are attributed with moral knowledge. To assess the contributions of moral knowledge attributions to mentalistic deity beliefs, two mixed-effect logistic regression models were examined (see Table 4.4). The dichotomized responses (0 = no/low; 1 = high) to the \u00E2\u0080\u0098think\u00E2\u0080\u0099 and \u00E2\u0080\u0098worry\u00E2\u0080\u0099 items for each deity were stacked and identifying-factors were created (item = think/worry; deity = local/moralistic). In both models, a random-intercept for site, item and random effect of deity by site were included. In Model 1, the fixed effects of deity (1 = moralistic), the moral knowledge index (standardized), and the interaction of deity and moral knowledge were included. In Model 2, the main effect of Christianity, its interaction with moral knowledge attributions and deity, and the three-way interaction site were included to assess differences between traditions on all the included relationships. In Model 1, moralistic deities were more mentalized than local deities (OR = 1.77 [1.56, 2.01]) when controlling for moral knowledge attributions. In support of the hypothesis that moral knowledge attributions predict mentalistic beliefs - the odds of mentalizing local deities increased by 2.44 to 2.98 times with greater moral knowledge attributions. For moralistic deities, the size of this increase was reduced by 66%, but the degree to which the targeted moralistic deities were explicitly moralized by participants still increased mentalistic beliefs (OR = 1.78 [1.36, 2.29]. This latter result likely reflecting the already higher mentalization of moralistic deities at most sites. In Model 2, the moderating effect of Christianity was examined. The results of which (depicted in Figure 4.4) indicate that greater attributions of moral knowledge predicted greater mentalization of deities across traditions. At high moral knowledge attribution \u00E2\u0080\u0093 both local and 114 moralistic deities at all sites were more mentalized. Remarkably, the difference between deities in mentalistic beliefs in Christian sites were reduced at high levels of moral knowledge attributions. Indeed, Figure 4.5. demonstrates that at all sites where local and moralistic deities were differentially mentalized at low levels of moral knowledge attribution converge at high levels of mentalistic beliefs when moral knowledge attributions are high. Table 4.4 The contributions of moral knowledge attributions to mentalistic beliefs Mentalistic deity beliefs (0 = no/low; 1 = high) Model 1 Model 2 OR CI OR CI (Intercept) .92 .59 \u00E2\u0080\u0093 1.42 .92 .42 \u00E2\u0080\u0093 2.01 Moral Knowledge Index 2.70 2.44 \u00E2\u0080\u0093 2.98 2.03 1.77 \u00E2\u0080\u0093 2.33 Deity (1 = Moralistic) 1.77 1.56 \u00E2\u0080\u0093 2.01 1.41 1.18 \u00E2\u0080\u0093 1.69 Deity * Moral Knowledge .66 .56 \u00E2\u0080\u0093 .77 .94 .77 \u00E2\u0080\u0093 1.15 Christian site? (1 = Yes) 1.13 .45 \u00E2\u0080\u0093 2.84 Christian * Moral Knowledge 1.71 1.40 \u00E2\u0080\u0093 2.08 Deity * Christian 2.01 .78 \u00E2\u0080\u0093 5.17 Deity * Christian * Moral Knowledge .44 .32 \u00E2\u0080\u0093 .61 Observations 7209 7209 AIC 8208.283 8181.133 115 Figure 4.4 Estimated odds ratios of mentalistic beliefs by tradition, deity, and moral knowledge attributions Notes: Values are exponentiated mean estimates of predictions made from Model 2, conditioning on random effects of site, item and deity by site. The x-axis is on a logarithmic scale. Error bars are 95% confidence intervals. Odds ratios > 1 indicate greater mentalistic beliefs; odds ratios < 1 indicate less mentalistic beliefs. 0.31.521.042.193.63.140.450.50.920.971.871.85Christian Other traditions0.25 0.50 1.00 2.00 3.00 0.25 0.50 1.00 2.00 3.00\u00E2\u0088\u00921 SDMean+1 SDOdds Ratio [95% CI]Moral Knowledge IndexDeityMoralisticLocal 116 Figure 4.5 Predicted probabilities of mentalistic beliefs by site, deity and moral knowledge attributions Notes: Blue curves are predicted probabilities for local deities, red curves are predicted probabilities for moralistic deities. The gray curve (same on every plot) is the average predicted contributions of moral knowledge to mentalistic beliefs across sites and deities. Plots were generated from Model 2 estimates (Table 4.4). 0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Cachoeira0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Coastal Tanna0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Huatasani0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Inland Tanna0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Kananga0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Lovu0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Marajo0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Mauritius0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Mysore0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Samburu0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Sursurunga0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Turkana0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Tyva Republic0.00.20.40.60.81.0Moral Knowledge IndexProb. of Mentalistic Belief\u00E2\u0088\u00922 \u00E2\u0088\u00921 0 1 2Yasawa 117 Discussion Psychological inquiry into the cognitive foundations of religion has focused a great deal on the contributions of mental state reasoning to belief (e.g., Bloom, 2006; Greenway, 2016; Jack, Friedman, Boyatzis, & Taylor, 2016; Kelemen, 2004; Maij et al., 2017; Norenzayan, Gervais, & Trzesniewski, 2012; Vonk & Pitzen, 2016; Willard & Cingl, 2017; and Chapter 3 of this dissertation). Moreover, this literature has overwhelmingly focused on the contributions of mentalizing to belief amongst Christians. Although often implicated in theories of the ultimate origins of religion (e.g., Boyer, 2001; Guthrie et al., 1980; Norenzayan et al., 2016), there has yet to be a systematic investigation of the contributions of individual differences in capacities for mentalizing to belief across cultures. In this chapter, a diverse data set was employed to assess the cross-cultural prevalence of mentalistic deity concepts, their antecedents, and contributions to belief in the types of populations otherwise ignored in the literature. Specifically, the mentalization of the mind of gods was assessed by the frequency with which participants thought about deities and worried about what deities thought about them. Although these are not measures of capacities for mentalizing per se, thinking about what a deity thinks (i.e., second-order mental state reasoning) certainly requires some capacities for thinking about minds (Baron-Cohen, Leslie, & Frith, 1985). And thus, while not specifically filling the gap in the literature of how individual differences in mentalizing contribute to religious belief across cultures, this work does assess the antecedents and correlates of mentalistic deity beliefs across cultures, and in non-Christian samples. On this last point, results indicate that the Christian God was by far the most consistently mentalized deity in this sample. The other moralistic deities examined (Shiva, Buddha Burgan and Kalpapan) were not mentalized to the same extent as the Christian God, but still more so than the 118 examined local deities. Moreover, local deities at Christian sites were consistently the least mentalized. This striking difference at Christian sites could be indicative of the extent to which the presence of Christianity may actively suppress this type of mentalistic engagement with traditional deities. Or rather, that even if local deities are believed to wield influence over one\u00E2\u0080\u0099s life, it is the Christian God that really matters (when present), and is worth thinking and worrying about. At Yasawa, for example, the local Kalou-vu are believed to punish norm violations by causing misfortune, illness and even death. But yet, participants