Masarykova univerzita Filozofická fakulta

Ústav religionistiky

Martin Lang

The Cognitive Science of Religion

Connecting the Humanities and the Sciences in the Study of Ritual Practice, Prosociality, and Anxiety

Dizertační práce

Školitel: Mgr. Aleš Chalupa, Ph.D.

Brno 2016

Prohlašuji, že jsem tuto dizertační práci vypracoval samostatně s využitím uvedených pramenů a literatury.

……………………………………… Mgr. Martin Lang

V Brně dne 22. září 2016

TABLE OF CONTENTS CZECH ABSTRACT I ENGLISH ABSTRACT II ACKNOWLEDGEMENTS III PREFACE IV INTRODUCTION - 1 -

THE COGNITIVE SCIENCE OF RELIGION - 4 - Two Cultures and the Divide between the Humanities and the Sciences - 4 - CSR and Consilience - 7 - CSR and the Mechanistic Approach - 12 - EFFECTS OF RELIGION AND RITUAL ON PROSOCIALITY - 18 -

RELIGION AND PROSOCIALITY - 23 - Introduction - 23 - The Religious Congruence Fallacy - 25 - What Religious People Say They Do - 26 - What Religious People Actually Do - 28 - Religion as Prime - 31 - Belief and Practice - 35 - Future Directions - 37 - Summary - 38 - MUSIC AS A SACRED CUE? EFFECTS OF RELIGIOUS MUSIC ON MORAL BEHAVIOR - 40 - Abstract - 40 - Introduction - 40 - Materials and Methods - 44 - Results - 50 - Discussion - 62 - LOST IN THE RHYTHM: EFFECTS OF RHYTHM ON SUBSEQUENT INTERPERSONAL COORDINATION - 68 - Abstract - 68 - Introduction - 68 - Materials and Methods - 70 - Results - 74 - Discussion - 79 - Appendix - 83 - EFFECTS OF RITUAL BEHAVIOR ON ANXIETY - 84 -

EFFECTS OF ANXIETY ON SPONTANEOUS RITUALIZED BEHAVIOR - 88 - Summary - 88 - Results - 89 - Discussion - 95 - Experimental Procedures - 98 - ANXIETY AND RITUALIZATION: CAN ATTENTION DISCRIMINATE COMPULSION FROM ROUTINE? - 100 - DISCUSSION - 103 - REFERENCES - 115 - SUPPLEMENTARY MATERIAL - 145 -

CZECH ABSTRACT

Náboženství je tradičním tématem mnohých humanitních a sociálních věd, jež při jeho studiu nabízí různorodé perspektivy a metodologické přístupy. Jelikož většina religionistů pracuje na humanitních fakultách, tak tradičními přístupy ve studiu náboženství jsou interpretace a komparace různých náboženských jevů, aniž by existovala snaha tyto jevy vysvětlit pomocí teorií dokumentujících obecné trendy v náboženském myšlení a jednání. Když však do studia náboženství vstoupily se svým důrazem na kauzální vysvětlení přírodovědné proudy, začaly vznikat dva epistemologicky a ontologicky neslučitelné přístupy ke studiu náboženství, které produkují vzájemně nekompatibilní vědění. V předložené dizertaci tuto situaci kritizuji a nabízím řešení v podobě kognitivní vědy o náboženství, která se snaží sjednotit humanitní a přírodní vědy v jejich přístupu ke studiu náboženství. Nejprve popisuji historické a epistemologické příčiny rozdílů mezi oběma přístupy, přičemž se soustředím na řešení, která navrhuje kognitivní religionistika a princip konsilience. Dále navrhuji mechanistický přístup jako epistemologický nástroj, který umožňuje překládat koncepty mez vědami na různých úrovních komplexity, a nabízí tedy způsob jak překlenout propast mez přírodními a humanitními vědami. V druhé kapitole se pak věnuji konkrétní aplikaci přístupu kognitivní religionistiky na teorii Émila Durkheima o náboženství jako instituci sdružující lidi do jednotných společenství. Ve třech předložených studiích dokumentuji různé mechanismy, jež jsou základem jevů popsaných Durkheimem. Další kapitola pak aplikuje stejný přístup na studium teorie o magických rituálech a úzkosti, která byla předložena Bronislawem Malinowským. Opět prezentuji své dvě originální studie na toto téma, přičemž jedna z nich zkoumá efekty úzkosti na rituální chování a druhá možné kognitivní mechanismy podmiňující tyto efekty. V poslední kapitole diskutuji výsledky těchto experimentů a identifikuji několik problémů, kterým současná kognitivní věda o náboženství čelí. Jako řešení těchto problémů pak na závěr nabízím kontextuální, proximátní, a ultimátní mechanistické analýzy, které mohou být implementovány v budoucích studiích.

Klíčová slova: náboženství, kognitivní věda o náboženství, propast mezi humanitními a přírodními vědami, konsilience, mechanistický přístup, Émile Durkheim, prosocialita, symbol, kolektivní vzrušení, hudba, rytmus, synchronizace, Bronislaw Malinowski, magické rituály, úzkost, ritualizované chování, proximátní and ultimátní vysvětlení

i

ENGLISH ABSTRACT

Religion has traditionally been one of the core topics for multiple humanistic and social scientific disciplines, affording diverse perspectives and methodological approaches. Based in humanistic departments, Religious Studies scholars mostly focus on an interpretation and comparison of various religious phenomena without an attempt to look for explanations generalizing their findings across different religions. However, when the life sciences entered the study of religion, their emphasis on generalizations and causal explanations has created an aggregation of theories that are epistemologically and ontologically incompatible with the humanistic theories. In this dissertation, I argue that the current situation hinders progress in the study of religion and that the Cognitive Science of Religion (CSR) offers a possible unifying framework that may bridge the gap between the humanities and the sciences. First, the roots of this gap are described with a particular emphasis on the principle of consilience and solutions devised by CSR scholars. Next, the mechanistic approach is advocated as an epistemological tool allowing to translate concepts between academic disciplines that are on different complexity levels, whereby the approach affords to connect the humanities and the sciences. In the second chapter, I apply the CSR methodology on Émile Durkheim’s theory of religion as a social glue and present my three original scientific studies that demonstrate underlying mechanisms of phenomena asserted by Durkheim. The next chapter emulates the previous approach and focuses on the theory of Bronislaw Malinowski that suggest a relationship between magical rituals and anxiety. Again, I present my two original studies illuminating mechanisms behind phenomena assumed by Malinowski: one on the effects of anxiety on spontaneous ritual behavior, and one on possible cognitive mechanisms underlying these effects. Finally, in the last chapter, I discuss the results of these experimental studies and identify peculiar difficulties in the CSR approach. As a remedy for these difficulties, I offer contextualized, proximate, and ultimate mechanistic analyses that can be implemented in future studies.

Key words: religion, Cognitive Science of Religion, sciences-humanities divide, consilience, mechanistic approach, Émile Durkheim, prosociality, symbol, collective effervescence, music, rhythm, synchrony, Bronislaw Malinowski, magical rituals, anxiety, ritualized behavior, proximate and ultimate explanations

ii

ACKNOWLEDGEMENTS

I would like to thank my doctoral supervisor Aleš Chalupa for pushing forward the CSR approach at our department, whereby he inspired many young and talented people to pursue this exciting path. Without his contribution and support, this dissertation would have never originated. I am also immensely grateful to Dimitris Xygalatas, the former director of LEVYNA, who introduced me to experimental approach, provided guidance in designing and reporting experiments, and shared all the joy and frustration of scientific investigation that every experimentalist is overly familiar with.

My further thanks go to all co-authors on the publications that are part of this dissertation. They all helped me to pursue my interests by offering always new research avenues that I eagerly explored, not to mention all their hard work with helping to design and conduct experiments. Specifically, I thank Radek Kundt for teaching me how to argue more carefully; Jan Krátký for a joint interest in stressing people and long evening discussions about rituals; and John Shaver for infecting me with his passion for field work. Furthermore, I would like to thank the whole LEVYNA team for the great intellectual environment that they created in our office. And, above all, I am also grateful to all the aforementioned people for their friendship that made our cooperation very enjoyable.

Last but not least, I wish to thank my family for supporting me during the long journey that my doctorate became. My beloved wife Agata was always there for me, despite the fact that it often meant spending countless hours and nights working next to each other. Even in our free time, she always provided a “reality check” of my abstract ideas, discussing them with a healthy dose of skepticism, yet sharing the enthusiasm for science and understanding human behavior. I am extremely grateful that Agata helped me to pursue the career I chose.

iii

PREFACE

The topics investigated in this dissertation are by their very nature complex and multi-layered, and as such require a multi-disciplinary exploration. This exploration entails studying scientific literature ranging from physics, over neuroscience, to sociology; working in a field to collect contextualized data; designing and conducting experiments; using advanced scientific equipment to measure behavioral and physiological processes; programming algorithms to extract and process the obtained physiological signals; performing statistical analyses to produce meaningful results; and drafting papers to be submitted in scientific journals. Although I have attempted to master all those skills, it would be impossible to conduct all the presented research without interdisciplinary cooperation and hard team-work. Thus, the current doctoral thesis is a result of group endeavor rather than of an individually working scholar, and while this approach is widely used in scientific departments, it is atypical for humanistic ones. As a consequence, a dissertation composed of team papers constitutes a precedent case at the Department for the Study of Religions at Masaryk University in Brno, and requires from the readers a slight change in perspective on what is being counted as a significant contribution. By presenting this doctoral thesis under my name, I believe that my share of work was not insufficient or inferior. Connecting the sciences and the humanities in the study of religion requires a lot of work atypical for humanities, and I had to spend a significant amount of time learning and implementing all the aforementioned procedures in my research. Nonetheless, readers can judge the sufficiency of my contributions from the following description of individual chapters.

The introduction and discussion chapters are partially based on a text that I have drafted for a prepared publication to be co-authored with Radek Kundt. The manuscript titled “The Cognitive Science of Religion: Looking for the Middle Ground” originated from discussions with Radek Kundt who also provided commentaries on previous versions of this manuscript. The earlier drafts of this manuscript were further improved by commentaries from Richard Sosis and Dimitris Xygalatas, and by language corrections from Vanessa Petroj. The introduction and discussion chapters themselves benefited from commentaries by Aleš Chalupa and proof reading by Lloyd Black.

The second chapter is composed from three papers. The first publication (to be printed in 2017) is titled “Religion and Prosociality” and was authored by Xygalatas and Lang. It is a

iv

chapter in a volume titled Macmillan Interdisciplinary Handbooks. Religion: Mental Religion, edited by Nicholas Clements, and printed in Farmington Hills, MI by Macmillan Reference USA. I drafted the chapter jointly with Dimitris Xygalatas, and we provided commentaries on each other’s writings. Furthermore Nicholas Clements provided a commentary on an earlier version of this chapter.

The second paper titled “Music as a Sacred Cue? Effects of Religious Music on Moral Behavior” was published in 2016 in Frontiers in Psychology (vol. 7, no. 814, pp. 1–13) by Lang, Mitkidis, Kundt, Nichols, Krajčíková, and Xygalatas. I designed the experiment together with Panagiotis Mitkidis, Radek Kundt, and Dimitris Xygalatas, and data were collected by me, Radek Kundt, Aaron Nichols, and Lenka Krajčíková. I analyzed all the data and drafted the paper. The final manuscript benefited from comments and corrections from the other co- authors, Guy Hochman, Mark Aveyard, Brendan Strejcek, and an anonymous reviewer. I am further grateful to Mijal Bucay, Will Fedder, Laura Hamon-Meunier, Anestis Karasaridis, Danijela Kurečková, and Mehreen Mey for help with data collection. I would also like to acknowledge that this research was supported by the project “LEVYNA Laboratory for the Experimental Research of Religion” (CZ.1.07/2.3.00/20.048), co-financed by the European Social Fund and the state budget of the Czech Republic; the Faculty of Arts, Masaryk University; the “Technologies of the Mind” project at the Interactive Minds Centre at Aarhus University, financed by the Velux Foundation; and the Cultural Evolution of Religion Research Consortium, financed by the Canadian Social Sciences and Humanities Research Council (SSHRC).

The third paper titled “Lost in the Rhythm: Effects of Rhythm on Subsequent Interpersonal Coordination” was published in 2015 in Cognitive Science as an e-print before the final paper version, and was authored by Lang, Shaw, Reddish, Wallot, Mitkidis, and Xygalatas. I designed the study together with Daniel Shaw, Panagiotis Mitkidis, and Dimitris Xygalatas. I collected the data with the help of research assistants. Sebastian Wallot provided novel analytical techniques, and I used those techniques to process the behavioral data that were further analyzed statistically together with Paul Reddish’s help. Finally, I drafted the paper and other co-authors provided commentaries and corrections. This paper also benefited from commentaries provided by Chris and Uta Frith, Ivana Konvalinka, Rick Dale, and three anonymous reviewers. I am further grateful to Lenka Brichová and Dagmar Adamcová for help with data collection. This research and its authors were supported by 1) the project “LEVYNA- Laboratory for Experimental Research of Religion” (CZ.1.07/2.3.00/20.048), co-financed by

v

the European Social Fund and the state budget of the Czech Republic, and by the Faculty of Arts, Masaryk University; 2) the project “CEITEC—Central European Institute of Technology” (CZ.1.05/1.1.00/ 02.0068) from the European Regional Development Fund; 3) the Marie-Curie Initial Training Network, “TESIS: Towards an Embodied Science of InterSubjectivity” (FP7-PEOPLE-2010-ITN, 264828); 4) the Velux core group “Technologies of the Mind”; and 5) the Social Sciences and Humanities Research Council of Canada-funded Cultural Evolution of Religion Research Consortium at the University of British Columbia.

The third chapter consists of two publications. The first one is titled “Effects of Anxiety on Spontaneous Ritualized Behavior” and was published in 2015 in Current Biology (vol. 25, no. 14, pp. 1892–1897) by Lang, Krátký, Shaver, Jerotijević, and Xygalatas. I designed the study together with all the other co-authors, provided novel analytical techniques, analyzed the data, and drafted the paper. Other co-authors provided comments and corrections. The manuscript also benefited from the comments of Chris and Uta Frith, Pierre Liénard, Geoffrey North, and four anonymous reviewers. I am further grateful to Daniela Kurečková and Anestis Karasaridis for help with data collection. This work was funded by the Laboratory for Experimental Research of Religion (CZ.1.07/2.3.00/ 20.048), co-financed by the European Social Fund and the state budget of the Czech Republic, and by the Faculty of Arts, Masaryk University. Other co-authors acknowledge support by the Velux core group ‘‘Technologies of the Mind’’, the Social Sciences and Humanities Research Council of Canada-funded Cultural Evolution of Religion Research Consortium and the University of British Columbia, and Royal Society of New Zealand Marsden Fund Grant (ID: VUW 1321).

The final publication in this dissertation is titled “Anxiety and ritualization: Can attention discriminate compulsion from routine?” and was published in 2016 in Communicative & Integrative Biology (vol. 9, no. 3, pp. e1174799) by Krátký, Lang, Shaver, Jerotijević, and Xygalatas. The paper was jointly drafted by Jan Krátký and me, and benefited from comments supplied by the other co-authors and by František Baluška.

Finally, I would like to kindly ask readers to bear in mind that experimenting and publishing usually does not go as one have planned, and the outcomes of these processes are often unpredictable. Composing a doctoral thesis from already published papers may sometimes look like a collection of scattered papers, especially when individual papers had to conform to a journal-specific styles and requirements. Hopefully, I achieved to provide a common link that joins the publication together into a meaningful doctoral thesis.

vi

INTRODUCTION

Religion has traditionally been one of the core topics for multiple social scientific disciplines, as reflected in the classic works such as those of Durkheim (1912/1964), Freud (1961), Tylor (1871), and Weber (1920/2002). While these grand theories brought significant scholarly merit to the understanding of religion, their interpretative lenses were often focused on single-cause explanations, thereby preventing formulation of theories compatible across disciplines. However, none of these grand theories achieved to dominate the study of religion either, because any such grand theory will always encounter plentiful counter-examples that do not exactly fit its predictions (Engler & Gardiner, 2010). Religion is an overwhelmingly complex phenomenon. Consequently, grand theories explaining religions, religiosity, and religious behavior were abandoned, and scholars have started to specialize in the study of specific traditions and symbolical systems, often embedding this particular knowledge in their discipline-specific interpretative frameworks. The increasing specialization of academic disciplines during the 20th century gave rise to many interpretative and explanatory frameworks that are encapsulated and often mutually exclusive. While this situation illustrates the multi- disciplinary nature of the study of religion dictated by the complexity of its object, the lack of cross-talk between disciplines significantly hinders advancements in the scholarly understanding of religion.

This issue has been further amplified by the entry of the life sciences into the study of religion. More so than in the humanities, the allure of biological causality and modern technology led scientists to initially reduce all variability in religions to a single gene or brain area (Albright, 2000; Hamer, 2005; Newberg, 2001). By proposing single-cause explanations, such as “the God gene” or “the God module” in the brain, the sciences brought a competing explanatory framework that was radically incompatible with the work in the humanities, which usually appreciate the contextualized variability of religions (Burlein, 2012). While these simplistic approaches were subsequently dismissed by the scientific community (e.g., A. W. Geertz, 2008; Schjoedt, 2009), and new research focused on the study of brain mechanisms related to particular religious phenomena (Inzlicht, McGregor, Hirsh, & Nash, 2009; McNamara, 2009; Schjoedt, Stødkilde-Jørgensen, Geertz, & Roepstorff, 2009), these explanatory frameworks are usually not compatible with the humanities either. Contrary to some incompatibilities between disciplines within the humanities, the sciences bring a suite of

- 1 -

ontological and epistemological assumptions that divide disciplines studying religion even more.1

For example, some scholars posit that the sciences and the humanities study different ontological aspects of reality, namely the physical and un-physical, and thus cannot be bridged (Shweder, 2011). While we could, in theory, maintain separate explanatory models for different ontological levels, problems occur when these two approaches aim to explain the same phenomenon or interpret the same data, such as a religious experience (Taves, 2011). Situating causal explanations of religious thought and behavior into the domain of physically anchored biological processes undermines theories operating with conscious human motivations, values, and intentions. Indeed, explaining particular religious experience as a consequence of complex inner workings of the evolved brain seemingly challenges classical ideas about humans as agents with qualities like free-will and socially constructed identities. Thus, it would seem, not only the sciences and the humanities cannot be bridged to work together, but they are often in explanatory contradiction.

However, maintaining the incommensurability of the scientific and humanistic perspectives only further propagates the explanatory gap and leads to exponentially diverging research trajectories manifested as a growing volume of competing explanations. This is not to say that competing explanations are not desired in academia, quite the contrary. However, if the competitiveness arises from metaphysical disagreements rather than testable theories (Dupré, 1996), research progress is not warranted. Consistency of theories should be the default principle of rationality (Slingerland & Bulbulia, 2011), and therefore finding the middle ground between the sciences and the humanities should be of high interest to scholars studying religion.

Moreover, explanations integrating the humanistic and scientific perspectives afford a complex understanding of studied phenomena. Instead of studying biological and culturo- historical layers separately, joining them into one research program can be fertile for both lines of research. Mutual inspiration and corroboration of theories are the main perks potentially gained by this fusion. While the thick layer of cultural observation, description, and interpretation may inspire questions about the structure and function of the human mind, the evolutionary and mechanistically focused scholars of religion can offer new insights into particular cultural events based on underlying biological processes. Furthermore, theorizing

1 The gap between the sciences and the humanities is not limited only to the study of religion; the incompatibility of these “two cultures” has been identified long since (Snow, 1961).

- 2 -

about the joint workings of culture and biology can spark a cascade of self-correcting processes by giving scholars an opportunity to test their theories against theories from different explanatory levels.