reported more frequent mentalistic engagement with the Christian God \u00E2\u0080\u0093 perhaps because they believe that the Christian God can protect against these local spirits (McNamara & Henrich, 2017). What then predicts whether deities are mentalized? The results presented here suggest that deities are increasingly mentalized the more they are moralized (i.e., deities are believed to be knowledgeable about thoughts and behaviours, and powerful enough to do something about this knowledge). Indeed, the reported difference between deities at Christian sites was found to be greatly reduced when local deities were attributed with moral knowledge. At the Hindu and Buddhist sites, mentalistic beliefs were closely related to moral knowledge attributions for both examined deities. Put differently, believing that a deity (1) knows about one\u00E2\u0080\u0099s thoughts and behaviours, and (2) cares enough about them to exert influence over individual lives (in this life, or the afterlife) makes it worthwhile (and perhaps necessary) to exert the cognitive effort to think and worry about the content of a deity\u00E2\u0080\u0099s mind because the contents of this moralized supernatural mind can have consequences. Put simply, this provides evidence for a cross-cultural mentalization bias of moralized supernatural agents. Purzycki (2013) demonstrated that despite explicitly endorsing the Christian God\u00E2\u0080\u0099s omniscience, American Christians were more likely to indicate that God knows more moral 119 information (e.g., if the participant was generous to someone) than non-moral information (e.g., the color of the participant\u00E2\u0080\u0099s shirt). And that, despite the lack of explicit concern about morality in cultural models of Tyvan spirit-masters; Tyvan Buddhists and Shamanists similarly privileged moral information in reasoning about what it is that spirit-masters know. These results suggesting a cross-cultural moralization bias of supernatural minds. In the current sample, Tyvan spirit-masters were increasingly mentalized the more they were explicitly moralized (as was the case for deities at all other sites). Taken together, thinking about any supernatural agents may intuitively cue moral reasoning (Purzycki, 2013); and in turn, explicit beliefs about an agent\u00E2\u0080\u0099s moralized concerns may encourage more mentalistic reasoning about that deity \u00E2\u0080\u0093 a potentially mutually reinforcing complex of cognitions. Limitations and future directions The cross-cultural approach undertaken in this chapter is still correlational, and cross-sectional. And as an important caveat, the data presented here isn\u00E2\u0080\u0099t necessarily representative of responses in the broader communities from which participants were sampled. That is, with the exception of the Inland Tanna site where the entire community was sampled. Indeed, sampling methods were mixed across field sites with some sites drawing participants from places of religious worship, others randomly asking participants on the street, others going door to door throughout specific neighbourhoods. Importantly, these sampling methods may have differentially restricted the range of observed religious beliefs (i.e., sampling at a place of religious worship is likely to draw from a population of relatively committed individuals). Although this study provides evidence for the prevalence and correlates of mentalistic deity concepts, the current data cannot speak directly to the question of how individual differences in the capacities for mentalizing come to be implicated in belief across cultures. In Chapter 3 of this 120 dissertation, it was demonstrated that individual differences in mentalizing\u00E2\u0080\u0099s contributions to belief was mediated by the active deployment of one\u00E2\u0080\u0099s mentalizing in reasoning about god\u00E2\u0080\u0099s mind. Here, only the secondary path was assessed. And thus, an important next step for accounting for the relationship between mentalizing and belief would be to assess \u00E2\u0080\u0093 cross-culturally \u00E2\u0080\u0093 the extent to which individual differences in mentalizing contribute to belief. In doing so in tandem with measuring the content of deities\u00E2\u0080\u0099 concerns, the cognitive sciences of religion stands to benefit from a deeper synthesis of how psychological processes and cultural knowledge interact in sustaining religious beliefs. 121 Chapter 5 Conclusion Throughout this dissertation, I have provided evidence and discussed the ways in which longstanding and leading theories in the cognitive science of religion \u00E2\u0080\u0093 as they are currently posited - cannot account for all the observed data. Here, I further contextualize the focal findings presented throughout this dissertation by taking a broader look at what the evidence collectively suggests for theories about the psychological foundations of religious beliefs. Specifically, I review what the current results suggest about how (1) analytical thinking and (2) mentalizing come to be implicated in belief, and (3) what taking a cross-cultural approach provides in furthering our accounts of how cultural variations in supernatural agent beliefs are psychologically represented in believers\u00E2\u0080\u0099 minds. In doing so, I point to places in existing theoretical models of how religious beliefs are shaped and sustained by human psychology that require some further specification. Lastly, I discuss the limitations of this work, and propose pertinent next steps in building on the evidence provided throughout this dissertation. What does analytical thinking have to do with belief? In Chapter 2, three models of how analytical thinking comes to be implicated in belief in god were assessed. Specifically, I tested in a large sample of undergraduate students at a Canadian university, Americans and Indians the replicability of the core prediction of the dual process model of belief that analytical thinking is robustly negatively related to belief in god. In all examined samples, a negative correlation between analytical thinking and belief in god was observed, as was the expected positive correlation of reliance on intuition and belief. The consistency in the direction of these relationships observed here, and elsewhere in the literature (e.g., Pennycook, Ross, et al., 2016) does suggest that analytical thinkers are less prone to endorsing belief in god. However, the consistency in the direction of these relationships tell us very little about what it is 122 about analytical thinking that leads to these differences in the first place. Moreover, the magnitude of these relationships was quite variable in different samples. Indeed, many of these relationships were lower amongst Canadians and undergraduate students than they were amongst Americans and Indians. This variability cannot be directly accounted for by the dual process model of belief as is currently posited. Strikingly, it is also the case that the effects of faith in intuition and analytical thinking in predicting belief in god were moderated (although weakly) by political orientation. Amongst conservatives, analytical thinking was largely unrelated to belief. These latter results are more in line with a weak version of the expressive rationality model of belief that (1) expects this variability and (2) accounts for it as a result of the precise moderation that was demonstrated here. Future research should specifically target individuals across the political spectrum to more stringently test the core predictions of the expressive rationality model as it pertains to belief in god. The expressive rationality model assumes that analytical thinking is employed to confirm one\u00E2\u0080\u0099s already held beliefs in an identity-protective manner. That being said, the data presented in this dissertation cannot speak to whether or not this mechanism is indeed the right explanatory model for this effect. However, it could be posited that in light of dwindling religiosity across Canada (Pew Research Center, 2013) and the potential \u00E2\u0080\u0098threat\u00E2\u0080\u0099 of being religious in a place of higher learning may be driving religious analytical thinkers to find new ways to justify their beliefs. The smaller effects observed in the Canadian and student samples are also closer to the effect size reported by Gervais et al. (2017)\u00E2\u0080\u0099s cross-cultural examination of the relationship between analytical thinking and belief in god. A clear next step forward would be to re-examine their published data for evidence of the moderating effect of political orientation. However, Gervais et al. (2017) also propose that analytical thinking may not be suppressing cognitive 123 intuitions but rather driving individuals to question the norms of their societies of which religion and its associated beliefs and practices are included. Here, I found no support for the counter-normative rationality model of belief. Taken together, these results indicate that analytical thinking is not always negatively related to belief in god. However, more work is certainly needed to specify precisely when and how analytical thinking comes to be related to belief. Importantly, it is likely the case that the key to deciphering with precision both when and how analytical thinking comes to be related to supernatural beliefs is to continue examining these questions across increasingly diverse cultural and religious contexts. Indeed, the identification of boundary conditions across cultures of when and how intuition comes to support, and analytical thinking comes to erode religious beliefs can help turn the intellectual discourse away from whether the effect exists or not to one that accounts for why. What does mentalizing have to do with belief? The dual process model of belief assumes that (1) religious cognition is grounded in intuitive cognitive processes and (2) that broad tendencies for overriding these intuitions (i.e., analytical thinking styles) leads to the erosion of belief by virtue of overriding the pull of one\u00E2\u0080\u0099s intuitions (e.g., Pennycook, Ross, et al., 2016). However, most of the work that has been generated thus far in support of the dual process model of belief has overtly relied on demonstrating that analytical thinking is related to not believing without a careful consideration of what kinds of intuitions support belief or how analytical thinking comes to override them. It is, however, often assumed that analytical thinking will limit the extent to which intuitions for mentalizing, for example, support belief. More specifically, it is often argued in line with the broad anthropomorphism hypothesis that overriding one\u00E2\u0080\u0099s intuitions for mentalizing prevents the over- 124 attribution of mental states to the world writ-large \u00E2\u0080\u0093 and thus leads to a reduction in belief. The results of Chapter 3, however, demonstrate that this is a particularly poor model of how mentalizing comes to be related to belief in god in the first place. That is, broad anthropomorphic tendencies were largely unrelated (and even weakly negatively) related to individual differences in predominantly Christian and Hindu samples. More importantly, broad anthropomorphism was unrelated to belief in god in all samples. Furthermore, the assumption that analytical thinking erodes mentalistic \u00E2\u0080\u0098biases\u00E2\u0080\u0099 is at odds with recent work demonstrating that mentalizing (i.e., the active process of inferring the mental states of others) is actually cognitively demanding and not-intuitive1 (e.g., Apperly & Butterfill, 2009). Indeed, analytical thinkers are more likely to deploy the necessary effort to mentalize the minds of other people (Ma-Kellams & Lerner, 2016). On average, humans are so restrictive in their willingness to engage with the minds of others that even other humans are often denied mental states when they certainly have them (e.g., members of social out-groups; Harris & Fiske, 2006); and it takes considerable effort to think about the perspectives of others in maintaining even some of the most motivating interpersonal relationships (e.g., with romantic partners; Davis & Oathout, 1 Of course, there are other intuitions that support religious beliefs and are simultaneously eroded by analytical thinking (i.e., teleological thinking; Kelemen, Rottman, & Seston, 2013). And thus, it is entirely plausible that the contributions of mentalizing to belief and whatever intuitions are eroded by analytical thinking are simply more orthogonal than has previously been argued. For instance, Willard & Norenzayan (2013) report quite small contributions (~ r = .10) of mentalizing to the related intuitions that come to otherwise support belief (e.g., dualism, teleology\u00E2\u0080\u00A6). 125 1987). And thus, a better model of how mentalizing comes to be implicated in supernatural agent beliefs is one that can account for why anyone would spend the cognitive effort in reasoning about the minds of disembodied supernatural agents in the first place. In Chapter 3, evidence is provided that mentalizing is deployed in reasoning about the minds of supernatural agents when the contents of those minds are believed to hold valuable (i.e., consequential) information. That is, mentalizing is employed when there is a heightened perceived need to make sense of the intentions and other mental states of others. Sitting on a commuter train, we might otherwise be agnostic to the contents of the minds of those around us \u00E2\u0080\u0093 let alone if they have minds at all \u00E2\u0080\u0093 until someone stands up and begins acting erratically and screaming obscenities. It is only when given a reason to make sense of another\u00E2\u0080\u0099s mind that individuals expend the cognitive effort to infer its contents. Indeed, Epley et al. (2007) call this an \u00E2\u0080\u0098effectance motivation\u00E2\u0080\u0099, and posit that a need to understand the mental states of some target is a primary impetus of when individuals anthropomorphize non-human targets. Applied to supernatural agents, the belief that an agent knows about one\u00E2\u0080\u0099s behaviours and cares enough to do something about it is a particularly compelling reason to mentalize the mind of god (i.e., thinking about what god\u00E2\u0080\u0099s plans to do about this knowledge). Especially so when these supernatural agents are believed to wield influence over your eternal soul. In support of this, it was observed in Chapter 3 that individual differences in mentalizing more strongly predicted belief in a more varied set of supernatural agents to the extent to which these agents/forces were moralized. And notably, the more moralized agents were reported to be, the less tendencies for anthropomorphising predicted belief. Put differently, when supernatural entities are not moralized \u00E2\u0080\u0093 broad tendencies for perceiving mental states in the world writ-large do come to predict supernatural cognitions \u00E2\u0080\u0093 and are perhaps driven by some intuitive process for 126 mind perception (Pennycook et al., 2012; Waytz, Gray, Epley, & Wegner, 2010). But, when agents are attributed with moral knowledge more careful mental state reasoning (i.e.., not just mental state perception) becomes implicated in belief. Moreover, this may be precisely why the broad anthropomorphism hypothesis fails to account for belief in god \u00E2\u0080\u0093 the most consistently moralized supernatural entity examined in this dissertation, and in the world. As individual differences in mentalizing were inconsistently related to anthropomorphic tendencies, this suggests that these may indeed be orthogonal processes \u00E2\u0080\u0093 both of which can come to be implicated in supernatural belief under different circumstances. In brief, the results of Chapter 3 delineate both how mentalizing is employed in reasoning about supernatural agents (i.e., by thinking about an entities\u00E2\u0080\u0099 agentic qualities) and when mentalizing is deployed at all in reasoning about an agent\u00E2\u0080\u0099s mind (i.e., when agents are moralized). You do not need to expend the mental effort in deciphering the contents of the minds of ghosts if you do not think ghosts can affect your life. But, your tendencies for anthropomorphism may just push you into believing that those bumps in the night are more than just old plumbing. In Chapter 4, further evidence was provided suggesting that across an immensely diverse sample of participants \u00E2\u0080\u0093 the extent to which deities were moralized was positively related to the extent to which participants mentalized (i.e., thought and worried) about the contents of their supernatural minds. When deities are less moralized, there is no need to expend the effort in applying one\u00E2\u0080\u0099s mentalizing to deciphering their minds. Why cross-cultural research is necessary? Throughout this dissertation, hypotheses were tested in diverse samples. And to the benefit of this work, cross-sample comparisons often helped reveal both the strengths and limits of core theoretical accounts of religious belief in the cognitive sciences of religion. For instance, the limits 127 of the dual process model of belief could not have been demonstrated without comparing the relationship between thinking style and belief in god even in between seemingly comparable samples as Canadians and Americans. On the other hand, the presented account of how and when mentalizing comes to be implicated in belief in god is greatly strengthened by its support in two distinct religious contexts (Hindus and Christians). Moreover, the consistency with which mentalistic representations of moralistic deity beliefs co-occur across the range of populations sampled in Chapter 4 suggests \u00E2\u0080\u0093 as has long been predicted in the cognitive sciences of religion \u00E2\u0080\u0093 that mentalizing is an important explanatory precursor to beliefs across cultures. Broadly speaking, the cognitive sciences of religion requires cross-cultural comparisons. If we are going to say anything about the psychological foundations of religious belief, we need more detailed maps of the contents of the diverse range of beliefs held by believers around the world. It is only after documenting this diversity of religious beliefs that the cognitive sciences of religion can begin to precisely test hypotheses regarding how and when they are shaped and sustained by psychological processes. Limitations and future directions Although the work presented in this dissertation takes the issue of sampling in the cognitive sciences of religion seriously, there is much more work to be done in increasing the extent to which researchers generate hypotheses based on a deep understanding of specific religious contexts to be tested in specific populations. In Chapter 2, many participants were recruited using Amazon\u00E2\u0080\u0099s Mechanical Turk. And although this certainly provides access to a broader slice of the population than does sampling with undergraduate students (as is typically the norm in psychology studies; Henrich et al., 2010) \u00E2\u0080\u0093 the guiding questions of this work are potentially much better suited to be tested in other samples. For example, no research has been conducted on religious experts (e.g., 128 theologians, monks, etc.) in regard to how analytical thinking comes to be related to belief despite this being a potentially ideal context to observe how and when analytical thinking might come to support belief. Moreover, the participant pool on Amazon\u00E2\u0080\u0099s Mechanical Turk is more liberal and urban than is the general American population (Arechar, Kraft-Todd, & Rand, 2017); and thus M-Turkers might be particularly poor subject pool in which to test how political orientation comes to be moderate the effects of analytical thinking. In addition to sampling concerns, the work presented here is limited in its scope by how it went about measuring complex psychological processes like \u00E2\u0080\u0098analytical thinking\u00E2\u0080\u0099 and \u00E2\u0080\u0098mentalizing\u00E2\u0080\u0099. In both cases, there is a great deal of variability in how these constructs are operationalized within and between research programs. Mentalizing, for instance, is a very broad and non-specific category of psychological processes. Indeed, much of the work assessing the contributions of mentalizing to belief (including the work presented here) has relied on a family of self-report measures (almost all developed by a single research group) of willingness to reason about the mental states of others that have only really been validated by the extent to which they discriminate between non-clinical populations and individuals diagnosed along the Autism Spectrum. In doing so, this work might be missing out on how other dimensions of mentalizing (e.g., ability to accurately reason about mental states) may be implicated in religious belief. Future research then should assess the contributions of mentalizing (measured in diverse ways) to belief in diverse religious contexts. Perhaps most importantly, all of the work presented in this dissertation is correlational \u00E2\u0080\u0093 and cannot speak directly to the causal mechanisms by which any of the examined psychological processes are directly implicated in religious beliefs. And thus, future research should make the exerted effort to experimentally test related hypotheses to those described in this dissertation. As 129 one example, the core prediction of the expressive rationality model of belief (i.e., that analytical thinking is employed in defense of one\u00E2\u0080\u0099s otherwise acquired beliefs in an identity protective manner) is suitably testable in an experimental framework. This could easily be tested by inducing social identity threat in a sample of politically conservative and politically liberal individuals and observing whether they are better able to or more willing to provide reasoned arguments to justify their belief (or non-belief) in god. Although demonstrating experimental effects on belief in god has proven to be difficult (e.g., Sanchez, Sundermeier, Gray, & Calin-Jageman, 2017), the cognitive sciences of religion can truly benefit from more careful experimental research. 130 Bibliography Apperly, I. A., & Butterfill, S. A. (2009). Do humans have two systems to track beliefs and belief-like states? Psychological Review, 116(4), 953\u00E2\u0080\u0093970. https://doi.org/10.1037/a0016923 Arechar, A. A., Kraft-Todd, G. T., & Rand, D. G. (2017). Turking Overtime: How Participant Characteristics and Behavior Vary Over Time and Day on Amazon Mechanical Turk (SSRN Scholarly Paper No. ID 2836946). Retrieved from Social Science Research Network website: https://papers.ssrn.com/abstract=2836946 Armstrong, K. (1994). A History of God: The 4,000-Year Quest of Judaism, Christianity and Islam (Reprint edition). New York: Ballantine Books. Astuti, R., & Harris, P. (2008). Understanding Mortality and the Life of the Ancestors in Rural Madagascar. Cognitive Science: A Multidisciplinary Journal, 32(4), 713\u00E2\u0080\u0093740. https://doi.org/10.1080/03640210802066907 Atran, S., & Medin, D. L. (2008). The native mind and the cultural construction of nature. Cambridge, Mass: MIT Press. Atran, S., & Norenzayan, A. (2004). Religion\u00E2\u0080\u0099s evolutionary landscape: Counterintuition, commitment, compassion, communion. Behavioral and Brain Sciences, 27(06), 713\u00E2\u0080\u0093730. https://doi.org/10.1017/S0140525X04000172 Banerjee, K., & Bloom, P. (2015). \u00E2\u0080\u009CEverything Happens for a Reason\u00E2\u0080\u009D: Children\u00E2\u0080\u0099s Beliefs About Purpose in Life Events. Child Development, 86(2), 503\u00E2\u0080\u0093518. https://doi.org/10.1111/cdev.12312 131 Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173\u00E2\u0080\u00931182. Baron-Cohen, S. (1999). Evolution of a theory of mind? In M. Corballis & S. Lea (Eds.), The descent of mind: psychological perspectives on hominid evolution (pp. 1\u00E2\u0080\u009331). Oxford: Oxford University Press. Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a \u00E2\u0080\u009Ctheory of mind\u00E2\u0080\u009D? Cognitive Development, 21, 37\u00E2\u0080\u009346. Baron-Cohen, S., & Wheelwright, S. (2004). The Empathy Quotient: An Investigation of Adults with Asperger Syndrome or High Functioning Autism, and Normal Sex Differences. Journal of Autism and Developmental Disorders, 34(2), 163\u00E2\u0080\u0093175. https://doi.org/10.1023/B:JADD.0000022607.19833.00 Baron-Cohen, S., Wheelwright, S., Spong, A., Scahill, V., & Lawson, J. (2001). Are intuitive physics and intuitive psychology independent? A test with children with Asperger Syndrome. Journal of Developmental and Learning Disorders, 5, 47\u00E2\u0080\u009378. Barrett, H. C. (2014). The Shape of Thought: How Mental Adaptations Evolve (1 edition). Oxford University Press. Barrett, J. L., Richert, R. A., & Driesenga, A. (2001). God\u00E2\u0080\u0099s beliefs versus mother\u00E2\u0080\u0099s: The development of nonhuman agent concepts. Child Development, 72(1), 50\u00E2\u0080\u009365. Barrett, Justin L. (2004). Why Would Anyone Believe in God? Walnut Creek, CA: AltaMira Press. 132 Bates, D., M\u00C3\u00A4chler, M., Bolker, B., & Walker, S. (2015). Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software, 67(1), 1\u00E2\u0080\u009348. https://doi.org/10.18637/jss.v067.i01 Beck, J. (2018, January 16). The New Age of Astrology. The Atlantic. Retrieved from https://www.theatlantic.com/health/archive/2018/01/the-new-age-of-astrology/550034/ Bellah, R. N. (2017). Religion in Human Evolution: From the Paleolithic to the Axial Age. Place of publication not identified: Belknap Press: An Imprint of Harvard University Press. Bering, J. M., & Bjorklund, D. F. (2004). The Natural Emergence of Reasoning About the Afterlife as a Developmental... Developmental Psychology, 40(2), 217\u00E2\u0080\u0093233. Bloom, P. (2006). Religion is natural. Developmental Science, 10(1), 147\u00E2\u0080\u0093151. https://doi.org/10.1111/j.1467-7687.2007.00577.x Boyer, P. (2001). Religion Explained: The Evolutionary Origins of Religious Thought. Basic Books. Caldwell-Harris, C. L., Murphy, C. F., Velazquez, T., & McNamara, P. (2011). Religious beleif systems of persons with high functioning autism. Proceedings of the 33rd Annual Meeting of the Cognitive Science Society, 3362\u00E2\u0080\u00933366. Retrieved from http://csjarchive.cogsci.rpi.edu/proceedings/2011/papers/0782/paper0782.pdf Crump, M. J. C., McDonnell, J. V., & Gureckis, T. M. (2013). Evaluating Amazon\u00E2\u0080\u0099s Mechanical Turk as a Tool for Experimental Behavioral Research. PLOS ONE, 8(3), e57410. https://doi.org/10.1371/journal.pone.0057410 Daniel L\u00C3\u00BCdecke. (2018). sjPlot - Data Visualization for Statistics in Social Science. https://doi.org/10.5281/zenodo.1310947 133 Davis, M. H., & Oathout, H. A. (1987). Maintenance of satisfaction in romantic relationships: Empathy and relational competence. Journal of Personality and Social Psychology, 53(2), 397\u00E2\u0080\u0093410. https://doi.org/10.1037/0022-3514.53.2.397 Deppe, K. D., Gonzalez, F. J., Neiman, J. L., Jacobs, C., Pahlke, J., Smith, K. B., & Hibbing, J. R. (2015). Reflective liberals and intuitive conservatives: A look at the Cognitive Reflection Test and ideology. Judgment and Decision Making, 10(4), 18. Enders, C. K., & Tofighi, D. (2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods, 12(2), 121\u00E2\u0080\u0093138. https://doi.org/10.1037/1082-989X.12.2.121 Epley, N., Akalis, S., Waytz, A., & Cacioppo, J. T. (2008). Creating Social Connection Through Inferential Reproduction: Loneliness and Perceived Agency in Gadgets, Gods, and Greyhounds. Psychological Science, 19(2), 114\u00E2\u0080\u0093120. https://doi.org/10.1111/j.1467-9280.2008.02056.x Epley, N., Waytz, A., & Cacioppo, J. T. (2007). On seeing human: A three-factor theory of anthropomorphism. Psychological Review, 114(4), 864\u00E2\u0080\u0093886. https://doi.org/10.1037/0033-295X.114.4.864 Evans, St. B. T. (2003). In two minds: Dual-process accounts of reasoning. Trends in Cognitive Sciences, 7(10), 454\u00E2\u0080\u0093459. https://doi.org/10.1016/j.tics.2003.08.012 Feuerbach, L. (1957). The essence of Christianity. Harper. Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 25\u00E2\u0080\u009342. 134 Gervais, W. M. (2013). Perceiving Minds and Gods: How Mind Perception Enables, Constrains, and Is Triggered by Belief in Gods. Perspectives on Psychological Science, 8(4), 380\u00E2\u0080\u0093394. https://doi.org/10.1177/1745691613489836 Gervais, W. M., Elk, M. van, Xygalatas, D., McKay, R., Aveyard, M., Buchtel, E. E. K., \u00E2\u0080\u00A6 Bulbulia, J. (2017). Analytic atheism: A cross-culturally weak and fickle phenomenon? PsyArXiv. https://doi.org/10.17605/OSF.IO/92R8X Gervais, W. M., & Najle, M. B. (2015). Learned Faith: The influences of evolved cultural learning mechanisms on belief in gods. Psychology of Religion and Spirituality. Gervais, W. M., & Norenzayan, A. (2012). Analytic Thinking Promotes Religious Disbelief. Science, 336(6080), 493\u00E2\u0080\u0093496. https://doi.org/10.1126/science.1215647 Gervais, W. M., Willard, A. K., Norenzayan, A., & Henrich, J. (2011). The Cultural Transmission of Faith: Why innate intuitions are necessary, but insufficient, to explain religious belief. Religion, 41(3), 389\u00E2\u0080\u0093410. https://doi.org/10.1080/0048721X.2011.604510 Gray, H. M., Gray, K., & Wegner, D. M. (2007). Dimensions of Mind Perception. Science, 315(5812), 619\u00E2\u0080\u0093619. https://doi.org/10.1126/science.1134475 Gray, K., Jenkins, A. C., Heberlein, A. S., & Wegner, D. M. (2011). Distortions of mind perception in psychopathology. Proceedings of the National Academy of Sciences, 108(2), 477\u00E2\u0080\u0093479. https://doi.org/10.1073/pnas.1015493108 Greenway, T. S. (2016). Mentalizing Domains and Belief in God. Journal for the Cognitive Science of Religion, 4(1), 91\u00E2\u0080\u0093109. https://doi.org/10.1558/jcsr.31063 135 Grice, J. W., Cohn, A., Ramsey, R. R., & Chaney, J. M. (2015). On Muddled Reasoning and Mediation Modeling. Basic and Applied Social Psychology, 37(4), 214\u00E2\u0080\u0093225. https://doi.org/10.1080/01973533.2015.1049350 Guthrie, S., Agassi, J., Andriolo, K. R., Buchdahl, D., Earhart, H. B., Greenberg, M., \u00E2\u0080\u00A6 Tissot, G. (1980). A Cognitive Theory of Religion [and Comments and Reply]. Current Anthropology, 21(2), 181\u00E2\u0080\u0093203. Haidt, J. (2012). The Righteous Mind: Why Good People Are Divided By Politics and Religion. New York: Pantheon Books. Harris, L. T., & Fiske, S. T. (2006). Dehumanizing the Lowest of the Low Neuroimaging Responses to Extreme Out-Groups. Psychological Science, 17(10), 847\u00E2\u0080\u0093853. https://doi.org/10.1111/j.1467-9280.2006.01793.x Henrich, J. (2015). The Secret of Our Success: How Culture Is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter. Princeton: Princeton University Press. Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? The Behavioral and Brain Sciences, 33(2\u00E2\u0080\u00933), 61\u00E2\u0080\u009383; discussion 83-135. https://doi.org/10.1017/S0140525X0999152X Heyes, C. (2018). Cognitive Gadgets: The Cultural Evolution of Thinking. Cambridge, Massachusetts: Belknap Press: An Imprint of Harvard University Press. Hooker, C. I., Verosky, S. C., Germine, L. T., Knight, R. T., & D\u00E2\u0080\u0099Esposito, M. (2008). Mentalizing about emotion and its relationship to empathy. Social Cognitive and Affective Neuroscience, 3(3), 204\u00E2\u0080\u0093217. https://doi.org/10.1093/scan/nsn019 Hume, D. (1757). The natural history of religion. Stanford, CA: Stanford University Press. 