Therefore, my goal in this dissertation is to show that scholars now have new methods and analytical tools that allow them to join the sciences with the humanities. In other words, I argue that combining humanistic observation, description, and understanding with rigorous scientific methodologies focused on empirical evidence can bring new insights to the multifaceted phenomenon recognized as religion. Instead of studying religion multi- disciplinarily (i.e., many disciplines in parallel), scholars can now study religious phenomena inter-disciplinarily (Kirkpatrick, 2011).

To support my inter-disciplinary argument, I first describe the roots of the explanatory gap between the sciences and the humanities, and argue that the Cognitive Science of Religion (CSR) can act as a platform for the facilitation of joint research programs. By employing the mechanistic approach, CSR has a potential to relate methods and results from different disciplines in order to create coherent knowledge of religious phenomena. Next, I review the theories of Durkheim and Malinowski, which provided inspiration for a large body of humanistic theorizing, and present my original CSR research exploring the hypotheses put forward by these classics. Durkheim and Malinowski were chosen because they both enjoy respect of contemporary scholars and their theories are still relevant, and also because they focused on different levels of religious phenomena, namely the social and the individual. Thus, I illustrate how CSR research can address humanistic theories on different levels of complexity.

Specifically, I present three studies related to Durkheim’s hypothesis of religion as a social glue, from which two discuss the contextualized effects of ritual behavior on cooperation (Lang et al., 2016; Xygalatas & Lang, 2017) and one investigates a specific cognitive mechanism mediating the effects of ritual behavior on social bonding (Lang, Shaw, et al., 2015). Furthermore, I present two studies exploring the relationship between anxiety and ritualized behavior on the individual level as proposed by Malinowski (Krátký, Lang, Shaver, Jerotijević, & Xygalatas, 2016; Lang, Krátký, Shaver, Jerotijević, & Xygalatas, 2015). These five studies demonstrate that the scientific method can bring empirical evidence to classical questions in the study of religions, and thus answer long-standing questions in the field of religious studies. Finally, I discuss the merit and shortcomings of this approach and suggest future directions that may advance the CSR research program.

- 3 -

The Cognitive Science of Religion

Two Cultures and the Divide between the Humanities and the Sciences In the past 25 years, CSR has started to develop an interdisciplinary framework for studying religion and formed a novel academic discipline focused on using scientific methods in the investigation of religious phenomena.2 Given that CSR scholars are predominantly rooted in religious-studies and anthropology departments, they are well aware of the intrinsic problems present in the study of religion. For example, CSR scholars acknowledge the socially constructed nature of the term religion and its problematic definition (Bulbulia & Slingerland, 2012; Franek, 2014). But in contrast to classical scholars of religion that work multi- disciplinarily, CSR scholars embrace the inter-disciplinary character of cognitive science that offers ways to combine methods and results from other sciences and, potentially, unify evidence from multiple levels of research (Boyer, 2001; Bulbulia & Slingerland, 2012; A. W. Geertz, 2010; Slingerland, 2008; Sorensen, 2005; Taves, 2010). In other words, CSR offers a framework for vertical integration of the sciences with the humanities in the study of religions (Slingerland & Collard, 2011).

To appreciate the CSR proposition of integrating the humanities and the sciences, we first need to understand the roots of the humanities-sciences divide. Therefore, in the next paragraphs, I identify four interconnected areas of tension where humanists and scientists occupy opposite ends of attitudinal spectra. Whereas I assume that most scholars of religion find themselves rather close to the middle of the presented spectra, it is useful to explicitly describe the extreme positions in order to expose the juxtaposition between the two cultures (Snow, 1961).

First, scientists programmatically use reduction to decompose studied phenomena into simpler parts that can be described by lower-level disciplines and afford a detailed naturalistic understanding of the components (Slingerland, 2008). However, reducing cultural diversity into universal natural laws discards the “thick” variability humanists study in the first place (C.

2 Rather than describing the various streams of CSR research, I focus here on the underlying theory of the CSR approach and on ways it suggests to ameliorate the humanities-sciences divide. While CSR is not a monolithic field of study, the meta-theoretical principles are mostly shared among the main proponents as illustrated in this dissertation by the work of Edward Slingerland, Pascal Boyer, or Joseph Bulbulia. Thus, I describe CSR as a specific epistemological proposition rather than a differentiated academic discipline. Readers interested in various examples of CSR research are recommended to explore the following papers: Konvalinka et al. (2011); Schjoedt et al. (2009); Shariff & Norenzayan (2007); Sosis & Ruffle (2003); and Xygalatas (2013).

- 4 -

Geertz, 1973), and some reductionist propositions even threaten the existence of the humanities and social sciences themselves. The New Wave Reductionism (Bickle, 2003), for example, proposes to reduce higher-level theories to lower-level ones because this approach asserts that all phenomena can be explained by reference to material underpinning. Such an intertheoretic reduction stipulates that psychological theory should be reduced to neuroscientific methods and vocabulary, because neuroscience can sufficiently describe all phenomena studied at the psychological level (Bickle, 2003). When applied to the study of religion, the New Wave Reductionism proponents would claim that one does not need religious studies because explaining religion can be achieved through neuroscientific methods. With the advance of science, moreover, even the neuroscientific level should be surpassed by lower-level sciences such as cellular biology (Bickle, 2003).3

Second, even the milder forms of reductionistic approaches that do not eliminate higher-level disciplines provoke disagreement between the sciences and the humanities. While many scientists assume that the mind is identical with the brain (Slingerland, 2008; Wilson, 1998, p. 66),4 humanists often assume the immaterial reality of the mind: “…there is something more, let's call them the unphysical realities, to which one should show respect because they are worthy of it” (Shweder, 2011, p. 68). By disassociating culture from biology, humanists often assume that the mind is independent of matter (Pinker, 2003). This dualistic view, however, is hard to reconcile with the scientific approach to the study of human behavior that seeks causal patterns within the material world. Edward Slingerland (one of the main spokesperson of the attempts to bridge the humanities-sciences divide) summarizes these two positions as “a Cartesian ghost in the machine” as opposed to a materialist view that “we are ‘little robots’ all the way down” (2008, p. 257). Although both are ontological propositions that cannot be clearly determined by the current state of knowledge, they influence research agendas in the in the most fundamental ways. Whereas scientists look for brain activations corresponding to particular behaviors, humanists study socially constructed worlds produced by minds independent of their materiality and, consequentially, of human nature (Berger, 1967, p. 7).

Third, scientists usually stress the role of genetic predispositions in behavior, but scholars in the humanities put emphasis on nurture; that is, on learning. Humanists do not deny

3 For a detailed critique of this claim see for example de Jong (2002); de Jong & Schouten (2005); and McCauley (2007). 4 That is to say, mind can be reduced to the brain.

- 5 -

that people are born with biological predispositions, but these predispositions are usually reduced to primitive instincts and reflexes that are equivalent for all humans (for a criticism of this approach see Tooby & Cosmides, 1992). Thus, these biological factors cannot explain the ample inter- and intra-group cultural variability. If the variability cannot be explained by inborn predispositions, it has to be learned (C. Geertz, 1973). Humanists assert that mental worlds are constructed and formed by socialization and language, and that the influence of culture extends even to cognition and emotions (Kay & Kempton, 1984; Shweder & Sullivan, 1993). By stressing the importance of nurture (vs. nature), “nurturists” see the mind as a content-free processing device that is organized by culture (for a critique see Slingerland, 2008, p. 15; Tooby & Cosmides, 1992). Scientists, on the other hand, investigate in-born predispositions and different rates of genetic or epigenetic expressions in specific environments and their influence on everyday human behavior. While these natural predispositions cannot fully explain cultural diversity, they point to shared patterns underlying human behavior in specific contexts (Boyer, 1994). For example, due to human cognitive abilities, the belief that the world was created in seven days by an omnipotent agent has certain transmission advantages over the belief that the world was created in 10-37 seconds ex nihilo. While the former is intuitive given the cognitive architecture of human mind, the latter requires a good amount of abstract rational thinking. Humanists, on the contrary, would assume that difference in acquiring these two beliefs will be purely contextual. If one happened to be born into a family professing one of the Abrahamic religions, he or she should acquire the former belief. If parents would be proponents of the Big Bang theory, a child would acquire the cosmological theory with the same ease as the story of creation by an omnipotent agent.5

Finally, while scientists mostly subscribe to the idea of knowledge accumulation, humanists usually reject the Enlightenment project and similar objectivist approaches to human epistemology (Dupré, 1996). In the post-Kuhnian world (Kuhn, 1962/1996) where methodologically “everything goes” (Feyerabend, 1975), the idea of gradual accumulation of knowledge is refused together with Popperian stress on falsification (Popper, 1935/2005). Instead of gradually accumulating empirical evidence, scholars in the humanities use mostly description and interpretation and their theories are often not concerned with causality or are not hypothesis driven.6 For example, Clifford Geertz suggested that scholars should interpret

5 Interestingly, Debroah Kelemen and colleagues demonstrated that even physicists show biases toward unwarranted teleological explanations under time constraints (Kelemen, Rottman, & Seston, 2012). 6 For a criticism of these two issues see Bulbulia, Wilson, & Sibley (2014); Penner (2000); and Stark (1997).

- 6 -

social action as a text and discard causal explanations as irrelevant (1983, p. 6). 7 Social scientists should, according to Geertz, abandon attempts to study society as “social physics” and turn to interpretative methodology (p. 23). By stressing interpretation, humanistic theories usually cannot be tested or falsified (Boyer, 2011). This leads to two very different approaches to the creation of new knowledge: understanding vs. explaining, typically represented by the German terms Verstehen vs. Erklären (Slingerland, 2008, p. 226).8

In summary, when studying religion, the humanists and the scientists have often incommensurable ontological and epistemological assumptions that manifest themselves as disparate research agendas. Without willingness to cross the middle point between these extremes, research programs on religion will suffer from the disparity of fragmentary approaches manifested as an emphasis on variability and nurture on one hand, and universality and nature on the other hand. The CSR approach strives to bring these two sides closer to the middle ground by proposing a methodological framework that would afford translating concepts between disciplines and a mutual agreement on the type of data relevant for these concepts. Crossing the middle ground between the interpretative and explanatory approaches can thus be understood as a vertical integration of academic disciplines, sometimes coined as a second-wave consilience (Slingerland & Collard, 2011; Tooby & Cosmides, 1992). The paths to vertical integration ameliorating the four points of tensions are described in the next subchapter.

CSR and Consilience In 1998, Edward O. Wilson published a highly influential book called Consilience: The Unity of Knowledge, where he argued that consilience is the most important scientific endeavor. For Wilson, consilience is “the linking of facts and fact-based theory across disciplines to create a common groundwork of explanation” (p. 8). Good explanations, according to Wilson, are those

7 This issue may partially stem from human reluctance to derive the particular from the general. In his book Thinking Fast, Thinking Slow Daniel Kahneman (2011) calls this bias “the pundit illusion” and suggests that experts often overestimate confidence in their professional intuitions gained from experiences with particular cases. This intuition, however, usually performs poorly in comparison with statistical algorithms based on thousands of cases. According to Kahneman, human minds have a tendency to overemphasize particular cases, regarding them as different, special, or unique, whereas in reality those cases are a subject to causal factors and randomness as any other cases. Thus, combining intuition with statistical algorithms might be a more valid approach toward theory construction. 8 These terms are derived from Wilhelm Dilthey’s distinction between the Geistewissenschaften and the Naturwissenschaften. For a further clarification and application of this distinction in modern sciences see Talmont-Kaminski (2014, pp. 23-26).

- 7 -

that are consistent with each other, and scientists should be careful to align their explanations with the causality principles of disciplines on lower explanatory levels (pp. 58 - 60). Assuming the hierarchical division of academic disciplines postulated by Aristotle and later elaborated by the French encyclopedists, one should be able to move smoothly from the highest-level disciplines to the lowest-level ones.

To illustrate the principles of consilience, Wilson (1998) describes his research program on the mechanisms of ant distress communication. Initially, Wilson speculated that ant distress signals are of chemical nature because ant communication usually takes place in areas without light, and the ant species studied by Wilson could not communicate by sound. He dissected an ant’s glands potentially containing alarm pheromones and observed what happens when he exposed other ants to these pheromones. Upon detecting the pheromones, these ants were galvanized, racing back and forth, confirming Wilson’s expectation. Wilson further studied the chemical composition of these pheromones and identified active substances called alkenes and terpenoids. When he presented these pure substances to ants, their behavior was again galvanized. Using known physical properties of these substances, Wilson further confirmed with mathematical models that alkenes and terpenoids can indeed be used as signals. The strength of this approach is the mutual corroboration of his hypotheses – if Wilson had not confirmed that ants have substances serving the signaling function, his theory would be invalid. The main point of this example, however, is that Wilson stepped out of usual entomology research boundaries and joined forces with researchers from other explanatory levels. By decomposing the studied phenomenon of ant communication into lower-level phenomena, he was able to corroborate his initial intuition gained through observing ant behavior.

Of course, with a growing complexity of studied phenomena, the consilient approach will start reaching its limits because the current state of knowledge does not allow one to jump from an observation of a particular cultural event to, for instance, a detailed neuroscientific description of this event. Indeed, Wilson admits that linking the sciences and the humanities is the most difficult endeavor (1998, p. 8). But the consilient approach is a useful starting point for endeavors in bridging the sciences with the humanities because it illustrates that willingness to work with people from other explanatory levels is absolutely crucial for any such endeavor. If scholars in the humanities would continue to maintain that human biology does not exert influence over culture and religion, the humanistic and scientific perspectives cannot not be aligned.

- 8 -

But CSR scholars also recognize that to find the middle ground, scientific approaches should not reduce the studied phenomena only to biological or neuroscientific levels of explanation (Wildman & McNamara, 2008). The method of reduction proposed by CSR scholars is opposed to eliminative intertheoretic reductionism (Slingerland, 2008; Slingerland & Bulbulia, 2011; Wildman, Sosis, & Mcnamara, 2012). CSR proponents do not see humans as genetically pre-programmed robots with closed behaviors that can be understood only by studying physiological processes (Boyer, 2011). Instead, CSR scholars employ a so called building-block or fractioning approach (McCauley, 1999; Taves, 2010, 2011; Whitehouse, 2005) whereby religious phenomena are disassembled to building blocks, which are not specific to religion and may be more easily defined and studied by the sciences (Boyer, 2001, p. 202).9 These blocks are usually framed as evolved mental mechanisms or modules (Whitehouse, 2005, 2008), and studying them in isolation uncovers what function they perform and what role they play in a complex system without necessarily postulating some religious faculty, brain structure, or genes for religiosity (McCauley & Cohen, 2010; Schjoedt, 2009). Furthermore, reduction opens doors for scholars from other disciplines who can use their expertise to create knowledge about a specific phenomenon (e.g., memory) placed within a religious context (Barrett, 2011). But this does not mean that every scholar of religion should disassemble religious phenomena and study the constitutive parts with scientific methodology, quite the contrary. As Slingerland and Collard put it:

“…even if every researcher in the humanities immediately embraced consilience with the sciences, the vast majority of humanistic work would still consist of what we are calling “horizontal analysis”: analyzing phenomena by tracing out connections between entities native to emergent levels of explanation. This is of course the case in any field of analysis, scientific or otherwise: organic chemists spend most of their time exploring connections that make sense only at the level of organic molecules, and even the most reductive evolutionary approach to poetry will necessarily focus primarily on problematic and modes of analysis native to the phenomenon of poetry” (Slingerland & Collard, 2011, p. 25).

Put simply, the work of scientists and humanists should continue primarily in their own fields, but they should share a common perspective that would allow them to employ their particular expertise and skills when investigating a shared topic.

9 See Asprem (2015) for an example of decomposing the term ‘esotericism’.

- 9 -

Some CSR scholars further emphasize that using reduction cannot explain everything about higher-level phenomena because those phenomena exhibit emergent properties (A. W. Geertz, 2010; Slingerland, 2008). Phenomena have unique organization on each complexity level, and this organization cannot be predicted from lower levels (Mitchell, 2009). The emergence principle connects the sciences and the humanities by limiting the explanatory possibilities of excessive reductionism and by stressing the unique validity of higher-level explanations.10 Applying the emergence principle on the body-mind dualism, CSR scholars acknowledge that the human mind has a biological basis that forms its functioning, but they also acknowledge that from this biological basis alone one cannot predict the mind’s full complexity. Thus, there is a lawful body-mind relationship that allows to some extent connect the mental with the physical. Instead of proposing exclusive ontological levels, the CSR approach relies on the so far accumulated evidence documenting the mind’s linking to physical brain structures (Pinker, 2003; Slingerland, 2011).

The emergence proposed by CSR scholars recognizes that higher levels are constrained by the laws of lower levels (e.g., humans cannot reproduce without passing on their genetic information), but the highest organizational level (human being) has its own laws and causal explanations. For instance, genetic information alone cannot explain why someone chose to worship one deity over another, and genetics will have only very little to say on this issue (Sosis & Bulbulia, 2011). But the developmental calibration of mental mechanisms engaged in religious belief will be influenced by genetic predispositions that are passed from one generation to another. This weak-emergence proposition allows for the decomposition of phenomena under investigation into less complex blocks (Slingerland, 2008), while simultaneously acknowledging that reduction cannot capture the whole complexity of studied phenomena. It affords understanding lower-level components individually, but also strives to find out how the lower-level phenomena interact and create the complex whole together. Since humanistic theories can capture unique complexity on higher levels, humanists can work together with lower-level scientists to find principles bridging different levels and thereby facilitate a complex understanding of observed phenomena.

A similar common ground can be found in the division between nature and nurture. While analyzing solely the influence of tempo, metricality, and pitch of a religious song would

10 Nonetheless, in its extreme form, emergence can be also understood as implying the body-mind dualism that further widens the explanatory gap. Strong emergence accounts posit that higher emergent levels (e.g., human mind) must be ontologically independent from lower levels (T. Nagel, 1974, 1986) and suggest that there are nonphysical qualities that cannot be studied scientifically.

- 10 -

not explain why it has particular words that were chosen over others, interpreting the symbolic meaning of such a song should (at least implicitly) take into account that different types of music have different effects on the brain and physiology. Meaning is not only learned, but it is also embodied (Slingerland, 2008). Likewise, an analysis of the symbolic meaning of a particular deity’s depiction should take into account that the mind has certain constraints when transmitting symbols and that some religious ideas will be transmitted easier than others (Boyer, 2000, 2001). For example, if the most common depiction of a deity in some religious society would be highly abstract (e.g., a combination of geometric patterns), one would need to posit some strong legitimization forces accounting for the depiction’s spread. If these forces could not be found in the historical records, some other explanation would be needed. Although very simplified, this example illustrates that aligning the knowledge of cognitive constraints with the knowledge of historical context may yield more plausible interpretations. Instead of emphasizing nurture or nature, this approach postulates nature via nurture (Ridley, 2003).

Finally, the CSR approach also attempts to diminish the distinction between understanding and explanation or, expressed by the German terms, Verstehen and Erklären. In his critique of the Verstehen approach, Slingerland argues for a pragmatic approach to the study of religion: by appreciating physical causation expressed in quantitative form, scholars can provide empirically testable claims and explanations (2008, p. 248).11 In the history of science, this approach turned out to be very productive as documented by, for instance, modern medicine or the development of information technology. Overcoming the negative heritage of naïve objectivism, the CSR approach offers formulating hypotheses and testing them against empirical evidence. However, these hypotheses are driven by the humanistic insights and understanding, which is crucial when selecting, for example, contextually relevant primes or forming theories of studied phenomena (Lawson & McCauley, 1990).