136 Hume, D. (1779). Dialogues Concerning Natural Religion (Reprint edition). London\u00E2\u0080\u0089; New York: Penguin Classics. Jack, A. I., Friedman, J. P., Boyatzis, R. E., & Taylor, S. N. (2016). Why Do You Believe in God? Relationships between Religious Belief, Analytic Thinking, Mentalizing and Moral Concern. PLOS ONE, 11(3), e0149989. https://doi.org/10.1371/journal.pone.0149989 Jacoby, S. (2017). Strange Gods: A Secular History of Conversion (Reprint edition). New York: Vintage. Kahan, D. M. (2017). Misconceptions, Misinformation, and the Logic of Identity-Protective Cognition (SSRN Scholarly Paper No. ID 2973067). Retrieved from Social Science Research Network website: https://papers.ssrn.com/abstract=2973067 Kahan, D. M., & Stanovich, K. (2016). Rationality and Belief in Human Evolution (SSRN Scholarly Paper No. ID 2838668). Retrieved from Social Science Research Network website: https://papers.ssrn.com/abstract=2838668 Kapogiannis, D., Barbey, A. K., Su, M., Zamboni, G., Krueger, F., & Grafman, J. (2009). Cognitive and neural foundations of religious belief. Proceedings of the National Academy of Sciences, 106(12), 4876\u00E2\u0080\u00934881. https://doi.org/10.1073/pnas.0811717106 Kelemen, D. (2004). Are Children \u00E2\u0080\u009CIntuitive Theists\u00E2\u0080\u009D?: Reasoning About Purpose and Design in Nature. Psychological Science, 15(5), 295\u00E2\u0080\u0093301. https://doi.org/10.1111/j.0956-7976.2004.00672.x Kelemen, D., Rottman, J., & Seston, R. (2013). Professional physical scientists display tenacious teleological tendencies: Purpose-based reasoning as a cognitive default. Journal of Experimental Psychology: General, 142(4), 1074\u00E2\u0080\u00931083. https://doi.org/10.1037/a0030399 137 Kline, R. B. (2015). The Mediation Myth. Basic and Applied Social Psychology, 37(4), 202\u00E2\u0080\u0093213. https://doi.org/10.1080/01973533.2015.1049349 Laland, K. N. (2017). Darwin\u00E2\u0080\u0099s Unfinished Symphony: How Culture Made the Human Mind. Princeton, NJ: Princeton University Press. Lanman, J. A., & Buhrmester, M. D. (2016). Religious actions speak louder than words: Exposure to credibility-enhancing displays predicts theism. Religion, Brain & Behavior, 1\u00E2\u0080\u009314. https://doi.org/10.1080/2153599X.2015.1117011 Lindeman, M., & Lipsanen, J. (2016a). Diverse Cognitive Profiles of Religious Believers and Nonbelievers. The International Journal for the Psychology of Religion, 26(3), 185\u00E2\u0080\u0093192. https://doi.org/10.1080/10508619.2015.1091695 Lindeman, M., & Lipsanen, J. (2016b). Mentalizing: Seeking the Underlying Dimensions. International Journal of Psychological Studies, 9(1), 10. https://doi.org/10.5539/ijps.v9n1p10 Lindeman, M., Svedholm-H\u00C3\u00A4kkinen, A. M., & Lipsanen, J. (2015). Ontological confusions but not mentalizing abilities predict religious belief, paranormal belief, and belief in supernatural purpose. Cognition, 134, 63\u00E2\u0080\u009376. https://doi.org/10.1016/j.cognition.2014.09.008 Luhrmann, T. M. (2012). When God Talks Back: Understanding the American Evangelical Relationship with God (Reprint edition). New York: Vintage. Maij, D. L. R., Harreveld, F. van, Gervais, W., Schrag, Y., Mohr, C., & Elk, M. van. (2017). Mentalizing skills do not differentiate believers from non-believers, but credibility enhancing displays do. PLOS ONE, 12(8), e0182764. https://doi.org/10.1371/journal.pone.0182764 138 Ma-Kellams, C., & Lerner, J. (2016). Trust Your Gut or Think Carefully? Examining Whether an Intuitive, Versus a Systematic, Mode of Thought Produces Greater Empathic Accuracy. Retrieved from http://papers.ssrn.com/abstract=2782596 McCauley, R. N. (2011). Why Religion is Natural and Science is Not. Oxford, New York: Oxford University Press. McNamara, R. A., & Henrich, J. (2017). Jesus vs. the ancestors: How specific religious beliefs shape prosociality on Yasawa Island, Fiji. Religion, Brain & Behavior, 0(0), 1\u00E2\u0080\u009320. https://doi.org/10.1080/2153599X.2016.1267030 Mogan, R., Fischer, R., & Bulbulia, J. A. (2017). To be in synchrony or not? A meta-analysis of synchrony\u00E2\u0080\u0099s effects on behavior, perception, cognition and affect. Journal of Experimental Social Psychology, 72, 13\u00E2\u0080\u009320. https://doi.org/10.1016/j.jesp.2017.03.009 Nail, P. R., McGregor, I., Drinkwater, A. E., Steele, G. M., & Thompson, A. W. (2009). Threat causes liberals to think like conservatives. Journal of Experimental Social Psychology, 45(4), 901\u00E2\u0080\u0093907. https://doi.org/10.1016/j.jesp.2009.04.013 Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175\u00E2\u0080\u0093220. https://doi.org/10.1037/1089-2680.2.2.175 Nishi, A., Christakis, N. A., & Rand, D. G. (2017). Cooperation, decision time, and culture: Online experiments with American and Indian participants. PLOS ONE, 12(2), e0171252. https://doi.org/10.1371/journal.pone.0171252 Norenzayan, A. (2013). Big Gods: How Religion Transformed Cooperation and Conflict (1 edition). Princeton: Princeton University Press. Norenzayan, A. (2016). Theodiversity. Annual Review of Psychology, 67(1). https://doi.org/10.1146/annurev-psych-122414-033426 139 Norenzayan, A., & Gervais, W. M. (2013). The origins of religious disbelief. Trends in Cognitive Sciences, 17(1), 20\u00E2\u0080\u009325. https://doi.org/10.1016/j.tics.2012.11.006 Norenzayan, A., Gervais, W. M., & Trzesniewski, K. H. (2012). Mentalizing Deficits Constrain Belief in a Personal God. PLOS ONE, 7(5), e36880. https://doi.org/10.1371/journal.pone.0036880 Norenzayan, A., & Shariff, A. F. (2008). The Origin and Evolution of Religious Prosociality. Science, 322(5898), 58\u00E2\u0080\u009362. https://doi.org/10.1126/science.1158757 Norenzayan, A., Shariff, A. F., Gervais, W. M., Willard, A. K., McNamara, R. A., Slingerland, E., & Henrich, J. (2016). The cultural evolution of prosocial religions. Behavioral and Brain Sciences, 39, e1 (19 pages). https://doi.org/10.1017/S0140525X14001356 Norris, P., & Inglehart, R. (2011). Sacred and Secular: Religion and Politics Worldwide (2 edition). Cambridge: Cambridge University Press. Pacini, R., & Epstein, S. (1999). The relation of rational and experiential information processing styles to personality, basic beliefs, and the ratio-bias phenomenon. Journal of Personality and Social Psychology, 76(6), 972\u00E2\u0080\u0093987. https://doi.org/10.1037/0022-3514.76.6.972 Paloutzian, R. F., & Kaufman, N. L. (1991). Manual for the spiritual well-being scale. Nyack: Life Advance, Inc. Pennycook, G., Cheyne, J. A., Koehler, D. J., & Fugelsang, J. A. (2016). Is the cognitive reflection test a measure of both reflection and intuition? Behavior Research Methods, 48(1), 341\u00E2\u0080\u0093348. https://doi.org/10.3758/s13428-015-0576-1 Pennycook, G., Cheyne, J. A., Seli, P., Koehler, D. J., & Fugelsang, J. A. (2012). Analytic cognitive style predicts religious and paranormal belief. Cognition, 123(3), 335\u00E2\u0080\u0093346. https://doi.org/10.1016/j.cognition.2012.03.003 140 Pennycook, G., Ross, R. M., Koehler, D. J., & Fugelsang, J. A. (2016). Atheists and Agnostics Are More Reflective than Religious Believers: Four Empirical Studies and a Meta-Analysis. PLOS ONE, 11(4), e0153039. https://doi.org/10.1371/journal.pone.0153039 Pew Research Center. (2009, December 9). Many Americans Mix Multiple Faiths. Retrieved August 20, 2018, from Pew Research Center\u00E2\u0080\u0099s Religion & Public Life Project website: http://www.pewforum.org/2009/12/09/many-americans-mix-multiple-faiths/ Pew Research Center. (2013, June 27). Canada\u00E2\u0080\u0099s Changing Religious