In summary, the CSR approach attempts to vertically integrate the sciences and the humanities in order to create consilient knowledge about religion. This integration asks humanists to relax their assumption that the mind is a blank slate and all its content is learned, and instead to embrace that biological processes have a formative influence on human culture.

11 For example, Schauer (2011) analyzed the popularity of motifs on painted pottery in ancient Greece. He was able to confirm some of his hypotheses about the changing popularity of various gods by looking at the frequency of their depictions. But more importantly, he also observed some unexpected variation in the historical records that demanded a new explanation. Specifically, the goddess Nike was very popular as a motif for pottery painting in the 5th century BC, yet this popularity was not predicted by any existing theory. Whereas traditional theories of gods’ importance were derived from the analysis of individual samples or specific depictions, a frequentist analysis challenged these theories and warranted further research.

- 11 -

The scientifically-minded scholars of religion, on the other hand, are expected to acknowledge the emergent properties of hierarchical levels and refrain from eliminating higher-level explanations. Nevertheless, whereas recognizing that the humanistic and scientific theories can be aligned is a crucial first step for vertical integration, the specific integrative process needs principles of inter-level translations affording a joint study of religious phenomena. The next sub-chapter describes the difficulties that any interdisciplinary framework faces and suggests that the mechanistic approach can help specify methodological principles for level transitions.

CSR and the Mechanistic Approach The first difficulty of the integrative approach is translating concepts between different explanatory levels (Bechtel & Hamilton, 2007). As noted by Wilson, the natural sciences and the humanities lack a common language (1998, p. 137). A solution to intertheoretic translations is offered for instance by classical reductionism as proposed by Ernest Nagel (1961), who formulates bridge laws between different explanatory levels and assumes that the phenomenon under investigation will have the same identity between different levels. It follows that lower- level explanations need to contain all laws that are present also on higher levels. Only then the phenomenon will have the same identity across levels.12 However, one might wonder how this type of reduction would work on higher-complexity levels; namely, humans. Translating complex terms like “the awe feeling” to cognitive or neuronal levels would require establishing what the feeling of awe means on both these lower levels and determining the full neuronal and cognitive operations mediating this feeling. Additionally, since CSR scholars renounce the eliminative reductionism and embrace emergence (Slingerland & Collard, 2011, p. 36), translating the awe feeling from a neuronal to cognitive level will require specifying the rules for emerging complexity between these two levels. Without understanding the emergence of complexity, reduction would only lead to admitting that lower-level explanations are just simplifications of what emerges on higher levels.

Second, it is not exactly clear what levels CSR models should bridge. Classically, the levels of explanation are conceived according to scientific disciplines (Bechtel & Hamilton, 2007), although, admittedly, this delineation is very raw because there are no clear boundaries

12 An exemplary case is the Boyle-Charles’ law that describes the reduction of the term “temperature” as it appears in classical thermodynamics to “mean kinetic energy of particles” that can be used in statistical mechanics (Bechtel & Hamilton, 2007). In this formulation, a physical law that is valid on one explanatory level can be equated with a physical law on another explanatory level (given certain boundary conditions). In other words, temperature is the mean kinetic energy of particles.

- 12 -

between disciplines, and many disciplines usually deal with several layers of complexity. For example, psychology encompasses research on interpersonal communication, language processing, or the interaction between visual input and the visual cortex. Looking at a particular ritual, scholars might employ a psychological perspective to study the ritual action patterns and hypothesize, for instance, about their functions on anxiety alleviation as proposed by Malinowski (1948/1992). But it is ambiguous whether one should use only observation and surveys to assess the perception of anxiety, or whether to also employ the study of hormonal mechanisms, such as the corticosteroids, that may imply stress. While all of these methods can be regarded as assessing a psychological phenomenon, the explanatory levels afforded by these methods will be different. Scholars aiming for a useful vertical integration need to specify the important levels of complexity that will be incorporated into multi-level models and delineate ways to recognize them rather than parse studied phenomena according to historically given disciplinary divisions.

Third, when building scientific models of religious phenomena, a lot of variability crucial for models’ functioning will be lost. For example, when modeling the hypothesized anxiolytic effects of ritual, is it important whether one performs rituals alone or in a collective? At home or in the temple? Does the elaborateness of ritualized movements matter? Is the model applicable to hand gestures and utterances alike? These questions are crucial when including the thick-description levels into the models of religious phenomena because the emergent properties of complex phenomena depend on contextual diversity. CSR models need to consider environmental variability that may potentially change the workings of some of the model’s parts. On the other hand, including environmental variability into the model would decrease the model’s generalizability, if not explanatory power, thus presenting researchers with a dilemma whether to strive for contextuality or universality.

These three sets of questions will arise in any integrative program, and a successful approach needs to assert an epistemological framework for crossing multiple levels of investigation. The theory of explanatory pluralism can offer a possible solution to the aforementioned problems of the multi-level study of religion. Explanatory pluralism as proposed by a philosopher and CSR scholar Robert McCauley and a philosopher William Bechtel (McCauley, 2009; McCauley & Bechtel, 2001) emphasizes a joint development of theories on different explanatory levels. To facilitate translations from one level to another, McCauley and Bechtel (2001) proposed the concept of heuristic identities that can function across different levels of scientific investigation. Instead of formulating strict rules that create

- 13 -

intertheoretic identities as in classical reductionism (mental state A is always equal to neuronal activation X; see Schaffner, 1967), heuristic identities are only hypothetical and have exploratory function, which can fuel a joined research of multiple disciplines. Once a heuristic identity is established, researchers can ask what structures mediate a phenomenon and why, or which parts of the structure have to be in place in order for the phenomenon to occur. The advantage of this approach is an accumulation of data that can help modify theoretical models on different explanatory levels.13

The heuristic identity theory is a first step toward determining the method of bridging different explanatory levels. Additional specification can be gained by employing Bechtel’s mechanistic approach (2008). At the heart of this approach is the notion that humans are complex systems, in other words, systems consisting of many components that interact (Bechtel & Richardson, 2010, p. xlvi). The mechanistic analysis disassembles a phenomenon into its constitutive parts and operations, and creates models of their organization (Bechtel & Hamilton, 2007). The elements and operations are specified on one level of complexity, but behavior of the whole mechanism is observed on a higher level. This implies that the mechanistic components have their own specific functions and that mechanisms can be part of a more complex mechanism that realizes a phenomenon on yet a higher level of complexity (e.g., neurons mediating vision are composed of mechanistically functioning nuclei, mitochondria, cytoplasm, etc.). Thus, the observed phenomenon is on a higher level of complexity than the mechanism itself. This also means that explanatory levels are defined by organizational complexity rather than by the classical division of disciplines. Hence the transition between explanatory levels might be smoother.

The final product of mechanistic operations is considered to be the phenomenon on the top of the mechanistic hierarchy. Thus, it can be decomposed into simpler phenomena and these sub-phenomena can be decomposed into mechanisms (see Bechtel, 2008, Chapters 2-3 for examples of phenomenal decomposition of memory and mechanistic decomposition of vision).

13 For example, a heuristic identity can be proposed between a mental state A and a neuronal activation X, but only when an environmental factor Z is present. If the connection between A and X would be found without the presence of Z or there would be an additional neuronal activation Y, the heuristic identity would need to be reformulated. Such mutual corrections proved to be useful, for instance, in specifying different functions of the human visual cortex (Bechtel, 2008, Chapter 3; McCauley & Bechtel, 2001). While cognitive neuroscience showed that vision is mediated by a dispersed neuronal structure with multiple functions (as opposed to a previous assumption in psychology that it is a specific brain area), cognitive psychology used experimental approaches to specify these functions. Cognitive neuroscience, in turn, connected these functions with particular structures.

- 14 -

Going back to the anxiolytic effects of ritual, anxiety decrease after ritual performance can be considered as a phenomenon on the top of the mechanistic hierarchy. Using heuristic identity theory, the anxiety alleviation can be assumed to be mediated by less complex phenomena such as compulsion, behavioral rigidity and repetitiveness, or working memory overload (Eilam, 2006; Lang, Krátký, et al., 2015). These phenomena can be further decomposed into particular mechanisms. In the case of compulsion, for instance, these mechanisms can include adrenaline and noradrenaline release, increased heart rate and breathing rate, etc. Working memory overload can be decomposed into pupillary dilation, increased sympathetic activation, or hyperactivity in the prefrontal cortex and parietal areas. Importantly, these mechanisms do not work in parallel but create a system of interconnected feedback loops that give rise to the more complex mechanism or phenomenon. This approach does not look for simple correlations between higher levels and lower-level components, but rather for a specific organizational setting that gives rise to the studied phenomenon.

Bechtel’s mechanistic approach has several important consequences for the understanding of reduction:

“The quest to explain phenomena by identifying responsible mechanisms involves an inherent reductionistic commitment—such research decomposes the mechanism into its parts and their operations. But unlike more traditional philosophical accounts of reduction, the mechanistic perspective is not exclusively reductionistic, for it requires also recomposing the mechanism by taking into account the organization among the parts of the mechanism and situating the mechanism in its environment. Studying the parts and operations, organization, and situatedness of a mechanism requires different sets of investigatory tools. Understanding mechanisms requires a pluralistic approach that looks not just down but around and up” (Bechtel, 2009, p. 559).

Note that in this specification, reduction does not mean replacement of higher-level theories with the lower-level ones, but rather it means decomposing a mechanism (similarly as in the CSR building-block approach). However, the process of decomposition is more complicated than just correlating a higher-level phenomenon with, say, neuronal activations. Correlational approaches can highlight the mechanism’s parts employed in the emerging phenomenon, but it is the specific organization of the mechanism’s parts that needs to be described in order to understand the mechanism’s operations (Bechtel & Hamilton, 2007). The mechanistic emergent properties are given by organizational structure that is not implied by individual components. This is especially important when dealing with mechanisms that do not have a

- 15 -

linear composition, but involve non-linear organizational features such as negative feedback loops or oscillations (Bechtel, 2011; Bechtel & Abrahamsen, 2010).14

However, in order to decompose and then recompose a mechanism (thereby fully understand its functioning), one needs to know all its components and their organization. Nevertheless, when dealing with complex system (e.g., weather), scholars are not usually able to specify all of its components, let alone the components’ organizational principles. The resolution is to identify “the backbone” of a mechanism and its main constitutional components, and hypothesize a heuristic identity.15 Subsequently, the organization of selected components can be described and experimentally tested. While such a description may be very crude, it can be a useful approximation that has good explanatory power. That is, it is not necessary to describe a phenomenon’s complete mechanistic functioning in order to understand how the phenomenon emerges.

In summary, the mechanistic approach offers a way to connect the emerging levels of complexity by decomposing both phenomena and mechanisms into particularly organized components. The possibility to incorporate various components into the reconstruction of mechanistic operations gives scholars considerable flexibility when choosing between a mechanism’s universality (“the backbone” working across different contexts) and specificity (including peculiar components important in a particular context). When deciding between universality and specificity of the CSR models, scholars need to account for the costs and benefits of these two modeling strategies and determine whether their models are sufficiently flexible to address contextual variability, yet also sufficiently general to have good explanatory power. Since scientists strive for simplicity rather than for creating complex and difficult-to-

14 A thermostat is a good example of a negative feedback loop (Bechtel, 2008, p. 25). It feeds information from a phenomenon on a higher level (ambient room temperature) back to parts of a lower-level mechanism (furnace), thereby regulating functioning of its components (radiators). Simply putting individual parts into a feed-forward linear chain (furnace -> radiators) would not give rise to the higher order phenomenon of ambient temperature (temperature would be constantly rising). 15 For instance, the mechanism of agency detection that is hypothesized to play an important role in reputation management can be identified as “the backbone” of mechanistic operations producing the positive effects of religious primes on moral behavior (M. Bateson et al., 2006; Krátký, McGraw, et al., 2016). Whereas the decision to behave morally will be enormously complex and, strictly speaking, specific for each individual, the agency detection will play a significant role for most people. Assuming different sensitivity of the agency- detection mechanism among people, the intensity of its activation should be normally distributed so as to yield mediocre effects on moral behavior for most people, and extremely low or high effects for a few people. Thus, in most cases, it should be possible to identify the activation of agency detection in contexts with religious primes and determine its causal role in moral decision-making. Despite the complexity and specificity of a mechanism, scholars can expose its main components through common statistical methods.

- 16 -

understand models (Kahneman, 2011), the trade-off between specificity and universality needs to be carefully weighted.

In the following chapters, the mechanistic approach will be illustrated on theories suggesting that religious rituals can have positive effects on A) group prosociality and B) anxiety alleviation. From the point of view of the mechanistic approach, these theories can be regarded as proposing phenomena on the highest level of investigations. That is, anxiety alleviation and group prosociality are the products of mechanistic operations that can be disassembled into its constitutive elements and investigated on lower behavioral and physiological levels. Such disassembling and experimental testing of the individual components can bring initial evidence to support or undermine the proposed theories of religious ritual.

First, I will review Durkheim’s theory (1912/1964) on the role of religious ritual in the enhancement of group prosociality, describe additional theories supporting this assertion, and present empirical studies on the two of the hypothesized mechanisms. The second empirical chapter will comprise Malinowski’s theory that ritual behavior may serve to assuage anxiety (1948/1992). I will describe theories of ritual behavior that disassemble the specific ritual features into sub-phenomena possibly mediating the effects on anxiety, and, finally, present my original research testing one of the purported mechanisms.

- 17 -

EFFECTS OF RELIGION AND RITUAL ON PROSOCIALITY

A classical theory on the relationship between religion and moral behavior was introduced into religious studies by one of the founding fathers of sociology Émile Durkheim. As a rationalist, Durkheim put emphasis on the empirical research of social facts16 and institutions buttressing them (1895/1982), and especially on studying how social facts “…produce socially useful effects” (p. 134). Since social facts can be uncovered by looking at general trends in people’s behaviors through statistical lenses (p. 55), sociology in Durkheim’s view should be open to the empirical research of large-scale behavioral patterns and their causal underpinnings. From this perspective one can select a specific social fact relevant for religion, such as moral norms or belief systems (suggested by Durkheim as a humanist), and test its effects with current experimental approaches (i.e., scientific methodology). In the following paragraphs, I will focus on the aspects of religion and ritual that according to Durkheim promote pro-group behavior and present theoretical and experimental studies investigating mechanisms underlying these aspects.17

Durkheim made a significant imprint on religious studies by his book The Elementary Forms of the Religious Life (1912/1964). In his perception, religion is one of the main institutions reinforcing and perpetuating society’s customs, morality, and moods. The group’s rules of conduct are established as objectivized social facts through ritual gatherings, giving rise to a phenomenon Durkheim called collective consciousness. This collective mind has moral power that coerces individuals into normative behaviors, thereby strengthening group cooperation. The community-binding aspect of religious behavior is famously expressed in

16 In Durkheim’s view (1895/1982, pp. 51 – 52), social facts are part of collective mind, which means they are external and objective to an individual and, thus, can have coercive power. Yet, paradoxically, social facts are carried by individual minds. Without individuals, there would be no society and no social facts (pp. 127 – 129). Here Durkheim antecedes the emergence principle whereby the combination of individual minds give rise to something more than just the sum of parts, that is, to social facts. Although one might question the ontological reality of collective consciousness and social facts, they are useful proxies for the phenomena studied in this dissertation. Without necessarily postulating the ontological existence of social facts, Durkheim’s insights can be translated into empirical studies of human behavior. 17 As we note in the first study (Xygalatas & Lang, 2017), the terms religion and prosociality are notoriously hard to define. For the purpose of this dissertation, I understand the terms religion and prosociality as multifaceted phenomena that need to be operationalized by individual studies in order to construct meaningful claims (e.g., as self-reported religiosity or helping an unknown person). However, since the introduction to this chapter is mostly theoretical, I use the term religion in Durkheim’s understanding (see his definition in the text) and prosociality can be substituted with Durkheim’s understanding of morality.

- 18 -

Durkheim’s definition of religion: “A religion is a unified system of beliefs and practices relative to sacred things, that is to say, things set apart and forbidden — beliefs and practices which unite into one single moral community called a Church, all those who adhere to them” (italics original; 1912/1964, p. 47). The positive effects on group cohesion and morality was according to Durkheim one of the main functions of religions, a function that distinguished it from magic (p. 44). Summarized in Durkheim’s words: “…the practices of the cult, whatever they may be, are something more than movements without importance and gestures without efficacy. By the mere fact that their apparent function is to strengthen the bonds attaching the believer to his god, they at the same time really strengthen the bonds attaching the individual to the society of which he is a member…” (p. 226).18

This original Durkheim’s thesis was further supported by theories and empirical evidence amassed during the 20th century, showing that religious people are more prosocial and emphasize pro-group behaviors (Galen, 2012). Nevertheless, the relationship between religion and prosociality may not be as straightforward as imagined by Durkheim. While some religious practices instill bonds among people, other practices may be violent, destructive, and harmful for groups and individuals (Boyer, 2003). Thus, in the first presented article co-authored with Dimitris Xygalatas (Xygalatas & Lang, 2017), we survey the literature on religion and prosociality and offer more nuanced view of this relationship. While there seems to be a substantial connection between religiosity and self-reported prosocial behavior, this is not always supported by experiments investigating people’s actual behavior. Rather, we suggest, religious people are more prosocial only in specific contexts that remind them of group values and moral norms. Although not explicitly embraced by Durkheim, this idea resonates with his analysis of totemism and religious symbols.

For Durkheim, totemism was the archetypal religion that illustrates how society projects its values into totems of worshiped deities.19 The collective consciousness, Durkheim

18 Although Durkheim’s hypothesis on religion and prosociality was a profound insight, his theory is criticized from several fronts, for example, for ignoring the role of charismatic religious leaders as highlighted by Max Weber (Paden, 1992; Weber, 1919/2008) or accentuating the sacred-profane divide (Taves, 2011). My favorable review of Durkheim’s theory in this introduction does not stem from its uncritical acceptance, but more from an endeavor to find common ground between Durkheim’s original observations and modern scientific approaches. 19 According to Durkheim, totemism is the most primitive religion and as such may provide insights into basic fundamental elements of religion and their evolution over the past centuries (1912/1964, p. 48). While dismissing animism and naturism as more primitive religions, he was inspired by Frazer (1894) and Smith (1898/1995) to look for the fundaments of religions in Australian aboriginal societies. These totemic clans fitted into Durkheim’s idea of the most elementary social stratification, and thus served well in illustrating how the collective consciousness arises in the most primitive societies. However, Durkheim was criticized for imposing his theoretical frameworks on the ethnographic data from Australia (C. Geertz, 1973), with some researchers failing to find the sacred-profane division in the Aboriginal cults (Stenner, 1960).

- 19 -

suggested, is transformed into religious symbols that serve as external placeholders for collective representations of social facts (p. 231). Put simply, society represents itself as deity (1912/1964, p. 206). Since the totemic principle is the society, it embodies all its values and moral norms, and the totemic symbols are charged with special sentiments that have coercive powers. In Durkheim’s words: “We say that an object, whether individual or collective, inspires respect when the representation expressing it in the mind is gifted with such a force that it automatically causes or inhibits actions, without regard for any consideration relative to their useful or injurious effects” (italics original; 1912/1964, p. 207). Any symbol can take up this role because the sentiments are not inherent in the symbol itself, but are superimposed on it by the community (p. 229). For example, Durkheim discussed how the color black symbolizes mourning and as such provokes sad emotions (p. 219).20 This proposition was later elaborated by Alcorta & Sosis (2005) who described ways that sacred symbols are emotionally charged through rituals in order to motivate people toward certain behaviors. Furthermore, a large body of experimental studies using religious symbols as primes documented positive effects of exposition to religious stimuli on prosocial behavior (Shariff, Willard, Andersen, & Norenzayan, 2016; van Elk et al., 2015). Durkheim’s initial insight about the moral power of religious symbols thus seems to be corroborated, although the mechanism of coercion is currently described as associative learning, departing from Durkheim’s original suggestion of “emanating mental energy.”

To extend the priming studies and further investigate Durkheim’s hypotheses, I conducted an experimental study with a team of co-authors and investigated the role of sacred music in normative behavior. Thus, in the second presented paper (Lang et al., 2016), we explore whether people display less anti-social behavior upon hearing sacred music. Participants first heard religious music and afterwards were given an incentive to cheat for a monetary reward. In line with Durkheim’s suggestion, the results showed that the effect of religious music on moral behavior was salient only for religious participants, implying that symbols require group specific emotional charging in order to affect one’s behavior. Additionally, by using only religious music lacking words, we corroborate Durkheim’s assumption that symbols representing collective norms are arbitrary and do not need to necessarily symbolize supernatural agents or other deities. Similarly as in the example with the

20 Although Durkheim described this coercive mechanism as “emanating mental energy” (1912/1984, p. 209), he was also aware that symbols work through associations as documented by his assertion that symbols are united with specific sentiments in individual minds (p. 219). Translating the metaphysical claim about mental energy into more contemporary terms, such as association, helps to operationalize this effect into experimental designs.

- 20 -

color black, religious music seems to be associated with particular sentiments that motivate people to adopt behavioral schemas that are in line with group norms.

Furthermore, Durkheim was also concerned with the ways that societies maintain the collective consciousness, which represents the group and motivates people to pro-group behaviors. Since he understood that individuals have different motivations when it comes to collective action, Durkheim postulated a mechanism through which people can communicate in unison and synchronize their internal states to strengthen the group mind (1912/1964, p. 230). This mechanism was coined “collective effervescence,” what can be understood as a state of shared arousal through simultaneous unified verbalization and movements. To enter the state of collective effervescence, people often chant rhythmically, sing, dance, and move in synchrony:

“In fact, if left to themselves, individual consciousnesses are closed to each other; they can, communicate only by means of signs which express their internal states. If the communication established between them is to become a real communion, that is to say, a fusion of all particular sentiments into one common sentiment, the signs expressing them must themselves be fused into one single and unique resultant. It is the appearance of this that informs individuals that they are in harmony and makes them conscious of their moral unity. It is by uttering the same cry, pronouncing the same word, or performing the same gesture in regard to some object that they become and feel themselves to be in unison.” (Durkheim, 1912/1964, p. 230).

Indeed, the perception of moving and vocalizing in unison creates group entitativity21 as documented by research on synchrony and mimicry. Synchronous movements were shown to increase liking of the group (Chartrand & Bargh, 1999), cooperation (Reddish, Fischer, & Bulbulia, 2013; Wiltermuth & Heath, 2009), and general prosociality (Reddish, Bulbulia, & Fischer, 2014). Moreover, synchronous behaviors are often accompanied by shared arousal as asserted by Durkheim. For example, Konvalinka et al. (2011) showed that people participating in a fire-walking ritual synchronize their heart rates during emotionally exalted parts of rituals. Such emotional exaltation can be further translated into elevated feelings of happiness (Fischer et al., 2014) and prosocial behaviors (Xygalatas, Mitkidis, et al., 2013).

21 This term was originally coined by Campbell (1958) to express the group perception of being one entity. For further information see for example Fischer, Callander, Reddish, & Bulbulia (2013); Ip, Chiu, & Wan (2006); and Lakens (2010).

- 21 -

In this chapter’s third study, we explore a low-level mechanism that promotes group synchrony and the related effects on prosociality (Lang, Shaw, et al., 2015). Specifically, we disassemble ritual into its individual components and select rhythmical music as a dominant feature underlying group entitativity. Since rhythm is intrinsically connected with movement, we demonstrate that listening to rhythmic music positively affects motor coupling, despite the fact that it hinders joint action. In other words, rhythmical music is a powerful tool that helps synchronize individual movements and serves as an external attractor for individuals wishing to fuse with the group.

- 22 -

Religion and Prosociality Xygalatas, D., & Lang, M.

Introduction The connection between religion and prosociality has long been debated. From the ancient Greeks to modern philosophers, scholars have often pondered whether morality can exist without religion. One of the best-known ancient texts dealing with this question is Plato’s Euthyphro, which dates back to the 4th century BCE. In this dialogue, Socrates asks Euthyphro: “Is an action morally good because the gods command it, or do the gods command it because it is morally good?” In other words, is acting justly in our nature, or do we need religion to tell us what to do? Socrates did not get a satisfactory answer from Euthyphro, and the question has continued to be debated throughout the centuries.

Some of the founders of contemporary social science described religion as a motivational force that binds groups together, deters immoral conduct, and promotes altruistic behavior. For example, Émile Durkheim (1858–1917) saw religion as the “social glue” that holds society together. Auguste Comte (1798–1857) established the “religion of humanity”, so that secular societies could still continue to function harmoniously in the absence of traditional forms of worship. And in his Letter Concerning Toleration, John Locke (1632–1704) excluded atheists from the right to be tolerated, as he thought they could not be trusted to behave morally: “Lastly, those are not at all to be tolerated who deny the being of a God. Promises, covenants, and oaths, which are the bonds of human society, can have no hold upon an atheist. The taking away of God, though but even in thought, dissolves all.”

Before proceeding any further, however, it is important to note some of the intrinsic complexities and problems with studying religious prosociality. The very terms “religion” and “religiosity” are notoriously hard to define, let alone quantify with any precision. It is often said that there are as many definitions of religion as there are scholars of religion. Indeed, religion lacks a universally accepted definition, and different researchers use the term in different ways. Moreover, since most researchers come from Western countries, their understanding of religion is often tied to Judeo-Christian ideas that might not be applicable to other religions (Henrich, Heine, & Norenzayan, 2010). The way a religiosity questionnaire is constructed or presented might greatly influence participants’ answers and, thus, limit the

- 23 -

potential generalizability of the results. For example, a study conducted in Egypt found that when interviewed about their religiosity by a woman wearing a headscarf, Muslims over- reported their religiosity, while Christians under-reported it (Blaydes & Gillum, 2013). Using cross-cultural participant samples and questionnaires adapted to specific populations might help address this problem, although it does not necessarily solve it entirely.

To make matters worse, the term morality is equally problematic. Although people generally have no problem understanding what the concept of morality is, the content of the term—what is and what is not moral—is highly variable across cultures and individuals. For example, ritualized genital mutilation may be seen as an act of piety in some cultures and as an instance of child abuse in other societies. Drinking wine may seem as acceptable in everyday life and as a blessing at communion to a Christian, but as a grave sin under any circumstances to a Muslim. Furthermore, morality is an umbrella term that encompasses a number of different aspects of life (e.g. social relationships, cooperative exchanges, family life, etc.), and each of those aspects may have different associations with religiosity. Trying to fit all those behaviors under a single term generates vagueness and confusion, which negatively impacts the potential of systematically studying the topic.

For example, when we want to examine whether someone is a moral person, do we look at whether they cooperate more, they are more altruistic, or they cheat less? Is stealing the opposite of helping, or are these two very different behaviors? Does motivation matter, or is morality solely based on the outcome of a behavior? For instance, is donating money to charity to get a tax write-off moral, selfish, or both? Likewise, is mutualistic behavior that is beneficial for both the actor and the recipient morally virtuous? And what about behaviors that religions often consider as highly immoral, although they involve no harm to anyone (e.g. taboos associated with sex and food)? Such important nuances suggest that the concepts of religion and morality are too broad to be used as monolithic variables that can be measured with any degree of precision. A more fruitful methodology involves breaking these concepts down into more concrete aspects that can be operationalized for the purpose of scientific research. Although the terms religion and prosociality will be used throughout this chapter, they will always refer to some more specific operationalization (e.g. belief in god, ritual participation, helping behavior, cooperation, etc.) according to each case study.

- 24 -

The Religious Congruence Fallacy To this day, most people around the world posit a causal link between religious belief and morality. This attitude has been demonstrated cross-culturally by an experimental study of intuitive attitudes towards nonbelievers across 13 societies on all continents (Gervais, Xygalatas, & McKay, under review). To target subjects’ visceral attitudes, the experimenters focused on the representativeness heuristic, a cognitive bias that can lead to stereotypical judgments of others based on superficial or irrelevant characteristics. For example, when asked whether a woman with a humanities degree and a track-record of human rights activism is more likely to be (A) a bank teller, or (B) a bank teller and a feminist, most people will choose option (B) as more probable. However, that option is by definition the less likely one, as there are more bank tellers overall than there are feminist bank tellers. This logical error is called the conjunction fallacy (Tversky & Kahneman, 1982).

For their study, Gervais and his colleagues gave subjects the description of a person who committed multiple murders and asked whether that it was more probable that this person was (A) a teacher, or (B) a teacher who either was a religious believer or (for another group of subjects) did not believe in God. Subjects in the atheistic condition produced much higher conjunction fallacy rates (picking option B), which suggests that people consider criminal behavior as more representative of atheists than of religious individuals. These results were stronger in more religious societies and among more religious individuals, however, even non- religious participants in the study exhibited moral distrust of atheists. Other studies have found that people consider immoral behavior as more characteristic of atheists than of any other group. There thus seems to be a widespread view that religion is a necessary precondition for moral behavior, that is, that because religious ideologies are commonly concerned with regulating moral behavior, religious people must be more moral and, inversely, atheists should be less moral. Mark Chaves (2010) has called this assumption the religious congruence fallacy.

In reality, however, there is little evidence to support the view that religious people are more prosocial. For example, when we compare moral attitudes between religious and non- religious individuals, no consistent pattern emerges as a whole. Religious people are often more likely to advocate compassion and forgiveness, but at the same time be less tolerant of other groups (Stokes & Regnerus, 2009) and less likely to support welfare for the poor (Stegmueller, Scheepers, Roßteutscher, & De Jong, 2012). In addition, whether religious or not, people do not always practice what they preach. Catholicism staunchly opposes abortion, but Catholic

- 25 -

women are no less likely to abort (Adamczyk, 2009). And communism is an egalitarian ideology that advocates the subjugation of the self to the common good, but people in communist societies are no less likely to be selfish. Furthermore, although no causality can be established, more secular countries generally have lower crime rates (Zuckerman, 2007), and atheists are far less likely to end up in prison (US Federal Bureau of Prisons, 2015).

But what is the empirical evidence for the relationship between religiosity and moral behavior? In this chapter, we will review various lines of research that have attempted to address this question, comparing self-reported with actual behavior; we will discuss some of the conceptual and methodological problems involved in providing a comprehensive answer; and we will suggest avenues for further research.

What Religious People Say They Do Although the close link between religion and morality has been postulated for centuries, systematic research on the topic only began in recent decades. The earliest evidence was provided by sociological studies that used surveys and questionnaires as their main tool for studying the subject. Such surveys typically seek to assess relationships between various aspects of religiosity on the one hand (belief in God, ritual attendance, religious upbringing, etc.), and various parameters of personality and social conduct on the other (charity, compassion, cooperativeness, etc).

For example, a survey of 180 Belgian high school students (Saroglou, Pichon, Trompette, Verschueren, & Dernelle, 2005) asked respondents to answer a series of questions about religiousness, empathy (perspective taking, concern about others, and personal distress from another’s suffering), honesty (fairness, sincerity, and humility), and altruism (for example, “How often do you help a handicapped person cross the street?”). Each respondent was also asked to give the survey to one sibling and one friend, who provided their own independent assessments of that respondent’s behavior. The results showed that altruism was positively correlated with religiousness (the higher one’s religiousness, the higher their altruism), and the same correlation held for all three types of respondents (oneself, friend, sibling). Furthermore, there was a positive relationship between spirituality and perspective- taking, although no other aspects of empathy and honesty were significantly correlated with religiosity. Together, these results suggest that religious people report being more altruistic and empathetic and that they are seen as such by their peers.

- 26 -

Similarly, a number of other studies show that in comparison with non-religious participants, religious interviewees report having higher willingness to help and cooperate, being friendlier, more empathic and forgiving, and having higher moral standards (Furrow, King, & White, 2004; Mccullough & Worthington, Everett L., 1999; Putnam, Campbell, & Garrett, 2010). And not only do religious people see themselves as having these qualities, but they are also seen as such by their peers (Saroglou et al., 2005). However, these studies only reported hypothetical behaviors and did not examine actual moral conduct. But does this self- professed prosociality manifest in real life?

The Problem With Self-reports

Survey data are helpful in uncovering participants’ views and opinions, and they provide valuable insights into the process of self-presentation. They provide a cheap and fast way to obtain data, even from large samples, and can often be used to measure constructs that would be hard to study through observation. However, they also have serious constraints and limitations, and do not always measure what they are supposed to measure.

Self-presentation is subject to various conscious and unconscious biases, such as impression management, people’s conscious or unconscious desires and attempts to influence the way others see them (Goffman, 1956; Schlenker & Pontari, 2000). As a result, our self- descriptions do not always accurately represent our actual behavior. In other words, participants can present themselves as altruistic and cooperative but in reality their behavior might be very different. This problem is particularly pronounced when answering questions about traits that are considered positive or desirable in the context of a particular culture. The tendency to over- report those traits is known as social desirability bias (Fisher, 1993). And since both religiosity and prosociality are socially desirable traits, they are both known to be consistently over- reported (Brenner, 2011).

In fact, there is evidence that religious people are particularly prone to social desirability bias. Will Gervais and Ara Norenzayan (2012) conducted a series of studies that focused on self-awareness among religious participants. When the researchers provided participants with reminders of religion (e.g. asking them to pick words to describe God), highly religious individuals were significantly more conscious of their external image and self-presentation, and more concerned about what others thought of them. The experimenters also gave participants an 11-item questionnaire that contained realistic questions on undesirable traits

- 27 -

(e.g. “I am sometimes irritated by people who ask favors of me”) and unrealistic socially desirable statements (“No matter who I am talking to, I am always a good listener”). The results showed that compared to participants scoring low on self-reported religiosity, highly religious participants chose significantly more socially desirable answers. These findings suggest that reminding religious people of God increases their motivation to care about their self- presentation and social desirability. This makes self-reported correlations between religiosity and prosociality even more problematic, as the more religious respondents are, the less reliable this relationship is.

Even in the absence of bias, our introspective abilities are limited, which means that we are not always fully aware of our own internal states or our external behavior. This was illustrated by a study of a Spanish fire-walking ritual (Xygalatas, Schjødt, et al., 2013). Those who took part in the ritual provided subjective evaluations of their stress levels, but the researchers also used heart rate monitors to measure their arousal. When the self-reports were compared to the physiological measurement, the results revealed a sharp contradiction between the two: people reported feeling entirely calm when their physiological arousal reached extreme levels, often approaching 200 beats per minute. And in fact, when these results were shown to the participants, they were shocked at how inaccurately they had perceived their own physiology. A subsequent study of facial expressions during the same ritual (Bulbulia et al., 2013) confirmed that the high levels of stress were also perceptible to observers, which suggests that in this case self-reports were the least reliable means of assessing arousal.

In sum, self-reports are plagued by a number of problems related to the way people perceive, understand, and portray themselves. Thus, although surveys can be useful for revealing what religious people claim or think about their moral behavior, they cannot provide any conclusive clues with respect to their actual behavior.

What Religious People Actually Do In order to examine whether religious people actually behave more prosocially, social scientists use behavioral experiments that look at how participants respond to particular situations. Those experiments show that there is a striking asymmetry between the findings of surveys and those of behavioral studies, suggesting that although religious people portray or think of themselves as more prosocial, this moral high ground does not manifest in their actual behavior.

- 28 -

For example, in a study conducted at Princeton University, John Darley and Daniel Bateson (1973) simulated the Biblical parable of the Good Samaritan (Luke 10: 29-37, New International Version) to see what factors would influence helping behavior. The researchers recruited 47 theology students and randomly assigned them into two groups. The first group read a text on the parable of the Good Samaritan, while the second group read a text about job opportunities. After reading either of these narratives, participants were instructed to go to another building where they were going to give a talk on their reading. These instructions differed in how much participants had to hurry: some were told that they are late and should hurry, while some were told to take their time because the other researcher was not yet ready. In an alley on the way to that building, participants encountered a “victim” (who was in fact an actor) sitting on the pavement and coughing, looking very sick. The researchers recorded which participants stopped and offered helped to the victim, and then looked at which situational or personality factors influenced their decisions.

The results of this experiment revealed that the only significant variable influencing participants’ behavior was how much they were in a hurry: when the experimenters told participants that they were running late, they were much less likely to offer help to the victim. The fact that some participants were reminded of the Good Samaritan parable did not play a significant role, and neither did their degree of religiosity. That is, even in the low-hurry condition, the more religious people were no more helpful than average. In fact, as the authors note, “on several occasions, a seminary student going to give his talk on the parable of the Good Samaritan literally stepped over the victim as he hurried on his way!” (Darley & Batson, 1973, p. 107).

Since then, numerous studies have disputed the claim that religious individuals behave in more prosocial ways than non-religious ones (Galen, 2012). But that does not mean that religion has no effect on moral behavior. A cross-cultural study that examined economic behavior across eight societies (Purzycki et al., 2016) found that people’s particular views about the personality of their gods was a significant predictor of their behavior towards distant co-religionists. Specifically, individuals who saw their gods as moralizing and punitive were more willing to favor those distant members of their group over themselves or local co- religionists.

This brings up an important observation: cases where religiosity does influence moral behavior are usually constrained to the religious ingroup (other members of the same religion),

- 29 -

and do not necessarily extend to outgroups (members of other religions). Thus, when considering the effects of religion on prosociality, it is important to distinguish between the possible recipients of this behavior (e.g. ingroup vs. outgroup). Religion might stabilize mutualistic exchange by increasing trust among its members while at the same time inducing hostility towards other groups. For example, Jordan LaBouff and his colleagues (2012) asked participants about their attitudes towards different religions either in front of a cathedral, or in front of a town hall. Those who answered in front of the cathedral reported higher religiosity, but also higher political conservatism and more negative attitudes towards other religious groups. Religion can bind people together, but can also divide them. A failure to distinguish between the identities of the recipients of those actions and attitudes could be a significant source of confusion.

Religious Disposition and Religious Situation

LaBouff’s study highlights another important aspect which has often been neglected in both surveys and experimental studies, namely the role of situational factors. Surveys most commonly only target dispositional religiosity (having to do with personality), looking for the effects of conscious religious beliefs on hypothetical moral conduct. A recent body of research, however, has revealed a significant effect of religious situation (contextual factors) on moral behavior. For instance, Deepak Malhotra (2008) measured people’s responses to an online appeal for charitable donations and found that religious people were significantly more charitable only on Sundays, while throughout the week religiosity made no difference in levels of generosity. Malhotra coined the term Sunday effect to describe this phenomenon, that is, that religious participants are prone to behave more prosocially only in religious contexts, such as upon returning from Sunday Mass. This paints a more interesting and complex picture of religious morality. As Malhotra acknowledges, the salient question is not “are religious people more moral?”, but rather “under which conditions do they behave more morally?”

To study those conditions, researchers have often employed priming methods – using different stimuli to elicit certain moods, attitudes, and behaviors in participants. In a priming study, participants are typically exposed to a stimulus without full conscious awareness of its presence or its role, and researchers observe the effects that this stimulus has on participants’ behavior and decision-making. This paradigm is often used in order to examine the

- 30 -

automaticity of behavior, that is, how we sometimes make decisions without being conscious of all the subtle factors that influence those decisions.