Landscape. Retrieved August 21, 2018, from Pew Research Center\u00E2\u0080\u0099s Religion & Public Life Project website: http://www.pewforum.org/2013/06/27/canadas-changing-religious-landscape/ Pew Research Center. (2014). Worldwide, Many See Belief in God as Essential to Morality. Retrieved August 20, 2018, from Pew Research Center\u00E2\u0080\u0099s Global Attitudes Project website: http://www.pewglobal.org/2014/03/13/worldwide-many-see-belief-in-god-as-essential-to-morality/ Pew Research Center. (2015). The Future of World Religions: Population Growth Projects, 2010-2050. Pew Research Center. (2017). The Partisan Divide on Political Values Grows Even Wider. Purzycki, Benjamin G., & Sosis, R. (2011). Our Gods: Variation in Supernatural Minds. In The Frontiers Collection. Essential Building Blocks of Human Nature (pp. 77\u00E2\u0080\u009393). https://doi.org/10.1007/978-3-642-13968-0_5 Purzycki, Benjamin Grant. (2013a). The minds of gods: A comparative study of supernatural agency. Cognition, 129(1), 163\u00E2\u0080\u0093179. https://doi.org/10.1016/j.cognition.2013.06.010 141 Purzycki, Benjamin Grant. (2013b). Toward a Cognitive Ecology of Religious Concepts: Evidence from the Tyva Republic. Journal for the Cognitive Science of Religion, 1(1). https://doi.org/doi:10.1558/jcsr.v1i1.99 Purzycki, Benjamin Grant. (2016). The Evolution of Gods\u00E2\u0080\u0099 Minds in the Tyva Republic. Current Anthropology, 57(S13), S88\u00E2\u0080\u0093S104. https://doi.org/10.1086/685729 Purzycki, Benjamin Grant, Apicella, C., Atkinson, Q. D., Cohen, E., McNamara, R. A., Willard, A. K., \u00E2\u0080\u00A6 Henrich, J. (2016a). Cross-cultural dataset for the evolution of religion and morality project. Scientific Data, 3, sdata201699. https://doi.org/10.1038/sdata.2016.99 Purzycki, Benjamin Grant, Apicella, C., Atkinson, Q. D., Cohen, E., McNamara, R. A., Willard, A. K., \u00E2\u0080\u00A6 Henrich, J. (2016b). Moralistic gods, supernatural punishment and the expansion of human sociality. Nature, 530(7590), 327\u00E2\u0080\u0093330. https://doi.org/10.1038/nature16980 Purzycki, Benjamin Grant, Henrich, J., Apicella, C., Atkinson, Q. D., Baimel, A., Cohen, E., \u00E2\u0080\u00A6 Norenzayan, A. (2018). The evolution of religion and morality: A synthesis of ethnographic and experimental evidence from eight societies. Religion, Brain & Behavior, 8(2), 101\u00E2\u0080\u0093132. https://doi.org/10.1080/2153599X.2016.1267027 Purzycki, Benjamin Grant, & McNamara, R. A. (2016). An ecological theory of gods\u00E2\u0080\u0099 minds. In Advances in Religion, Cognitive Science, and Expreimental Philosophy. Bloomsbury Publishing. R Core Team. (2017). R: A Language and Environment for Statistical Computing. Retrieved from https://www.R-project.org Raihani, N. J., Mace, R., & Lamba, S. (2013). The Effect of $1, $5 and $10 Stakes in an Online Dictator Game. PLOS ONE, 8(8), e73131. https://doi.org/10.1371/journal.pone.0073131 142 Rand, D. G. (2016). Cooperation, Fast and Slow: Meta-Analytic Evidence for a Theory of Social Heuristics and Self-Interested Deliberation. Psychological Science. https://doi.org/10.1177/0956797616654455 Rappaport, R. A. (1999). Ritual and Religion in the Making of Humanity (1 edition). Cambridge: Cambridge University Press. Reddish, P., Tok, P., & Kundt, R. (2016). Religious Cognition and Behaviour in Autism: The Role of Mentalizing. The International Journal for the Psychology of Religion, 26(2), 95\u00E2\u0080\u0093112. https://doi.org/10.1080/10508619.2014.1003518 Sanchez, C., Sundermeier, B., Gray, K., & Calin-Jageman, R. J. (2017). Direct replication of Gervais & Norenzayan (2012): No evidence that analytic thinking decreases religious belief. PLOS ONE, 12(2), e0172636. https://doi.org/10.1371/journal.pone.0172636 Saribay, S. A., & Yilmaz, O. (2017). Analytic cognitive style and cognitive ability differentially predict religiosity and social conservatism. Personality and Individual Differences, 114, 24\u00E2\u0080\u009329. https://doi.org/10.1016/j.paid.2017.03.056 Schaafsma, S. M., Pfaff, D. W., Spunt, R. P., & Adolphs, R. (2015). Deconstructing and reconstructing theory of mind. Trends in Cognitive Sciences, 19(2), 65\u00E2\u0080\u009372. https://doi.org/10.1016/j.tics.2014.11.007 Schjoedt, U., St\u00C3\u00B8dkilde-J\u00C3\u00B8rgensen, H., Geertz, A. W., & Roepstorff, A. (2009). Highly religious participants recruit areas of social cognition in personal prayer. Social Cognitive and Affective Neuroscience, 4(2), 199\u00E2\u0080\u0093207. https://doi.org/10.1093/scan/nsn050 Schmidt, M. F. H., Butler, L. P., Heinz, J., & Tomasello, M. (2016). Young Children See a Single Action and Infer a Social Norm Promiscuous Normativity in 3-Year-Olds. Psychological Science, 27(10), 1360\u00E2\u0080\u00931370. https://doi.org/10.1177/0956797616661182 143 Severson, R. L., & Lemm, K. M. (2016). Kids See Human Too: Adapting an Individual Differences Measure of Anthropomorphism for a Child Sample. Journal of Cognition and Development, 17(1), 122\u00E2\u0080\u0093141. https://doi.org/10.1080/15248372.2014.989445 Shariff, A. F., Willard, A. K., Andersen, T., & Norenzayan, A. (2015). Religious Priming: A Meta-Analysis With a Focus on Prosociality. Personality and Social Psychology Review: An Official Journal of the Society for Personality and Social Psychology, Inc. https://doi.org/10.1177/1088868314568811 Shenhav, A., Rand, D. G., & Greene, J. D. (2012). Divine intuition: Cognitive style influences belief in God. Journal of Experimental Psychology: General, 141(3), 423\u00E2\u0080\u0093428. https://doi.org/10.1037/a0025391 Spencer, S. J., Zanna, M. P., & Fong, G. T. (2005). Establishing a causal chain: Why experiments are often more effective than mediational analyses in examining psychological processes. Journal of Personality and Social Psychology, 89, 845\u00E2\u0080\u0093851. Stagnaro, M., Pennycook, G., & Rand, D. G. (2018). Cognitive Reflection is a Stable Trait (SSRN Scholarly Paper No. ID 3115809). Retrieved from Social Science Research Network website: https://papers.ssrn.com/abstract=3115809 Thoemmes, F. (2015). Reversing Arrows in Mediation Models Does Not Distinguish Plausible Models. Basic and Applied Social Psychology, 37(4), 226\u00E2\u0080\u0093234. https://doi.org/10.1080/01973533.2015.1049351 Vonk, J., & Pitzen, J. (2016). Believing in other minds: Accurate mentalizing does not predict religiosity. Personality and Individual Differences. https://doi.org/10.1016/j.paid.2016.06.008 144 Wakabayashi, A., Baron-Cohen, S., Wheelwright, S., Goldenfeld, N., Delaney, J., Fine, D., \u00E2\u0080\u00A6 Weil, L. (2006). Development of short forms of the Empathy Quotient (EQ-Short) and the Systemizing Quotient (SQ-Short). Personality and Individual Differences, 41(5), 929\u00E2\u0080\u0093940. https://doi.org/10.1016/j.paid.2006.03.017 Waytz, A., Cacioppo, J., & Epley, N. (2014). Who Sees Human? The Stability and Importance of Individual Differences in Anthropomorphism. Perspectives on Psychological Science\u00E2\u0080\u0089: A Journal of the Association for Psychological Science, 5(3), 219\u00E2\u0080\u0093232. https://doi.org/10.1177/1745691610369336 Waytz, A., Gray, K., Epley, N., & Wegner, D. M. (2010). Causes and consequences of mind perception. Trends in Cognitive Sciences, 14(8), 383\u00E2\u0080\u0093388. https://doi.org/10.1016/j.tics.2010.05.006 White, C. J. M. (2017). Belief in karma: The content and correlates of supernatural justice beliefs across cultures. University of British Columbia, Vancouver, Canada. White, Cindel J. M., Norenzayan, A., & Schaller, M. (2018). The Content and Correlates of Belief in Karma Across Cultures. Personality and Social Psychology Bulletin, 0146167218808502. https://doi.org/10.1177/0146167218808502 Willard, A. K., & Cingl, L. (2017). Testing Theories of Secularization and Religious Belief in the Czech Republic and Slovakia. Evolution and Human Behavior, 0(0). https://doi.org/10.1016/j.evolhumbehav.2017.01.002 Willard, A. K., & Norenzayan, A. (2013). Cognitive biases explain religious belief, paranormal belief, and belief in life\u00E2\u0080\u0099s purpose. Cognition, 129(2), 379\u00E2\u0080\u0093391. https://doi.org/10.1016/j.cognition.2013.07.016 145 Willard, A. K., & Norenzayan, A. (2017). \u00E2\u0080\u009CSpiritual but not religious\u00E2\u0080\u009D: Cognition, schizotypy, and conversion in alternative beliefs. Cognition, 165, 137\u00E2\u0080\u0093146. https://doi.org/10.1016/j.cognition.2017.05.018 Wlodarski, R., & Pearce, E. (2016). The God Allusion. Human Nature, 27(2), 160\u00E2\u0080\u0093172. https://doi.org/10.1007/s12110-016-9256-9 Xygalatas, D., Mitkidis, P., Fischer, R., Reddish, P., Skewes, J., Geertz, A. W., \u00E2\u0080\u00A6 Bulbulia, J. (2013). Extreme rituals promote prosociality. Psychological Science, 24(8), 1602\u00E2\u0080\u00931605. https://doi.org/10.1177/0956797612472910 Yilmaz, O., & Saribay, S. A. (2016). An attempt to clarify the link between cognitive style and political ideology: A non-western replication and extension. Judgment and Decision Making, 11(3), 287. 