For example, in a study conducted by John Bargh and his colleagues (1996), participants were presented with different versions of a scrambled-sentence task, where they had to choose four out of five randomly ordered words in order to form a meaningful sentence. The experiment had three different conditions: polite, neutral, and rude. In the polite condition, participants had to order sentences that contained words like “honor”, “respect”, or “patiently”; in the neutral condition, the sentences contained words like “normally”, “send”, or “watches”; and in the rude condition, participants had to use words like “bold”, “bother”, or “disturb.” While participants were solving the scrambled-sentence task, the experimenter engaged in conversation with a confederate (an actor who is part of the experimenter’s team unbeknown to participants). Upon finishing the task, participants were instructed to bring the finished sentences to the experimenter; however, since the experimenter was engaged in conversation, they had to decide whether to interrupt him or not. The main measurement in this study was whether participants interrupted the experimenter’s conversation within 10 minutes. The results showed that the percentage of those who interrupted the conversation significantly differed between conditions: 18% of participants interrupted in the “polite” condition; 38% in the neutral condition; and 64% in the “rude” condition.

Thus, although participants did not make a conscious link between the scrambled sentences and their decision to interrupt, priming with emotionally charged words increased their tendency to behave in a specific way. The priming paradigm is an important tool that enables researchers to discover situational influences on decision making. In the context of the study of religion and morality, it opened new avenues for exploring how various religious concepts and contexts might affect people’s prosocial behavior.

Religion as Prime Religious Concepts

In a controlled experiment that used the priming paradigm, Azim Shariff and Ara Norenzayan (2007) looked at the effects of implicit religious primes on prosocial behavior using a scrambled-sentence task. Half of the participants were given words that contained religious concepts like “spirit” or “divine”, while the other half were only given neutral words. In other

- 31 -

words, one group was primed with religious concepts, while the other group received no prime. Following this task, participants played an anonymous dictator game, where they had the power to decide whether they would share any of their earnings with another player. The results showed that people who were primed with religious concepts acted more prosocially, that is, shared more of their earnings with other players. But while the religious primes had an impact on their behavior, the players’ religiosity had no effect on their decisions.

Shariff and Norenzayan proposed two possible mechanisms that might be responsible for this effect. First, religious primes might trigger conceptual associations with norms and behaviors which are then unconsciously and automatically enacted: God is semantically related with morality, so reminding people of God makes them behave more morally. Second, religious primes might trigger a sense of being monitored by a supernatural watcher. It is well-known that people change their behavior in the presence of cues of being watched, even if it is simply a pair of stylized eyes on the wall. This sense of being monitored may activate reputational concerns and thus lead people to behave in more moral ways. Notably, these two mechanisms are not mutually exclusive, and indeed have been independently shown to operate in the presence of religious cues.

Religious Contexts

More recently, a number of studies have applied the priming paradigm in real-life settings. Although those studies often do not have the same level of control as laboratory experiments, they use more culturally salient, naturalistic stimuli and thus have greater ecological validity, that is, they better approximate the conditions under which these phenomena would occur in real life. In addition, since these studies are conducted in more natural settings, they provide an opportunity for cross-cultural comparisons beyond the typical samples of Western undergraduate students. This approach is particularly valuable when it comes to studying religion, which is heavily laden with culturally-specific meanings and values that cannot easily be replicated in sterilized laboratory environments.

In a field experiment conducted in Mauritius (Xygalatas, 2013), local Hindu participants were randomly assigned into on two groups, who played a common pool resource game. Those in the first group played the game in a religious temple, and those in the second group played in a restaurant. Each player had to make a financial decision independently and anonymously. The game consisted in deciding how much to withdraw from a “common pool”

- 32 -

of 500 Mauritian rupees (a substantial amount of money equivalent to 2-3 days’ salary for an unskilled worker). If the cumulative total withdrawn by both players was lower than 500 rupees, the remaining amount would be increased by 50% and shared among the two players, in addition to their withdrawal. But if the cumulative amount exceeded 500 rupees, both players would lose and receive nothing. This game provides a measure of cooperation, as players have to balance between their self-interest to withdraw as much as possible and the collective benefit that depends on sharing as much as possible. The results showed that those who played in the temple were more cooperative (withdrew less money) compared to those in the restaurant, although the players’ religiosity had no effect on their decisions. In a similar study conducted in Chile, Ali Ahmed and Osvaldo Salas (2013) found that people who played an economic game in a chapel were significantly more cooperative than those who played in a lecture hall.

Religious Practices

Another line of research has examined the impact of ritual participation on prosocial behavior. For example, a study conducted in Israel (Sosis & Ruffle, 2003) found that among the members of a Kibbutz (a collectivist community), those who participated more frequently in communal prayers were more cooperative in an economic game.

Indeed, rituals involve a variety of elements that may promote prosociality. Such elements include music, synchrony, arousal, and suffering. For example, studies have shown that when a group of people move in synchrony (e.g. marching, dancing), this can increase inter-personal rapport. In a field study conducted in a Spanish town (Konvalinka et al., 2011), heart-rate measurements were obtained during the performance of a fire-walking ritual. The researchers found that even in the absence of any motor synchrony, participation in this ritual led to the alignment of people’s physiological states, and that these effects extended to the entire community –not merely to fire-walkers themselves. Such emotional synchrony can strengthen bonds within a community and foster group cohesion.

Similar effects have been demonstrated at the behavioral level. In a field experiment conducted in Mauritius, Xygalatas and his colleagues (2013) studied one of the most widely performed high-intensity rituals in the world, the Tamil practice of Kavadi. This ritual involves piercing of the body with needles and skewers, carrying heavy bamboo objects (kavadi) in a long procession, and dragging chariots the size of cars with chains attached to the skin through hooks. The researchers compared donations to a charity among those who had taken part in this

- 33 -

ritual to those who had participated in a collective prayer. While participation in both rituals led to higher donations than a control group, the painful ritual produced much higher donations than the low-arousal one. Notably, the level of pain that participants experienced predicted the level of donations. Finally, just like we saw in the Spanish study, these effects also extended to observers. Similar effects of pain and suffering have been well-documented in controlled studies, which show that painful experiences elicit prosocial responses (Bastian, Jetten, & Ferris, 2014).

Is Belief Necessary?

Given that religious priming often works at a subconscious level, can we assume that even non- religious people will be primed with religion, o are the effects of religious primes constrained only to believers? Answering this question requires a better understanding of the cognitive mechanisms that facilitate the effects of religious priming. We need to understand how the associations between religious words and symbols and moral behavior originate, as well as the role of socialization in the formation of these associations. For example, some religious words such as God might be universally recognized as religious among English speakers and thus activate an association with religion, while some other symbols such as religious music might be known only to the practitioners who have been exposed to them during religious services. This was documented in a study conducted in three countries (the Czech Republic, Mauritius, and the USA), in which Martin Lang and his colleagues (2016) examined the influence of religious music on participants’ moral behavior. After being exposed to either a religious or a secular piece of music, participants had to solve a series of mathematical problems, and they received a monetary reward for each successfully solved problem. However, the research design intentionally provided the opportunity to cheat in order to increase the payoff. The results showed that only religious participants behaved more morally after hearing religious music, but the music did not have any significant effects on non-believers.

Another study conducted among Mauritian Christians used a within-subject design (examined how the same people behave in different situations) to study the effects of religious context (Xygalatas et al., 2015). Participants in that study were asked to solve a series of puzzles, each in a different location: a Christian Church, a Hindu temple, and a restaurant. Each time they solved a puzzle, they were rewarded with 100 rupees, but then they also had the opportunity to contribute part of their earnings to a charity organized for those who could not

- 34 -

successfully solve the puzzle. The identities of the recipients were unknown to the givers. The results showed that donations were significantly higher in both the Hindu temple and the Christian church compared to the restaurant, suggesting that the effects of religious priming are not restricted to primes pertaining to the group but extend to reminders of religion in general. But the results also revealed an interaction between religiosity and location, indicating that belief played an important role: religious people donated more when they were located in a religious setting. As we saw in the music study, belief and context interact in meaningful ways to produce results stronger than either of those factors alone can produce.

But does this mean that people need to be religious in order to be affected by religious stimuli? So far, results coming from different studies have been mixed. Some primes affected both religious and non-religious participants, while some only affected believers, and in a recent meta-analysis (a statistical procedure calculating the cumulative results of multiple studies), Shariff and his colleagues (2016) found no overall effects of religious primes on non- believers. However, the evidence is far from conclusive, as we lack specific data from non- religious samples. Most existing studies simply examined religiosity on a spectrum and compared more with less religious participants, but low religiosity is not the same as no religiosity. This lack of evidence suggests that we need to be cautious when it comes to interpreting the existing data, and that more research is needed to resolve the issue.

Belief and Practice Both belief and practice are important parts of religion, but we need to better understand how these aspects work together. There are deeply religious individuals who do not participate in organized ritual activities. On the other hand, many people regularly attend religious rituals without having strong religious convictions. Does religion impact those two groups in the same way? Can we predict, for example, that they will behave similarly in the presence of religious symbols or after the performance of a collective ritual? Is it the social aspect of collective rituals that motivates people to behave morally, as Comte and Durkheim would have it, or is it their belief systems, as Locke argued? Or is it both?

In order to answer this question, we need to gain a better understanding of how belief and ritual practice might independently affect moral behavior. For example, we can look at secular rituals devoid of any belief in the supernatural, as well as religious beliefs without ritual participation. For instance, high-intensity rituals are frequently used by secular groups like

- 35 -

armies, fraternities, and corporations to boost cohesion and cooperative behavior among their members. In the USA, fire-walking rituals can be purchased as team-building activities, while in New Zealand the ritual dance of haka is performed by sports teams before each match to increase bonding between team members. On the other hand, some people might believe that they are one with the universe and thus have the moral imperative to behave altruistically towards others without engaging in any ritual practices. A deeper understanding of how religious beliefs and practices work independently might help us identify some of the underlying mechanisms affecting morality.

For instance, while belief in ancestral spirits or some impersonal spiritual power might not pose any demands on one’s behavior, belief in an omnipotent god can motivate people to behave morally because of this deity’s ability to punish transgressors. Likewise, participation in low-arousal communal prayers might affect people’s motivations in different ways than participation in high-intensity rituals that involve pain and suffering.

In a study conducted in Mauritius, Xygalatas and his colleagues (in press) gave participants the chance to cheat in order to maximize their profit from a monetary task, and examined some of the factors affecting the levels of cheating. The two most important factors, which were negatively associated with cheating, were belief in an omnipotent punishing version of god and regular participation in the Kavadi ritual that involves prolonged suffering. Both factors independently predicted lower cheating rates, but no such effects were observed for belief in other kinds of deities or participation in low arousal rituals. These results stress the importance of a more detailed understanding of participants’ religious beliefs and practices, but also demonstrate that specific beliefs and practices can be effective irrespective of each other. But if those factors can independently affect moral behavior, what happens when they are combined?

The combined effects of religious belief and practice have been so far only hypothesized. For example, one of the big theoreticians of ritual behavior, Roy Rappaport (1999), wrote that by participating in rituals, individuals signal their commitment to the rest of the community. However, their actual intentions to behave morally or to cheat on other community members are concealed from others. While their behavior might be moral in public, they might still be inclined towards immoral conduct in private. In contrast, religious belief can provide a strong motivation for moral behavior even in private. But such belief might quickly dry up without a supporting community and a frequent re-establishment of moral order through

- 36 -

the performance of collective rituals. As shown above, religious people do not necessarily behave more morally. The repetition of moral axioms and a collective renewal of group identity might be important instruments that reinforce individual belief. In this respect, religious belief and practice might interact and reinforce each other to promote a particularly effective complex that influences moral behavior. In order to test this theory, however, we need to look at the combined effects of specific beliefs and practices in experimental and field studies.

Future Directions This overview of the state of the art in the research on religion and morality reveals a number of important questions that still remain unanswered and will need to be addressed by future studies. Some of the main challenges are related to the lack of precise measures of religious belief; the need for more contextualized studies of moral behavior; and the need to obtain a better understanding of how different religious beliefs and practices and their interaction affect moral behavior.

To address the first issue, i.e. developing better measures of religiosity, we need to develop measures that will combine emic (insider) and etic (outsider) perspectives to provide more comprehensive and contextually relevant research instruments. For example, in polytheistic religions, adherents’ preferences, commitments and affection may vary significantly between different deities. In those contexts, asking questions abstractly about “god” or generally about “gods” may not adequately capture people’s understanding of the divine.

In addition, it is important to distinguish between people who are non-religious and those that score low on religiosity. If we can put religiosity on a scale from zero to ten, the difference between zero and one is probably not equidistant to that between any other two consecutive points on that scale. In other words, a difference in degree of religiosity is qualitatively not the same as the difference between being a believer and being an atheist. Clarifying this conceptual problem might lead to a better methodological treatment of the issue of measurement.

The second issue, i.e. the need for more contextualization, has recently received attention as various scholars have raised criticism of the way psychological research is conducted. The vast majority of psychological experiments is conducted with samples

- 37 -

consisting of Western undergraduate students. Joseph Henrich, Steven Heine, and Ara Norenzayan (2010) showed that while researchers typically take those samples to be representative of the global population as a whole, in fact they are often outliers in many domains of cognition and behavior. Clearly, this has severe implications on the potential replicability and generalizability of psychological research.

To address this problem, we need cross-cultural, interdisciplinary research, which will move beyond sterilized laboratory settings, into the real-world contexts where religion actually takes place to increase the ecological validity of the findings. This requires a synergy between anthropology, psychology, and other related disciplines, who have traditionally studied religion in isolation and without interaction with one another.

The third issue, i.e. the interaction between religious belief and practice, is particularly complex and requires systematic research. Although the connection between doctrine and ritual is observed worldwide, there is no particular theoretical reason why the two should always co- occur. After all, in the secular domain there are plenty of ideologies without rituals, and rituals without belief systems. However, in the religious domain, the two are intricately related and interacting in complicated ways.

To get a better understanding of religion’s link to morality, we need to be able to account for the effects of beliefs and practices both independently and cumulatively. This requires implementing a systematic division of labor while at the same time maintaining a cohesive bird-eye’s view. Although this is certainly easier said than done, being able to isolate these factors will give us a better understanding of what makes religion so successful.

Summary As we have seen, the relationship between religion and morality is far more complicated than one might expect. The challenges of defining, operationalizing, and measuring both religion and morality require a fractionating approach. This approach involves examining various aspects of the problem separately and then trying to put the pieces of the puzzle together to look at the broader picture rather than relying on isolated studies. Furthermore, the observed discrepancy between self-reported and actual behavior demonstrates additional problems with measuring socially desirable traits like religion and morality.

- 38 -

A careful look at the available evidence shows that religious people are no more or less moral than non-believers, despite what they often report, and in the face of widespread popular assumptions and stereotypes. But although religious disposition plays little role in moral behavior, religious situation can exert significant influence on it. Religious concepts, contexts and practices can independently influence social conduct, and their interaction can make religion a powerful social force. This force can be used for better or for worse, either directed towards building cohesive communities and increasing in-group cooperation or producing hostility and suspicion towards outgroups.

- 39 -

Music as a Sacred Cue? Effects of Religious Music on Moral Behavior Lang, M., Mitkidis, P., Kundt, R., Nichols, A., Krajčíková, L., & Xygalatas, D.

Abstract Religion can have an important influence in moral decision-making, and religious reminders may deter people from unethical behavior. Previous research indicated that religious contexts may increase prosocial behavior and reduce cheating. However, the perceptual-behavioral link between religious contexts and decision-making lacks thorough scientific understanding. This study adds to the current literature by testing the effects of purely audial religious symbols (instrumental music) on moral behavior across three different sites: Mauritius, the Czech Republic, and the USA. Participants were exposed to one of three kinds of auditory stimuli (religious, secular, or white noise), and subsequently were given a chance to dishonestly report on solved mathematical equations in order to increase their monetary reward. The results showed cross-cultural differences in the effects of religious music on moral behavior, as well as a significant interaction between condition and religiosity across all sites, suggesting that religious participants were more influenced by the auditory religious stimuli than non-religious participants. We propose that religious music can function as a subtle cue associated with moral standards via cultural socialization and ritual participation. Such associative learning can charge music with specific meanings and create sacred cues that influence normative behavior. Our findings provide preliminary support for this view, which we hope further research will investigate more closely.

Keywords: religion, music, associative learning, morality, priming

Introduction Much psychological research conducted over the past decade has attempted to further scientific understanding of morality and ethical behavior by observing how environmental cues can enhance or degrade ethical behavior (L. K. John, Loewenstein, & Rick, 2014; Mazar & Zhong, 2010; Mead, Baumeister, Gino, Schweitzer, & Ariely, 2009; Shariff & Norenzayan, 2007). Inferred social norms (Gino, Ayal, & Ariely, 2009), ethical reminders (Mazar, Amir, & Ariely, 2008), and even decorative objects in a room (Krátký, McGraw, Xygalatas, Mitkidis, & Reddish, 2016), have all been observed to affect dishonest behavior. This evidence suggests

- 40 -

that automaticity plays an important role in moral decision-making based on perceptual cues (Bargh, Schwader, Hailey, Dyer, & Boothby, 2012; Newell & Shanks, 2014). Making internalized norms salient via contextual cues can push people toward normative behavioral strategies (Cialdini, Reno, & Kallgren, 1990; Hirsh, Galinsky, & Zhong, 2011), often without a conscious link between the two (Bargh & Morsella, 2008). As such, behavioral responses to moral dilemmas might result from the interplay between individual norms and contextual percepts, especially in a structured environment that is regulated by normative institutions (Graham, Meindl, & Beall, 2012).

A prime example of such a normative institution is religion. Religions often strongly impact the individual’s socialization process, and through the use of reminders such as symbols and repeated rituals make group-specific norms salient (Durkheim, 1912/1964; Norenzayan & Shariff, 2008; Xygalatas, 2013). Research in recent years has shown that religious situational factors enhance the saliency of norms and play a significant role in moral decision-making (for an extensive meta-analysis see Shariff et al., 2016). However, despite an ample body of research on religious prosociality, the effects of religious contextual cues on unethical behavior are less well-documented. Only a handful of studies have looked at the effects of religious cues on deterring cheating (Bering, McLeod, & Shackelford, 2005; Mazar et al., 2008; Piazza, Bering, & Ingram, 2011; Randolph-Seng & Nielsen, 2007). For example, Mazar et al. (2008) found lower cheating rates amongst participants who were asked to recall the 10 Commandments compared to those who had to recall 10 book titles. Similar results were observed when using other environmental cues, such as the Islamic call to prayer (Aveyard, 2014).

These studies suggest that people modify their decisions in response to sacred cues, similarly to the way they respond to other environmental cues (for instance light in the room – Mazar & Zhong, 2010), and that religious environments might have complex effects on people’s social behavior. However, the exact mechanisms underlying the perceptual-behavioral links that affect decision-making under the influence of sacred cues are still not fully understood. Researchers have traditionally primed concepts of spirituality implicitly through the use of religiously infused anagrams (Srull & Wyer, 1979). For example, “dessert divine was fork the” would be unscrambled by participants to “the dessert was divine” (Shariff and Norenzayan, 2007). Such priming can carry semantic associations with moral norms and might also invoke fear of supernatural punishment thereby inhibiting immoral behavior. Similarly, anthropomorphic depictions of eyes might evoke a feeling of being observed and trigger reputational concerns (M. Bateson, Nettle, & Roberts, 2006; Krátký, McGraw, et al., 2016).

- 41 -

But would the same effects on moral behavior hold for arbitrary stimuli associated with religion, for instance, specific objects, gestures, or music? While the meanings of words are formed during the process of early socialization, and associations with specific actions are reinforced by everyday use, religious symbols are often confined to specific domains of one’s life. Their tentative influence on moral decisions is moderated by associative learning, but it is not yet clear whether such influence would be strong enough to deter cheating. Could religious environments affect moral behavior through the accumulation of arbitrary, subtle sensory cues associated with morality?