146 Appendix Table S2.1 Testing the expressive rationality model \u00E2\u0080\u0093 belief that religion is necessary for morality Model 1 Model 2 Model 3 Predictors b CI b CI b CI (Intercept) 0.17 0.11 \u00E2\u0080\u0093 0.23 0.19 0.08 \u00E2\u0080\u0093 0.29 0.19 0.08 \u00E2\u0080\u0093 0.29 CRT -0.12 -0.15 \u00E2\u0080\u0093 -0.09 -0.12 -0.18 \u00E2\u0080\u0093 -0.07 -0.12 -0.18 \u00E2\u0080\u0093 -0.07 Conservativism 0.32 0.26 \u00E2\u0080\u0093 0.38 0.24 0.13 \u00E2\u0080\u0093 0.35 0.24 0.13 \u00E2\u0080\u0093 0.35 CRT * Conservativism 0.01 -0.02 \u00E2\u0080\u0093 0.04 0.03 -0.02 \u00E2\u0080\u0093 0.09 0.03 -0.02 \u00E2\u0080\u0093 0.09 Sample (1 = USA) -0.03 -0.15 \u00E2\u0080\u0093 0.10 -0.03 -0.16 \u00E2\u0080\u0093 0.10 CRT * Sample 0.01 -0.06 \u00E2\u0080\u0093 0.08 0.01 -0.05 \u00E2\u0080\u0093 0.08 Conservativism * Sample 0.12 -0.01 \u00E2\u0080\u0093 0.25 0.13 0.00 \u00E2\u0080\u0093 0.26 CRT * Sample * Conservativism -0.03 -0.09 \u00E2\u0080\u0093 0.04 -0.03 -0.09 \u00E2\u0080\u0093 0.04 Age (years, std.) -0.04 -0.08 \u00E2\u0080\u0093 -0.00 Male (1 = Male) -0.00 -0.09 \u00E2\u0080\u0093 0.08 Observations 2256 2256 2256 AIC 6099.370 6121.616 6132.045 147 Table S3.1 Mediated linear regression models from Studies 2A and 2B in Chapter 3 Broad Anthropomorphism Hypothesis DV: Belief in a personal god MED: Anthropomorphism Narrow Anthropomorphism Hypothesis DV: Belief in a personal god MED: Mental state attribution Anthropocentric Knowledge Hypothesis DV: Mental state attribution MED: Belief in a personal god Predictors b CI b CI b CI (Intercept) Hindu -.15 -.27 \u00E2\u0080\u0093 -.04 -.13 -.23 \u00E2\u0080\u0093 -.03 .01 -.09 \u00E2\u0080\u0093 .11 Christ -.11 -.20 \u00E2\u0080\u0093 -.01 -.11 -.20 \u00E2\u0080\u0093 -.02 .02 -.07 \u00E2\u0080\u0093 .12 Mentalizing (mediated) Hindu .22 .15 \u00E2\u0080\u0093 .29 .19 .13 \u00E2\u0080\u0093 .26 -.02 -.09 \u00E2\u0080\u0093 .05 Christ .27 .20 \u00E2\u0080\u0093 .34 .20 .14 \u00E2\u0080\u0093 .26 .05 -.01 \u00E2\u0080\u0093 .12 Anthropomorphism Hindu -.00 -.07 \u00E2\u0080\u0093 .07 Christ -.12 -.19, -.05 Mental state attribution to god Hindu .40 .33 \u00E2\u0080\u0093 .46 Christ .40 .34 \u00E2\u0080\u0093 .47 Belief in a personal god Hindu .41 .34 \u00E2\u0080\u0093 .48 Christ .46 .39 \u00E2\u0080\u0093 .53 Age (years) Hindu .05 -.03 \u00E2\u0080\u0093 .12 .11 .04 \u00E2\u0080\u0093 .18 -.18 -.24 \u00E2\u0080\u0093 -.11 Christ .16 .09 \u00E2\u0080\u0093 .23 .18 .12 \u00E2\u0080\u0093 .24 -.08 -.15 \u00E2\u0080\u0093 -.02 Female (1 = yes) Hindu .26 .11 \u00E2\u0080\u0093 .41 .23 .09 \u00E2\u0080\u0093 .36 -.02 -.16 \u00E2\u0080\u0093 .12 Christ .19 .06 \u00E2\u0080\u0093 .33 .19 .07 \u00E2\u0080\u0093 .32 -.04 -.17 \u00E2\u0080\u0093 .09 Formal education (years) Hindu -.02 -.09 \u00E2\u0080\u0093 .06 .02 -.05 \u00E2\u0080\u0093 .08 -.08 -.14 \u00E2\u0080\u0093 -.01 Christ -.02 -.09 \u00E2\u0080\u0093 .04 .02 -.04 \u00E2\u0080\u0093 .08 -.07 -.13 \u00E2\u0080\u0093 -.00 R2 / adjusted R2 Hindu .076 / .070 .227 / .222 .206 / .201 Christ .156 / .151 .302 / .298 .216 / .211 148 Table S3.2 The mediating effects of moral knowledge attributions on the relationship between agency attributions and belief in a personal god in the Hindu sample \u00E2\u0080\u0093 Study 2A Chapter 3 Belief in a personal god Moral knowledge attributions Belief in a personal god Predictors b CI b CI b CI (Intercept) -0.16 -0.25 \u00E2\u0080\u0093 -0.07 -0.11 -0.20 \u00E2\u0080\u0093 -0.02 -0.10 -0.19 \u00E2\u0080\u0093 -0.02 Agency 0.54 0.48 \u00E2\u0080\u0093 0.60 0.64 0.59 \u00E2\u0080\u0093 0.70 0.21 0.14 \u00E2\u0080\u0093 0.28 Age 0.12 0.05 \u00E2\u0080\u0093 0.18 0.05 -0.01 \u00E2\u0080\u0093 0.11 0.09 0.04 \u00E2\u0080\u0093 0.15 Female (1=yes) 0.27 0.15 \u00E2\u0080\u0093 0.40 0.19 0.07 \u00E2\u0080\u0093 0.30 0.17 0.06 \u00E2\u0080\u0093 0.29 Formal education 0.02 -0.05 \u00E2\u0080\u0093 0.08 -0.03 -0.09 \u00E2\u0080\u0093 0.03 0.03 -0.02 \u00E2\u0080\u0093 0.09 Moral knowledge 0.52 0.45 \u00E2\u0080\u0093 0.59 Observations 749 749 749 R2 / adjusted R2 0.312 / 0.308 0.432 / 0.429 0.464 / 0.460 149 Table S3.3 The mediating effects of moral knowledge attributions on the relationship between agency attributions and belief in a personal god in the Christian sample - Study 2B Chapter 3 Belief in a personal god Moral knowledge attributions Belief in a personal god Predictors b CI b CI b CI (Intercept) -0.13 -0.21 \u00E2\u0080\u0093 -0.05 0.02 -0.06 \u00E2\u0080\u0093 0.10 -0.14 -0.21 \u00E2\u0080\u0093 -0.07 Agency 0.55 0.49 \u00E2\u0080\u0093 0.60 0.64 0.59 \u00E2\u0080\u0093 0.70 0.25 0.18 \u00E2\u0080\u0093 0.32 Age 0.14 0.08 \u00E2\u0080\u0093 0.20 -0.04 -0.10 \u00E2\u0080\u0093 0.02 0.15 0.10 \u00E2\u0080\u0093 0.21 Female (1=yes) 0.25 0.13 \u00E2\u0080\u0093 0.36 -0.03 -0.14 \u00E2\u0080\u0093 0.08 0.26 0.16 \u00E2\u0080\u0093 0.37 Formal education 0.03 -0.03 \u00E2\u0080\u0093 0.08 -0.02 -0.08 \u00E2\u0080\u0093 0.03 0.04 -0.02 \u00E2\u0080\u0093 0.09 Moral knowledge 0.46 0.39 \u00E2\u0080\u0093 0.52 Observations 771 771 771 R2 / adjusted R2 0.366 / 0.363 0.411 / 0.408 0.490 / 0.487 150 Table S4.1 Means (standard deviations) of demographic variables by site \u00E2\u0080\u0093 Chapter 4 Culture Wave N Males Age Yrs. Formal Ed Material Insecurity Children Cachoeira II 274 83 34.19 (12.87) 8.58 (4.02) .86 (.29) 1.81 (1.92) Coastal Tanna I + II 178 88 35.14 (14.33) 7.76 (4.22) .28 (.36) 2.62 (2.06) Huatasani II 94 37 38.51 (15.92) 8.96 (3.80) .79 (.30) 2.47 (2.04) Inland Tanna I + II 112 57 36.25 (15.40) .68 (2.04) .28 (.38) 3.39 (3.35) Kananga II 200 79 38.09 (14.46) 9.51 (3.32) .83 (.34) 4.49 (2.98) Lovu I 76 24 44.56 (16.94) 8.77 (3.78) .83 (.33) 2.24 (1.59) Maraj\u00C3\u00B3 I 77 37 34.12 (13.08) 8.00 (3.53) .86 (.24) 2.18 (2.56) Mauritius I + II 245 144 36.93 (15.80) 8.84 (3.57) .36 (.38) 1.34 (1.72) Mysore II 165 94 33.56 (12.34) 13.35 (5.42) .10 (.28) .91 (1.10) Samburu II 40 12 51.27 (12.48) .70 (1.76) .64 (.42) 8.43 (4.13) Sursurunga II 163 73 37.60 (14.13) 7.51 (2.63) .57 (.40) 3.01 (2.49) Turkana II 247 91 38.03 (16.38) .48 (1.23) .20 (.29) 3.96 (3.85) Tyva Republic I 81 23 33.53 (12.52) 15.44 (2.29) .47 (.28) 1.70 (1.43) Yasawa I 75 34 38.04 (15.91) 9.66 (2.42) .50 (.40) 2.00 (2.07) Total 2027 876 36.98 (14.79) 7.15 (5.39) .51 (.44) 2.74 (2.83) "@en . "Thesis/Dissertation"@en . "2019-09"@en . "10.14288/1.0379879"@en . "eng"@en . "Psychology"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "Attribution-NonCommercial-NoDerivatives 4.0 International"@* . "http://creativecommons.org/licenses/by-nc-nd/4.0/"@* . "Graduate"@en . "Reasoning about the supernatural : a cross-cultural examination of how and when intuitions shape belief"@en . "Text"@en . "http://hdl.handle.net/2429/71014"@en .