To answer this question, we suggest a novel approach to religious priming. We selected a stimulus that does not bear any inherent meaning by itself: instrumental music. While religions employ multiple symbols that could have been chosen, music is a widespread feature of religious environments that can be translated between different cultures (as opposed to specific symbols like Shiva lingam, Christian crosses, etc.). Moreover, numerous researchers have noted that music can play a significant role in social cohesion and cooperative behavior (Dunbar, Kaskatis, MacDonald, & Barra, 2012; Kirschner & Tomasello, 2009; Lang, Shaw, et al., 2015; Pearce, Launay, & Dunbar, 2015). It has been suggested that music can function as a proto-symbolic system that encompasses the structure of rituals, and that religious environments might have complex effects on people’s social behavior (Alcorta & Sosis, 2005). Indeed, such a connection can be described as extra-musical meaning (Koelsch, 2011) or culturally enactive meaning (Ian Cross & Morley, 2008), referring to explicit and conventional associations of music with real-world situations (e.g., anthems making people aware of their identity; Brown, 2000). This association may work similarly to the association with linguistic concepts. In an EEG study by Koelsch et al. (2004), participants were primed with sentences or musical excerpts that were semantically either related or unrelated to a word that followed. The authors recorded an event-related brain potential that is sensitive to a semantic fit (N400) and found no difference between sentences and musical excerpts. That is, when musical excerpts were semantically unrelated to the words that followed, the same error occurred as in the case of sentences. This result suggests that music can convey linguistic concepts and prime the meaning of a word (Koelsch, 2010). Such primes have been used, for instance, in a study of purchasing behavior, showing that when music is associated with information congruent with an advertised product, participants are more likely to be persuaded by the advertisement (North, Mackenzie, Law, & Hargreaves, 2004).

- 42 -

Besides the extra-musical meaning, musical stimuli carry information and messages that can elicit specific emotional responses, which in turn affect mood (Thompson, Schellenberg, & Husain, 2001) and morality judgments (Seidel & Prinz, 2013). For example, musical stimuli with positive valence decrease concerns regarding immoral messages and increase compliance with a request to harm others (Ziv, 2015; Ziv, Hoftman, & Geyer, 2012). Negatively valenced music, on the other hand, can increase participants’ critical thinking (Sinclair, Lovsin, & Moore, 2007). Furthermore, it has been shown that the tempo of musical stimuli can influence emotional arousal (Webster & Weir, 2005) and cognitive performance (Schellenberg, 2005; Schellenberg, Nakata, Hunter, & Tamoto, 2007). However, we lack robust evidence showing that music influences participants’ actual moral behavior (Ziv et al., 2012). And if it does, does this happen via the induction of specific emotions, through an association with conceptual complexes, or both?

The current study explored whether priming participants with instrumental religious music would decrease the rate of dishonest behavior. To isolate the effects of religious music, we designed three conditions: religious, secular, and control. After exposure to one of the three stimuli, participants’ task was to solve a set of 20 matrices, and for each correctly solved matrix they received a monetary reward (Mazar et al., 2008). The number of correctly solved matrices was self-reported, thereby giving participants an opportunity to report dishonestly to increase their monetary reward and inflate their performance. We predicted that participants in the religious condition would behave less dishonestly than in the other two conditions. However, because instrumental religious music is not universally recognized as sacred (compared to religious concepts) and is thus less salient, we also expected that the effect of religious music would be moderated by participants’ religiosity (congruent with the extra-musical meaning). That is, only religious participants would respond to this environmental cue that should activate an internalized behavioral schema (honesty). An additional supplementary hypothesis assumed the moderating effects of ritual participation frequency. The emotional characteristics, tempo, and impact of the presented stimuli were also assessed in order to test the hypothesis that music can affect decision-making through its affective component.

Addressing current debates on the generalizability of psychological studies (Henrich et al., 2010) and criticisms of religious priming literature and related meta-analytical research (Gomes & McCullough, 2015; Shariff et al., 2016; van Elk et al., 2015), we collected data from three different samples: a general population sample in Mauritius, and student population samples in the Czech Republic and the USA. By diversifying our participant pool, our goal was

- 43 -

to control for possible culturally unique responses to religious primes. Despite demographic differences between these sites, we did not expect that priming with religious music should have different effects. We hypothesized that the learned link between religion and morality should work similarly in all sites. We were also interested to see whether general religiosity rates might play an important role in the effectiveness of religious primes, and we thus selected these three countries due to their different rates of general religiosity (Gervais et al., under review; Zuckerman, 2007).

Materials and Methods Participants

Data were collected from May 2014 to July 2015 in three sites: we recruited participants from the general Hindu population in Point aux Piments in Mauritius; a student population at Masaryk University in the Czech Republic; and a student population at Duke University, North Carolina, USA. Across the three sites, 254 participants were randomly assigned to one of three conditions: religious, secular, and control. Participants who previously took part in a similar experiment or showed suspicion about the experiment’s goals were excluded from the final analysis (5 in Mauritius, 4 in the Czech Republic, and 13 in the USA). Overall, we tested 73 participants in Mauritius (20 females; Mage= 30.29, SD = 12.95); 78 participants in the Czech

Republic (40 females; Mage = 24.05, SD = 3.69); and 81 in the USA (47 females; Mage = 22.74, SD = 3.77). Participants who did not fill out the parts of our questionnaire regarding musical stimuli (n = 12) were retained in the analysis of behavioral data, but were omitted from the analysis of musical stimuli. Participants were tested alone in rooms that contained only a chair, table, and computer. All materials, questionnaires, and consent forms were translated into the local languages (Mauritian Creole, Czech, and English). Informed consent was obtained from all participants. The study was approved by the Institutional Review Boards of Masaryk University, University of Connecticut, and Duke University.

Material

In a double-blind design, participants were randomly assigned to one of three conditions defined by the type of stimulus they were exposed to: religious, secular, or control. Because we were specifically interested in the effects of music, none of the used musical excerpts contained any lyrics. All stimuli were of identical duration (2 min) and were administered via

- 44 -

headphones in order to prevent interference from external noise. In the control condition, participants were exposed to white noise in order to control for possible effects of sound manipulation. While the control stimulus was the same across the three sites,22 music in the religious and secular conditions was site-specific.

In Mauritius, we selected the appropriate religious music after consulting local religious experts, and comparable secular music after discussing with local research assistants. For the religious condition, we chose music that is often played during collective rituals in the local temple, and in particular during the annual religious festival of Thaipusam Kavadi.23 This musical piece has dominant fast drums and a flute sound that is characteristic of the Kavadi ritual. For the secular condition we chose a popular Bollywood song (Mera Mahi Bada Sohna Hai—“Dhaai Akshar Prem Ke”24) that had similar sound and tempo to the music in the religious condition by sampling the first minute without any lyrics. This minute was looped in order to create a 2 min music sample.

In the Czech Republic and the USA, we pre-screened four Christian religious songs that are used during Catholic mass and four comparable secular songs. Participants from the Czech Republic and the USA rated them on 14 characteristics. These characteristics were combined into measures of stimuli’s positivity, negativity, holiness, tempo, and impact (see Supplementary Material 1.1,1.2; and Tables S1, S2). In order to select secular stimuli that would be comparable with religious stimuli, we compared the most holy stimulus with the four pre-selected secular stimuli on the ratings of positivity, negativity, tempo, and impact, and selected the least different secular stimulus.

In the Czech Republic, 40 students from Masaryk University rated the eight selected stimuli. Since all of the religious songs had similar ratings of holiness (ranging from 4.28 to 4.43 out of 6), we chose the one that had the least mean difference in all other ratings with a secular song. Using this procedure, we selected an organ version of J. S. Bach’s Ave Maria 25 interpreted by Charles Gounod as the religious song (Mholy = 4.33, SD = 1.54), and Tchaikovsky’s Romance for piano in F Minor, Op 526 as the secular song. Ave Maria was performed on organs and Tchaikovsky’s piece on piano, and both songs had similar tempos. The same procedure was used in the USA to select appropriate stimuli. We used Amazon

22 See Audio 7 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full# 23 See Audio 1 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full# 24 See Audio 2 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full# 25 See Audio 3 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full# 26 See Audio 4 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full#

- 45 -

Mechanical Turk to recruit 102 participants who rated the same songs as participants in the Czech Republic. For the Religious condition, we selected J.S. Bach’s BWV 147 Jesu joy of 27 man’s desiring, which was rated as the most sacred song (Mholy = 2.94, SD = 2.10). The most similar secular song was J.S. Bach’s BWV 140 Sleepers Wake. 28 Although both songs were from the same composer, the religious one was performed on organs, while the secular one on piano.

Procedure

Our experiment was conducted using Cogent 2000 developed by the Cogent 2000 team at the FIL and the ICN, and Cogent Graphics developed by John Romaya at the LON at the Wellcome Department of Imaging Neuroscience. Cogent 2000 was run as a Matlab Toolbox (MathWorks Inc., 2013). Participants were seated in individual rooms in front of a table with a computer, and a local research assistant explained that the purpose of the study was to investigate the effects of music on cognitive performance. Subsequently, the research assistant made sure that every participant understood the instructions (a practice item was presented) and instructed participants to keep their headphones on for the rest of the experiment. The research assistant then left the room, informing the participant that she or he would be working in the adjacent room and could be called when needed. The condition-specific musical stimulus played for 2 min, after which low-volume white noise was played for the rest of the experiment. This served to eliminate any possible disturbing noises.

Once the music ended, a series of mathematical tasks appeared on the screen. The participants’ task was to solve as many as they could out of a total of 20 given matrices (adapted from Mazar et al., 2008). Each matrix was presented on the screen in the form of a 3 × 3 table of numbers (see Figure 1).

27 See Audio 5 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full# 28 See Audio 6 at: http://journal.frontiersin.org/article/10.3389/fpsyg.2016.00814/full#

- 46 -

Figure 1. A comparison of easy (A) and difficult matrices (B) used in the experiment.

In each matrix, participants had to find two numbers that added up to 10 and remember their coordinates. There was always only one correct solution. Each matrix was presented for 15 s, after which participants had 6 s to think about the correct solution. Subsequently, the correct answer appeared on the screen for 3 s, and if it matched the solution that participants had in mind, they would make a mark on a prepared sheet. The prepared sheet contained one previously filled-out row, suggesting how many matrices the previous participant had successfully solved. Because almost no cheating was observed in a pilot that was run in the Czech Republic before the current study, we decided to encourage participants to cheat by suggesting that a previous participant cheated as well (Gino et al., 2009). Thus, the pre-filled row always contained eight marks. The matrix-solving task lasted 8 min in total.

After participants went through all 20 matrices they were instructed by the program to call the research assistant who then administered a post-study questionnaire and compensated participants based on their self-reported number of correctly solved matrices. The questionnaire assessed participants’ religiosity, familiarity with the musical piece, ratings of the stimuli’s positivity, negativity, holiness, tempo, and impact, and contained basic demographics (see

- 47 -

Supplementary Material 1.4). We used the same approach to the construction of the stimuli’s measures as during pre-screening the stimuli (see Supplementary Material 1.3). Debriefing was performed at the end of data collection.

For each correctly solved equation, participants were paid 5 MUR/10 CZK/0.5 USD. The maximum possible amount that participants could earn in each site was roughly equivalent to a budget restaurant meal. We did not control how many equations participants really solved correctly. However, in contrast to Mazar et al. (2008) who used the overall number of claimed matrices in their analyses, we approached the approximation of actual cheating in a more robust way. Using the raw untransformed data would introduce variability where, in theory, there should not be any. In other words, two participants might have solved three and five matrices respectively, seemingly showing variability in cheating while actually having chosen the same behavioral strategy (honesty). Whereas this problem could be addressed with a large sample, adding predictors at the level of an individual (e.g., religiosity) could bias a predictor’s explanatory power. Furthermore, using the raw data would inflate the cheating scale and any differences in cheating would appear smaller than they were in reality.

To approximate the actual levels of cheating, we designed the experiment in such a way that most participants would solve five matrices. The first two equations were easy enough that everyone who passed the comprehension test should solve (adding up two numbers from 1 to 9), while the third matrix included numbers with three decimals, making it possible to solve in 15 s. In the rest of the matrices, the numbers contained 4 or 5 decimals, making it very difficult to solve in 15 s. However, participants could also guess the correct answer with a chance of 1:36 in each of the 17 remaining equations. According to the Bernoulli probability distribution, there is a 99% probability that a participant will not guess more than two solutions correctly. We thus assumed that participants who reported five or fewer solved equations were honest (i.e., possibly solving three and guessing two matrices).

To test this assumption, we recruited 112 participants from a student population at Masaryk University in the Czech Republic who were presented with the matrix task during a lecture in a large classroom. The matrices were projected on a wall and participants were instructed to write down answers (coordinates of two numbers adding up to 10) on a piece of paper. Participants who did not answer correctly any of the first two matrices without decimals were removed from the subsequent analysis (n = 12; such participants would not pass a comprehension test in our experiment), where we computed the average number of correctly solved matrices. Although the correspondence of pretest results with our assumption would not

- 48 -

mean that we measured actual cheating, we believe that a low SD of pretest results together with a relatively large sample size provides sufficient precision for assessing the effects of our manipulation on participants’ behavior.

Data Analysis

All data were analyzed in R version 3.2.3 (R Core Team, 2014). Since our data were bounded on the possible amount of dishonesty, we considered four different models: normal, normal censored, beta, and zero-inflated beta. While the untransformed data on cheating looked almost normally distributed (albeit leptokurtic; see Figure S2A), the values between 2 to 5 solved matrices mask the actual censoring. Since we considered five or fewer reported matrices as ethical behavior (see section Procedure), these values were collapsed to zero unfairly reported matrices. Thus, when this boundary was taken into account, histogram data showed a significant positive skew (see Figure S2B). Although the zero-inflation could be modeled by a censored model with normal distribution, the rest of the distribution (from the value of one and higher) was not normal either. We therefore considered a beta regression that uses a logit link to model means and variation in order to account for heteroscedasticity and skewness often present in bounded data (Cribari-Neto & Zeileis, 2009; Stasinopoulos & Rigby, 2007). To test whether a model with beta distribution would better fit the data, we transformed the number of claimed matrices to a percentage, with 15 being 100% - maximal dishonest behavior. Because our data also contained extreme values of 0 and 1 that are unacceptable for a beta regression model, we transformed the dependent variable using the formula (y’=(y·(n − 1) + 0.5)/n), where y is the transformed variable and n is the sample size (Smithson & Verkuilen, 2006). For the beta zero-inflated model, we used percentage data without transforming 0 and 1. A difference in Akaike Information Criterion (AIC) was used to compare models with different distributions (modeling only the intercept). From the four considered models, the one with beta distribution had significantly lower AIC than the other models (AICbeta = −137.24, AICnormal = 37.05). Thus, we used beta regression on the transformed data to model our dependent variable.

We fitted a beta regression model (Eskelson & Madsen, 2011; Smithson & Verkuilen, 2006) using the function gamlss (gamlss package; Stasinopoulos and Rigby, 2007). We built four sets of models. In the first set, we kept site as an independent factor in all models, controlling for differences between our sites. First, we modeled the main condition effect across all sites; subsequently, we added a Condition∗Religiosity interaction to the model and

- 49 -

compared it with a model that included a Condition∗Ritual participation interaction; and lastly, we added possible covariates. In the first addition, age and sex were considered. The second addition comprised of the stimuli’s positivity, negativity, tempo, and impact. In the second set of models, we analyzed condition effects and a Condition∗Religiosity interaction at each site. In the third set, we considered covariates that could explain tentative differences between the sites. Namely, we looked at between-site differences in religiosity; ritual participation frequency; perceived holiness of the religious stimuli; perceived negative and positive emotional valence of the stimuli; and perceived tempo and impact of the stimuli. Finally, in the fourth set, we looked at the musical characteristics of the religious stimuli and their predictive power regarding unethical behavior in the religious condition. In all models with condition effects, we set the religious condition as a reference category for comparisons. That is, we were interested only in differences between the religious condition and the other two conditions. We assumed there should be no differences between the secular and control conditions. For the models of cheating that included site as a predictor, the USA was set as the reference category, but this choice was arbitrary. Specific between-site differences in overall cheating were not of interest in the current study—we used site only as a control for effects that were outside of our interest.

Results Pretest

Results from the pretest confirmed our assumption that people on average solve five matrices (n = 100, M = 4.53, SD = 1.57). The minimum number of solved equations was two, while the maximum was nine. Although this range seems high at first, the frequency of participants that solved more than five matrices is exponentially decreasing (see Figure S1). We decided to set the cut-off at five as suggested by the mean number of solved matrices and Bernoulli probability distribution (see Procedure). In other words, we treated all participants in our experiment as behaving ethically if they reported five or fewer solved matrices. Six or more reported matrices were regarded as a scale of cheating.

- 50 -

Manipulation Check

An analysis of the perceived holiness of the stimuli across the three sites revealed a significant difference between conditions [F (2, 217) = 20.63, p < 0.001]. Specifically, the religious condition had significantly higher ratings than the secular condition, and the control condition (ps < 0.001; see Table 1 for descriptive statistics). Looking at the emotional valence of the stimuli [F (2, 217) = 4.64, p = 0.010], we found that the religious condition was perceived as significantly less negative than the control condition (p = 0.047). We did not observe any significant differences between the religious and secular conditions (p = 0.347). These results were replicated also for the positivity of stimuli [F (2, 217) = 18.06, p < 0.001]: the religious stimuli were rated as significantly more positive compared to the control stimuli (p < 0.001), but not compared to the secular stimuli (p = 0.573). Similar results were obtained for our measures of tempo [F (2, 217) = 6.90, p = 0.001] and impact [F (2, 217) = 4.97, p = 0.008] of the stimuli. The religious stimuli were rated as significantly slower than the control stimuli (p = 0.001), but there was no difference between the religious and secular stimuli (p = 0.874). In terms of impact, the religious condition had significantly higher impact than the control condition (p = 0.002). The difference between the religious and secular condition was not significant (p = 0.219).

Dishonest Behavior

To assess the amount of dishonest behavior among participants, we measured the percentage of matrices that were claimed as correctly solved and used beta regressions to estimate differences between predictors. We did not observe a significant difference between the religious and the secular (p = 0.44) and control conditions (p = 0.14). The estimates with significance levels from a beta regression are displayed in Table 2, Model 1 and plotted in Figure 2A. Looking at differences between the sites, participants in Mauritius claimed significantly more solved matrices than participants in the USA (p = 0.007), while participants in the Czech Republic claimed significantly fewer (p = 0.004; Table 2, Model 1). We observed a significant Condition∗Religiosity interaction, with religious people cheating significantly less in the religious condition (p = 0.027). Compared to the religious condition, religiosity played a significantly smaller role in the secular (p = 0.026) and control conditions (p = 0.039; see Table 2, Model 2 and Figure 2B). That is, the more religious participants were, the less they cheated in the religious condition, while in the other two conditions religiosity did not significantly

- 51 -

affect cheating. The model comprising a Condition∗Ritual participation interaction suggested the same trend (for religious condition, p = 0.294), but the interaction was significant only for the secular condition (p = 0.011) and not for the control condition (p = 0.086; see Table 2, Model 3). From the considered covariates, only sex significantly improved the model fit. Aggregating across the three sites, on average males reported more matrices than females (p =

0.025; see Table 2, Model 4). There was no effect of perceived valence (pnegativity = 0.203; ppositivity = 0.335; ptempo = 0.382; pimpact = 0.286) of the stimuli or of age (p = 0.847) on participants’ behavior (Table 2, Model 5).

- 52 -

Table 1 Descriptive Statistics of Dishonest Behavior and Musical-stimuli Ratings

Religious (n = 74) Secular (n = 80) Control (n = 78)

Variable M SD CI d M SD CI d M SD CI d 31.5 % Claimed 30.27 27.35 24.04-36.05 - 24.41 26.37-36.63 0.05 34.96 27.71 28.81-41.12 0.17 0 Holiness 3.84 1.58 3.47-4.21 - 2.86 1.32 2.56-3.15 0.68 2.42 1.16 2.15-2.68 1.03

Negativity 2.28 0.92 2.10-2.49 - 2.13 0.80 1.95-2.31 0.17 2.59 1.09 2.34-2.84 0.31

Positivity 3.11 0.84 2.91-3.31 - 3.20 0.89 2.99-3.40 0.10 2.34 1.10 2.09-2.59 0.78

Tempo 2.73 0.96 2.50-2.95 - 2.76 0.83 2.57-2.94 0.03 3.23 0.96 3.01-3.45 0.52

Impact 3.26 1.13 3.01-3.53 - 3.01 1.28 2.73-3.30 0.21 2.63 1.63 2.34-2.91 0.53

Note. CI = 95% Confidence intervals. Cohen’s d is the effect size of comparisons between the religious condition and the other conditions.

- 53 -

Tab1e 2 Estimates with SE from Beta Regressions for the Percentage of Matrices Claimed as Correct. Predictor Model 1 Model 2 Model 3 Model 4 Model 5 29.84 (3.61) 30.32 (6.27) 30.27 (6.19) 29.86 (6.49) Intercept 31.18 (6.78) *** *** *** *** *** Mauritius 11.32 (4.18)** 11.10 (4.44)* 9.28 (4.21)* 7.93 (4.77) Ϯ 9.27 (5.19) Ϯ

Czech Republic -9.61 (3.34)** -9.63 (3.44)** -9.38 (3.39)* -10.50 (3.48)** -9.75 (4.20)*

Secular 3.041 (3.93) 2.45 (3.92) 3.48 (3.95) 2.74 (3.97) 3.03 (3.93)

Control 5.99 (4.07) 5.40 (4.05) 6.01 (4.06) 6.19 (4.16) 7.60 (4.38) Ϯ

Religiosity -5.31 (2.40)* -4.97 (2.48) * -4.97 (2.43)*

Secular*Religiosity 7.55 (3.37)* 7.54 (3.45) * 7.32 (3.37)*

Control*Religiosity 6.60 (3.18)* 6.49 (3.28) * 6.26 (3.18)*

Ritual -1.55 (1.47)

Secular*Ritual 5.40 (2.12)*

Control*Ritual 3.54 (2.06) Ϯ

Females vs. Males 7.90 (3.50)* 8.47 (3.48)*

Age 0.04 (0.20) 0.04 (0.20)

Positivity -2.16 (2.24)

Negativity -2.70 (2.12)

Tempo -1.61 (1.84)

Impact 1.99 (1.86)

Cox-Snell R2 .124 .147 .157 .166 .175

Note. In all models, we control for the effects of Site. Religious condition and USA site were set as reference categories (intercept). The first model contains only the effects of condition (compared to the religious condition) while controlling for the effects of site. The second model includes a Condition*Religiosity interaction, describing the effects of religiosity on cheating in the Religious condition. The two predictors specified as interactions (Secular*Religiosity and Noise*Religiosity) are comparisons with this effect. Again, we control for site. The third model has an identical design to the second, only with a Condition*Ritual participation interaction. Since the effects of ritual participation on morality were not as strong as those of religiosity, we retained the latter factor for subsequent models. The fourth model contains site and condition effects, significant interaction, and demographic covariates. The fifth model controls also for different characteristics of our musical stimuli. Ϯp < 0.1; *p < .05; **p < .01; ***p < .001

- 54 -

Figure 2. (A) The effects of different stimuli on the percent of matrices that were claimed as correctly solved above the expected levels with ±SEM. While controlling for the effects of site, there were no significant differences between conditions (see Table 2, Model 1). (B) Predicted values with 95% confidence intervals for the Condition*Religiosity interaction. The significantly different slopes suggest that religious participants cheated less upon being exposed to religious music (Table 2, Model 2).

- 55 -

Between-Sites Differences

Focusing on the differences between our three sites (Mauritius, the Czech Republic, and the USA), we built separate models for the condition effects (see Table 3 and Figure 3 for descriptive statistics and Table 4 for model estimates). First, there was a significant difference between the religious condition and the other two conditions in Mauritius. Specifically, participants in the religious condition claimed a lower percentage of solved matrices than participants in the secular condition (p = 0.043) and participants in the control condition (p = 0.044). We did not observe a significant main effect of condition in the Czech Republic (religious vs. secular: p = 0.581; religious vs. control: p = 0.891). Likewise, the condition effect was not significant in the USA (religious vs. secular: p = 0.718; religious vs. control: p = 0.695).

Looking at the Condition∗Religiosity interactions, we observed a marginally significant interaction in the USA sample (Religiosity∗Secular: p = 0.068; Religiosity∗Control: p = 0.052), but this interaction did not replicate in the other sites (ps > 0.3; Table 4, Models B).

In order to better understand why the results from Mauritius differed from the other two sites, we used site as an independent variable (with Mauritius as the reference category) in predicting religiosity and ritual participation; and holiness, tempo, impact, and valence of the religious stimuli (see Table 5 for descriptive statistics). Mauritian participants reported being significantly more religious [F (2, 229) = 13.31, p < 0.001] than those in the Czech Republic (p = 0.003) and the USA (p < 0.001). Similarly, participants in Mauritius reported significantly more frequent ritual participation [F (2, 229) = 14.41, p < 0.001] compared to participants in the Czech Republic (p < 0.001) and the USA (p = 0.010). Religiosity and ritual participation are plotted in Figure 4.

There were no significant differences [F (2, 67) = 1.03, p = 0.364] between Mauritius and the other sites in perceived holiness of the religious stimuli (Czech Rep.: p = 0.370; USA: p = 0.157). However, there were significant differences in perceived negativity of the religious stimuli [F (2, 67) = 20.55, p < 0.001], with the Mauritian stimulus rated as significantly more negative compared to the USA (p < 0.001), but not to the Czech Republic (p = 0.592). Conversely, this pattern of significance was reversed for the positivity of the religious stimuli [F (2, 67) = 8.83, p < 0.001], with the Mauritian stimulus being significantly less positive than the stimulus in the Czech Republic (p < 0.001) but not compared to the stimulus used in the USA (p = 0.093). Similar results were obtained for the tempo [F (2, 67) = 5.37, p = 0.007] and impact [F (2, 67) = 12.67, p < 0.001] of the religious stimuli. The Mauritian stimulus was rated

- 56 -

as significantly faster than the stimulus used in the Czech Republic (p = 0.003), but there was no significant difference between Mauritius and the USA (p = 0.393). The Czech religious stimulus had a higher impact on participants compared to the Mauritian one (p < 0.001), but again, no significant difference was found between Mauritius and the USA (p = 0.664). In order to investigate whether these differences affected decision-making in the Religious condition, we built a model with the number of matrices claimed as a dependent variable, and the religious stimuli’s characteristics as predictors. However, none of these characteristics explained any significant amount of variation in dishonest behavior in the religious condition (all ps > 0.29; see Table 6).

- 57 -

Figure 3. The condition effect divided by site with ±SEM. The only significant difference between conditions was found in Mauritius.

- 58 -

Table 3 Descriptive Statistics of Between-sites Differences in Dishonest Behavior (% Claimed)

Religious Secular Control Site n M SD CI d n M SD CI d n M SD CI d Mauritius 21 36.83 32.91 22.75-50.90 - 25 46.67 22.36 37.90-55.43 0.35 27 49.83 30.11 38.02-6074 0.40 Czech Rep. 27 21.73 19.27 14.46-28.30 - 27 20.00 22.57 11.49-28.51 0.08 24 20.56 18.43 13.18-27.93 0.06 USA 26 33.85 28.34 22.95-44.74 - 28 29.05 17.80 22.45-35.64 0.20 27 33.33 25.62 23.67-42.00 0.02

Note. CI = 95% Confidence intervals. Cohen’s d is the effect size of comparisons between the religious condition and the other conditions.

Tab1e 4 Estimates with SE from Beta Regressions for the Percentage of Matrices Claimed as Correct across our Three Sites. Mauritius Czech Republic USA

Model A Model B Model A Model B Model A Model B

Intercept 35.49 (15.24)** 35.49 (15.24) ** 20.20 (3.57) *** 21.70 (3.50) *** 33.82 (5.09) ** 32.46 (9.78) ** Secular 17.13 (11.33) * 17.13 (11.33) -2.54 (4.43) -4.29 (4.46) -3.17 (6.75) 0.49 (6.77) Control 13.83 (8.27) * 13.83 (8.27) Ϯ 0.93 (4.88) 0.23 (4.83) 1.00 (7.03) 4.48 (6.97)

Religiosity -4.72 (5.23) -4.40 (3.22) -5.04 (3.60) Secular*Religiosity 2.41 (9.87) 1.58 (4.20) 9.91 (5.36) Ϯ Control*Religiosity 7.58 (7.52) -0.57 (3.81) 9.89 (5.01) Ϯ Cox-Snell R2 .081 .085 .008 .077 .005 .068

Note. Models A describe condition effects for the three sites: Mauritius, Czech Republic, and USA. Models B display a Condition*Religiosity interaction for each site. In all models, the religious condition was set as a reference category. Ϯp < 0.1; *p < .05; **p < .01; ***p < .001

- 59 -

Figure 4. Differences between sites in religiosity and ritual participation frequency with ±SEM. Mauritian participants were significantly more religious and attended rituals more frequently than participants in the Czech Republic and the USA.

- 60 -

Table 5 Descriptive Statistics of Between-sites Differences in Religiosity and Religious-stimuli Ratings Mauritius (n = 73) Czech Republic (n = 78) USA (n = 81) Variable M SD CI d M SD CI d M SD CI d Religiosity 3.81 0.89 3.60-4.01 - 3.30 1.09 3.05-3.54 0.51 2.89 1.28 2.61-3.17 0.84

Ritual 4.21 1.59 3.84-4.57 - 2.65 1.73 2.27-3.04 0.94 3.33 1.98 2.90-3.76 0.49 participation Holiness 3.41 2.00 2.46-4.36 - 3.85 1.32 3.35-4.35 0.26 4.12 1.51 3.54-4.69 0.40 Negativity 2.78 0.76 2.42-3.14 - 2.66 0.89 2.32-2.99 0.15 1.55 0.50 1.36-1.74 1.93 Positivity 2.59 0.69 2.26-2.92 - 3.55 0.73 3.27-3.83 1.35 2.99 0.83 2.68-3.31 0.53 Tempo 3.15 1.21 2.57-3.72 - 2.30 0.72 2.02-2.57 0.85 2.90 0.85 2.58-3.23 0.23 Impact 2.88 1.10 2.36-3.40 - 4.00 1.07 3.50-4.41 1.03 2.75 0.75 2.46-3.04 0.14 Note. CI = 95% Confidence intervals. Cohen’s d is the effect size of comparisons between Mauritius and the sites.

Tab1e 6 Estimates with SE from a Beta Regression for the Percentage of Matrices Claimed as Correct in the Religious Condition. Intercept 30.87 (5.890)***

Positivity -0.539 (4.473) Negativity -0.703 (3.794) Tempo -3.227 (3.312) Impact -3.573 (3.365)

2 Cox-Snell R 0.028 Note. Differences between sites in the characteristics of religious stimuli do not explain differences in the number of claimed matrices. Ϯp < 0.1; *p < .05; **p < .01; ***p < .001

- 61 -

Discussion We tested the hypothesis that non-verbal religious primes in the form of religious music would decrease dishonest behavior compared to secular music and white noise. Whereas it has been previously shown that religious words and complex religious contexts (e.g., a church environment) can increase participants’ prosociality (Xygalatas, 2013), a possible effect of religion on deterring antisocial behavior was tested only by priming with religious words. We were interested in whether moral decision- making would be influenced by such a subtle cue as instrumental music. Participants in Mauritius, the Czech Republic, and the USA were given an opportunity to dishonestly inflate their performance in order to maximize their profit. This incentive to behave dishonestly was shown to be effective across all three sites. When collapsing all three sites together, we did not observe a significant effect of religious music on the rate of dishonest behavior. However, breaking down the condition effect by site revealed that religious music significantly decreased the incentive to cheat in Mauritius, but no such effect was observed in the other two sites. To test the hypothesis that the condition effect would be moderated by religiosity, we included a Condition∗Religiosity interaction in our models. Religious music significantly reduced dishonest behavior in religious participants, while ritual participation frequency played a marginally significant role in the religious condition. Males displayed higher rates of dishonesty across the three conditions. Finally, participants’ age and musical characteristics of the stimuli did not play a significant role. Together, these results offer a more nuanced interpretation of the influence of religious contexts on moral behavior.

It is important to acknowledge that the current study has several limitations. First, given the effect sizes for the differences between conditions at each site, we need to exert caution in interpreting the observed differences. While the collapsed sample across all sites is robust enough to detect medium effect sizes, the sample sizes at each site do not warrant generalizations due to low statistical power (Button et al., 2013). Furthermore, since the effect sizes of the differences between conditions in Mauritius are rather small (0.3 and 0.4), this finding needs to be further probed by future studies. Second, we did not collect exact data on actual cheating. While our procedure should secure confident estimates of unethical behavior, it is still possible that some participants correctly solved more than 5 matrices and vice versa. Similarly, some participants could feel that they found a correct answer and that the answer we provided was incorrect. Since the mathematical equations were computed under time-pressure, participants could make a small mistake without noticing and feel righteous to claim their answer as correct. However, given our overall sample size, such participants should constitute

- 62 -

only a minimal portion of our sample. Third, since the musical stimuli were played before the mathematical task, their effects could be concealed by the time delay or the cognitive demands of the task. Perhaps if the stimuli were played during the whole experiment, the primes would be more salient and thus capable of influencing participants’ behavior to a greater extent. Such a proposition needs further empirical testing. Fourth, the religiosity effect could have been mediated by some other mental process than by an association to normative behavior. For example, the thought of religion could have primed global processing, which has been previously shown to increase prosocial behavior (Mukherjee, Srinivasan, & Manjaly, 2014).

The lack of a main condition effect in the overall sample suggests that religious music might not always be salient enough to deter people from dishonest behavior. Although our religious stimuli were recognized as significantly more holy than the other two stimuli, honesty was only affected in one of three sites. A significantly lower amount of dishonest behavior in the religious condition was observed only in Mauritius, which points to the need for a more thorough understanding of differences between our sites. There are at least three possible interpretations: (a) this finding is a false positive; (b) participants in Mauritius were induced with different emotions that influenced their behavior; or (c) the association between religious music and normative behavior is stronger in Mauritius due to higher religiosity.

The observed difference between different conditions in Mauritius could have been caused by different characteristics of our religious stimuli. While we used organ music in the Czech Republic and the USA, the Mauritian religious stimulus had significantly higher tempo and dominant drums. A comparison of religious stimuli across sites revealed mixed results. The Mauritian religious music was perceived as significantly more negative than the religious stimulus in the USA, while there was no difference between Mauritius and the Czech Republic. We can speculate that, for example, Mauritian participants were more avoidant and critical due to higher negativity evoked by the religious stimulus and, consequently, avoided the cheating behavior. However, we find this interpretation unlikely because the perceived negativity of the stimuli was not significantly different between Mauritius and the Czech Republic. Similarly, differences between Mauritius and the other sites in positivity, tempo, and impact were always only between two sites, suggesting that no systematic differences were related to those properties. Furthermore, looking at the overall effects of musical characteristics on cheating rates, we did not observe any significant influence of these variables. This is in contrast with previous research which suggested that positively valenced music decreases moral concerns (Ziv et al., 2012). The lack of such effects might stem from the fact that the link between

- 63 -

positive music and cheating was previously tested only by self-reports (Ziv et al., 2012). Alternatively, the cognitive demands of our task might have concealed any tentative subtle effects of musical characteristics.

The overall higher rates of self-reported religiosity and ritual participation frequency in Mauritius appear to be a more probable explanation of the behavioral differences between our sites. Religiosity is entrenched into Mauritian everyday life much more than in the other two sites, and might play a more important normative role (Xygalatas, 2013). This was confirmed by the significant differences in reported religiosity and frequency of ritual participation between Mauritius and the other two sites, and might indicate that higher religiosity could be associated with heightened sensitivity to religious cues (for similar results on prosocial behavior see Xygalatas et al., 2015). This interpretation is further supported by the significant

Condition∗Religiosity interaction. Collapsing all three sites, higher religiosity was associated with decreased rates of dishonest behavior in the religious condition. Although participants recognized our stimuli as religious, the less religious participants seemed to be unaffected. This result is in contrast with previous studies that showed no effect of religiosity on overall cheating rates (Randolph-Seng and Nielsen, 2007; Mazar et al., 2008; Aveyard, 2014). Our study thus offers new preliminary evidence on the role of religiosity, in congruence with the research on religious prosocial behavior (Shariff et al., 2016).

The fact that religiosity had a significant impact on dishonest behavior only in the religious condition supports the important role of religious situational factors in decision- making. We propose that dispositional religiosity does not affect participants’ honesty to a large extent, unless it is activated by environmental sacred cues (Darley and Batson, 1973; Norenzayan and Shariff, 2008; Xygalatas, 2013; Xygalatas et al., 2015). While Mauritian participants reported significantly higher religiosity than participants at the other sites, the Mauritian cheating rates were significantly higher than those in the Czech Republic and the USA. Such a finding suggests that participants needed to be reminded of their religiosity in order for it to affect their moral decision-making. However, such a “reminder effect” is probably temporary (Malhotra, 2008) and confined only to religious participants. When religious cues are salient and general enough (e.g., the word God), they might affect non- religious participants, thus masking the effect of dispositional religiosity. But when subtle (as in the case of our study), these sacred cues only influence religious people who are more sensitive to them. This could also explain why studies that used linguistic primes (Randolph- Seng and Nielsen, 2007; Mazar et al., 2008) did not find a significant moderating effect of

- 64 -

religiosity. Religious words are part of the standard cultural language toolbox and have stronger behavioral associations than specific religious symbols. For example, the Islamic call to prayer is a public, omnipresent cue that is directly associated with specific behaviors. As such, these cues are less ambiguous than music (Cross and Morley, 2008). Instrumental religious music, on the other hand, is generally less known, and associative learning is rather accomplished via communal socialization that reinforces the association of symbols with religion. Music is rarely associated with specific behavioral requirements, especially those regarding moral conduct. Behavioral schemas are thus not directly accessible to those who have not undergone religious socialization and do not participate in communal ritual gatherings (while they might be accessible to the majority of people through words). The fact that music is such a subtle cue can explain why we did not observe a significant Condition∗Religiosity interaction in each of our sites. We would probably need larger sample sizes in order to show such an interactive effect.

The importance of ritual participation in the accessibility of behavioral schemas is further supported by a trend in the Condition∗Ritual participation interaction. The fact that this trend did not reach statistical significance, however, suggests that ritual participation alone might not be enough to promote honest behavior (Mitkidis, Lienard, Nielbo, & Sørensen, 2014). It may reinforce the link between symbolic and behavioral schemas, but this link without an overarching religious worldview is probably a weak motivational force. Although participation in public rituals usually signals acceptance of religious norms (Rappaport, 1999), it is not necessarily tied to actual normative behavior and people can participate in these rituals for various reasons, for instance, reducing anxiety (Lang, Krátký, et al., 2015), including no specific reason at all (Xygalatas, 2012). Such participants might be less inclined to follow normative schemas prescribed by their respective religions, especially if different behaviors have momentarily higher pay-offs (free-riding). Furthermore, ritual intensity may play an important role in the reinforcement of the link between symbol and behavior. High-intensity rituals are usually extremely arousing events (Xygalatas, Mitkidis, et al., 2013; Xygalatas, Schjødt, et al., 2013), and as such might yield stronger affective bonds between symbols and conceptual complexes (Alcorta and Sosis, 2005). This might provide additional support for the suggested explanation of the differences in dishonest behavior between our sites. In Mauritius, we used music from the Kavadi ritual as the religious stimulus. The Kavadi is a high-intensity ritual that involves multiple body piercings, walking on nails, carrying heavy objects, and other forms of prolonged suffering. As such, it might be especially powerful in associating the

- 65 -

musical stimulus with specific behavioral requirements and might have provided sufficient motivation for moral behavior that was not reached by religious stimuli that referred to less intense rituals in the other sites. This interpretation gains additional support by field experimental evidence that self-reported frequency of participation in the Kavadi ritual significantly predicted lower amounts of dishonest behavior in an economic game (Xygalatas et al., in press). We thus suggest that participation in high-intensity rituals might be effective in transforming behavioral requirements into symbols and as such be a powerful motivational force.

Our findings might be of importance for evolutionary models of music and its functions. Evolutionary theorists have disagreed on whether music is an evolutionary by-product or an adaptation. The by-product thesis argues that music parasitizes upon our evolved language abilities. In fact, Steven Pinker (1998) has dubbed music an “auditory cheesecake.” According to this view, our love for music is a by-product of specific cognitive-linguistic capacities, just like our love for junk food is a by-product of our adaptive need for fat, salt, and sugar. Others, however, point to the ubiquity of music across all cultures, as well as the fact that language and musical abilities are not strictly cognitively overlapping, and argue that music-making might have evolved as an adaptive trait (Fitch, 2006). For example, it might be an important tool for sexual selection, much like in birds (Miller, 2000), as suggested by the sex appeal of musical celebrities. Another important function might be related to an endorphin-based social binding mechanism (Dunbar et al., 2012) whereby music can function as social glue, a sort of “vocal grooming” (Weinstein, Launay, Pearce, Dunbar, & Stewart, 2015). While these functions are not mutually exclusive, here we demonstrate that music may serve yet another function, that of representing norms and influencing behavioral schemas. We suggest that it does so via associative learning in communal gatherings where conceptual complexes are encoded in memory together with music. This link might be even stronger when norm-related words are included to create a song. Such songs can trigger outbursts of connotations, and thus function as a compact version of normative conceptual complexes, becoming effective vehicles for the transmission of social norms.

In summary, the current study provides preliminary support for the hypothesis that instrumental music can serve as a reminder of normative behavior, but only for participants who previously formed an association between religion and specific music. This result suggests that while socialization into group norms is crucial for ethical behavior, people need to be reminded of these norms to ensure an activation of normative behavioral schemes. In this

- 66 -

respect, religion is a powerful institution that fosters normative behavior via shared rituals, repetitive songs and prayers, and other symbols that can act as associative triggers of ethical behavior. Further research should also investigate whether a combination of these triggers might possibly amplify their effects on participants’ decision making. Likewise, using multiple sites within different cultural contexts in future research might help increase the reliability of priming studies and address the reproducibility crisis in psychological research (Open Science Collaboration, 2015).

- 67 -

Lost in the Rhythm: Effects of Rhythm on Subsequent Interpersonal Coordination Lang, M., Shaw, D.J., Reddish, P., Wallot, S., Mitkidis, P., & Xygalatas, D.

Abstract Music is a natural human expression present in all cultures, but the functions it serves are still debated. Previous research indicates that rhythm, an essential feature of music, can enhance coordination of movement and increase social bonding. However, the prolonged effects of rhythm have not yet been investigated. In this study, pairs of participants were exposed to one of three kinds of auditory stimuli (rhythmic, arrhythmic, or white noise) and subsequently engaged in five trials of a joint action task demanding interpersonal coordination. We show that when compared with the other two stimuli, exposure to the rhythmic beat reduced the practice effect in task performance. Analysis of the behavioral data suggests that this reduction results from more temporally coupled motor movements over successive trials and that shared exposure to rhythm facilitates interpersonal motor coupling, which in this context serves to impede the attainment of necessary dynamic coordination. We propose that rhythm has the potential to enhance interpersonal motor coupling, which might serve as a mechanism behind its facilitation of positive social attitudes.

Keywords: Rhythm; Interpersonal coordination; Motor coupling; Social bonding;

Introduction Music is omnipresent across all cultures and dates back to the evolutionary origins of Homo sapiens (Adler, 2009; Conard, Malina, & Münzel, 2009). Its ubiquity and conservation has led to various speculations about the societal functions music might serve (Dissanayake, 2006; Fitch, 2006; Huron, 2001). One key function appears to be the facilitation of cooperation and interpersonal coordination (Ian Cross & Morley, 2008; Dunbar et al., 2012; Kirschner & Tomasello, 2010), yet the way in which music exerts this effect remains unknown. Brown (2000) proposes four possible routes through which music might function: group identity (e.g., anthems), group cognition (communication of ideas), group catharsis (synchronizing of emotions), and group coordination (synchronization and harmonization). While all four aspects are highly relevant, we were interested particularly in coordination; through shared sensory

- 68 -

input, we believe that music facilitates the temporal organization of movements between individuals, and a number of studies have shown that such interpersonal coordination can subsequently increase social bonding (Hove & Risen, 2009; Reddish et al., 2013; Shaw, Czekóová, Chromec, Mareček, & Brázdil, 2013; Wiltermuth & Heath, 2009).

The key element of music facilitating temporal coordination is rhythm—the metric organization of sounds. Rhythm enables groups of individuals to entrain to a common beat (Merker, Madison, & Eckerdal, 2009) and sustain stable motor patterns (McNeill, 1995). Interestingly, rhythmic entrainment appears to be an inborn human ability, observed even in infants (Phillips-Silver & Trainor, 2005; Winkler, Háden, Ladinig, Sziller, & Honing, 2009; Zentner & Eerola, 2010). Such audiomotor integration, or sensorimotor synchronization (SMS), has been investigated in numerous studies (for an extensive review see Repp & Su, 2013); these include tapping tasks (J. Chen, Zatorre, & Penhune, 2006; Dhamala et al., 2003; Konvalinka, Vuust, Roepstorff, & Frith, 2010), pendulum swinging (Schmidt, Richardson, Arsenault, & Galantucci, 2007; Varlet, Marin, Issartel, Schmidt, & Bardy, 2012), walking (Styns, van Noorden, Moelants, & Leman, 2007), rocking in chairs (Demos, Chaffin, Begosh, Daniels, & Marsh, 2012), and dancing (Miura, Kudo, Ohtsuki, & Kanehisa, 2011; Van Dyck et al., 2013). Together, these studies demonstrate the influence of rhythm on various aspects of motor coordination.

In all of the aforementioned studies, the effect of an auditory rhythmic stimulus was measured on an intrinsically rhythmic motor task (e.g., rhythmic tapping) performed synchronously to music. None of them assessed if the effects of rhythm on interpersonal coordination and social behavior can (a) extend beyond the period of listening and (b) manifest in tasks unrelated to an auditory stimulus. In other words, if rhythm remains an attractor of coordination, does this manifest even in tasks that pull participants away from rhythmic movement and create dissonance in motor structures?

The present study explored whether exposure to an auditory rhythm influences subsequent interpersonal coordination on a rhythmically unrelated task; specifically, a task that requires dynamic coordination rather than coupled motor performance. To isolate the effects of rhythm, we designed three conditions in which dyads listened passively to either a metrically structured beat (Rhythmic condition), a chaotic, unpredictable beat (Arrhythmic condition), or white noise (Control condition). Following exposure to one of these stimuli, pairs of participants were tested on their ability to coordinate their movements on the labyrinth task employed by Valdesolo, Ouyang, and DeSteno (2010). These authors showed that joint

- 69 -

synchronous action increased perceptual sensitivity to the movements of the other participant and subsequently enhanced coordination on this labyrinth task.

We selected this task specifically because it permits a dissociation between complementary motor coordination and motor coupling. While the former can consist of two different movements performed asynchronously but with the same underlying goal (tilting the labyrinth in opposite directions with varying angles to control the trajectory of the ball), the latter refers to movements performed with tight temporal and spatial synchrony. In terms of our research question, it allowed us to see if rhythm exerts a prolonged influence on participants’ motor structures by pushing them towards motor coupling.

Since the labyrinth task requires dynamic, responsive, and fluid movements (thus high perceptual sensitivity to the other), we expected the rhythmic beat to attract dyads to more rhythmic, temporally coupled movements that are detrimental to performance in the labyrinth task. That is, we hypothesized that the rhythmic pattern will remain an attractor of coordination, thereby interfering with the responsiveness of participants’ movements to one another. Furthermore, we predicted that motor coupling of participants in the Rhythmic condition would extend over multiple trials of the task, demonstrating the persistent influence of the rhythmic stimulus. We also assessed whether rhythm influences social attitudes toward interaction partners, predicting more positive attitudes among dyads in the Rhythmic condition.

Materials and Methods Participants

One hundred subjects (50 females; Mage = 23.8 years, range = 21–29 years) were recruited from the student population of Masaryk University, Brno, Czech Republic, and rewarded with course credits for participation. From this sample, 50 dyads comprised individuals of the same sex, same dominant hand, and similar height. Due to a malfunction of recording equipment, three dyads were omitted from the final analyses. The study protocol was approved by the ethical committee of Faculty of Arts, Masaryk University, and informed consent was obtained from all subjects.

- 70 -

Materials

In a double blind design, dyads were divided randomly into three conditions defined by the type of auditory stimulus to which they were exposed before the labyrinth task – Rhythmic (15 dyads), Arrhythmic (16 dyads), and Control (16 dyads). The rhythmic and arrhythmic stimuli comprised the same number of kick and snare drum beats with strong bass line and tempo of 120 BPM. The only difference between these two stimuli was the metric pattern: Beats in the Rhythmic condition had a 4/4 metrical pattern with an inter onset interval (IOI; or inter beat interval) of 500 ms, and with a distinctive syncopation added to the “third” beat (the syncopation index described by Longuet-Higgins & Lee was 1; Fitch & Rosenfeld, 2007; Longuet-Higgins & Lee, 1984). Here, the kick drum arrives also on the pause between beat 3 and 4: “(1)kick-500 ms-(2)snare-500 ms-(3)kick-250 ms-kick-250 ms-(4)snare-500 ms” (Butler, 2006). A syncopated rhythmic pattern was chosen over a simple metronome to simulate a more natural musical environment, which usually contains complex metric patterns (Large & Palmer, 2002; London, 1995). For the Arrhythmic condition, the beats comprising the rhythmic stimulus were distributed with random IOIs so as to eliminate any meter (IOIs ranged between 0 and 800 ms, with SD = 202.476 ms). The control stimulus was created by converting the rhythmic auditory stimulus to numbers via MATLAB; randomizing those numbers; and converting them back to auditory stimulus, producing a constant white-noise sound.

All stimuli were of identical durations (4 min). The rhythmic and arrhythmic stimuli were presented at 70 dB, while the control stimulus had, naturally, a lower volume of 43 dB. To ensure full and comparable attention was paid to the stimuli in all conditions, each stimulus was supplemented with 16 bell-ringing sounds that participants counted and reported in a subsequent questionnaire. Participants were told that the purpose of the study was to investigate the memory of sounds, and emphasis was placed on the bell-counting task.

Procedure

Dyads were seated back-to-back (0.7 m distance) and instructed not to move or talk during the auditory stimulus, thereby avoiding any social interaction. Subsequently, they stood facing each other at a distance of 0.5 m holding a wooden labyrinth (12 X 9 X 14 cm) with a steel ball (adopted from Valdesolo et al., 2010). Their task was to jointly navigate the steel ball through

- 71 -

the labyrinth, and success was possible through one path only. Before the labyrinth task, participants were instructed to try to achieve the best possible time during five trials of the task.

To assess whether the rhythmic stimuli exerted a prolonged effect after exposure, completion times were recorded for each of the five trials. During the labyrinth task, hand- movement acceleration was recorded in three dimensions by ActiGraph Motion Sensors GT3X (D. John & Freedson, 2012) positioned on participants’ wrists, operating at a sampling rate of 30 Hz.

At the end of the procedure, participants were separated into different rooms and asked to fill out a questionnaire. Items concerned the auditory stimuli (8 items), the labyrinth task (9 items), and the participants’ partner (8 items), each scored by way of 140 mm visual analog scales (Bond & Lader, 1974) anchored by positive and negative extremes. Items about participants’ partners were combined into four measures reflecting distinct attitudinal dimensions: closeness (the Inclusion of Other in the Self Scale; Aron, Aron, & Smollan, 1992), liking, dominance, and cooperation. Debriefing was performed at the end of data collection.

Data analysis

Raw hand-acceleration data were first preprocessed so as to allow the extraction of movements irrespective of directionality (see Appendix). Next, we investigated if the rhythmic beat to which participants in the Rhythmic condition were previously exposed remained an active attractor of coordination during the subsequent labyrinth task. Since successful performance on the labyrinth task demands movements of an aperiodic nature, it serves as a strong attractor that opposes simple phase locking to the beat. As such, we considered a measure of relative phase to be inappropriate (M. J. Richardson, Marsh, Isenhower, Goodman, & Schmidt, 2007). Instead, we estimated unintentional entrainment to the frequency of the rhythm, which would reduce the degrees of freedom necessary for coordination. In other words, rather than a simple one-to-one mapping of the beat and movement, we expected the movements of participants in the Rhythmic condition to be attracted by the most prominent frequency of the beat to which they were exposed previously. This situation is illustrated in Figure 6A, where movements appear attracted to the frequency of the beat comprising the rhythmic stimulus, but performance on the labyrinth task serves to mask this entrainment.

To assess the strength of the beat as an attractor of coordination, a Fast Fourier Transform (FFT) was applied on the preprocessed hand-movement data. The frequency with

- 72 -

maximal power was selected within a non-overlapping moving window of 7s duration (as a compromise between temporal resolution and frequency resolution), and this was compared with the dominant frequency of the beat (Bruyn, Leman, Moelants, & Demey, 2009; Desmet, Leman, & Lesaffre, 2010). Since the beat had a dominant frequency of 2 Hz, synchronization within a window was detected when the movements of participants’ hands had maximal power in this frequency, with a tolerance of ± 2 data points (1.76 and 2.31 Hz respectively). Importantly, participants with longer completion times experienced a greater interval after exposure to the auditory stimulus in their last trials compared to participants with shorter completion times. To control for this, we performed the FFT analysis on a time window equal to the shortest hand-movement time series (87.55 s = 12 windows). The ratio of synchronous to asynchronous windows was counted to obtain a percentage of time in which participants moved with a dominant movement frequency similar to the frequency of the rhythmic stimulus. Due to the length of the moving window, we were not able to analyze individual trials – there would be only approximately two windows per trial. Therefore, we report just the condition effects.

To quantify participants’ interpersonal motor coupling, we employed cross-recurrence quantification analysis (CRQA) on the hand-movement data. CRQA is a phase space based analysis that quantifies the coupling between two signals (Shockley, Butwill, Zbilut, & Webber, 2002), in our case the acceleration of participants’ hands. When two signals follow the same trajectories in time, they run through the same phase space neighborhood (defined by radius) and CRQA quantifies the extent to which two signals share that same neighborhood (Marwan, Romano, Thiel, & Kurths, 2007).

The time series for each hand of participant A and B were paired on the basis of movements required for operating the labyrinth (i.e. dominant A–dominant B; and non- dominant A–non-dominant B; see Supplementary Materials, Figure S3), and the measure of determinism (%DET) was computed using the MATLAB CRP Toolbox 5.17 (Marwan & Kurths, 2002). %DET is a percentage of recurrence points forming a diagonal line in a recurrence plot, thereby reflecting similar developments of trajectories and indicating predictability of the system (Marwan et al., 2007), in our case, shared acceleration of hand movements (for details on CRQA analysis see Appendix, CRQA computation).

Due to the clustered nature of the hand-movement data, results from CRQA were analyzed with linear mixed-model regression (LMM) using SPSS (Version 21.0; IBM corp., Armonk, NY, USA). Using a stepdown approach (West, Welch, & Galecki, 2007), all the logical fixed

- 73 -

effects, interactions, repeated, and random effects were added initially to a base model and removed subsequently on the basis of improvement in the log-likelihood ratio (p < .05). This technique yielded the most parsimonious model, which still accounted for effects of clustering in the data (see Supplementary Material, Equation 1). In addition to CRQA measures, we were interested in the mean of movement acceleration and the total number of movements, as measured by the ActiGraphs (see Appendix, Movement computation). For the analysis of these data we created a second LMM (Supplementary Material, Equation 2). The Rhythmic condition was used as reference for both models, against which the Control and Arrhythmic conditions were compared.

Results As a manipulation check, a one-way ANOVA was performed on ratings of perceived rhythmicity of the stimulus. This revealed a main effect of Condition [F (2, 91) = 18.104, p < .001], with post hoc comparisons confirming that perceived rhythmicity was greater in the Rhythmic compared with the Arrhythmic and the Control condition (Gabriel correction; ps < .001).

Joint action completion time

The time needed to complete the labyrinth task for each dyad and on each trial was analyzed with a two-way mixed-model ANOVA, in which Condition comprised the between subject factor and Trial the repeated measures factor. Contrary with our hypotheses, there was no main effect of Condition [F (2, 44) = 1.258, p = .294, r = .06]. There was, however, a significant main effect of Trial [F (4, 41) = 16.372, p < .001]29 revealing a linear decrease in completion time. We refer to this as the “practice” effect herein. Furthermore, there was a Condition*Trial interaction [F (8, 84) = 2.427, p = .021], revealing differences in the practice effect between conditions. Specifically, the differences in completion time across trials were significant for the Arrhythmic [F (4, 41) = 14.397, p < .001] and Control conditions [F (4, 41) = 6.143, p = .001], but not for the Rhythmic condition [F (4, 41) = 1.432, p = .241].

Post hoc pairwise comparisons (Sidak corrections) revealed significant decreases in completion time between the first trial and all subsequent trials for the Arrhythmic condition

29 We are reporting multivariate tests because Muachley’s test indicated that the assumption of sphericity had been violated (χ2 (9) = 17.489, p = .042).

- 74 -

(ps < .001). For the Control condition, significant decreases existed between the first trial and the third and fourth trials (ps < .05). There were no significant differences between trials in the Rhythmic condition. In other words, a greater practice effect was observed in the Control and Arrhythmic condition relative to the Rhythmic condition. This is presented in Figure 5.

Figure 5: Mean completion times of the labyrinth task for five trials with ±SEM, showing linear decrease in the Arrhythmic and Control conditions, but not in the Rhythmic condition.

Movement kinematics

To achieve a more precise understanding of movement dynamics between individuals during the labyrinth task, we examined movement kinematics; specifically, we measured hand acceleration with ActiGraph Motion Sensors GT3X (ActiGraph; Pensacola, FL, USA). First, we tested if the dominant frequencies of the beat comprising the rhythmic stimulus manifests in the movements of individuals exposed to it previously; second, we assessed the number of

- 75 -

movements and mean acceleration of movements for each individual; and third, we quantified interpersonal motor coupling in the labyrinth task.

The percentage of seconds that dyads moved at the same dominant frequency as the beat comprising the rhythmic stimulus was analyzed with a one-way ANOVA, where Condition served as a between subject factor. This demonstrated a significant effect of Condition [F (2, 91) = 5.090, p = .008], with post-hoc pairwise comparisons (Gabriel correction) revealing a significantly higher percent of synchronized seconds in the Rhythmic compared to the Arrhythmic (p = .011) and Control conditions (p = .041). This effect is illustrated in Figure 6.

The results of the linear mixed-model regressions (LMMs) applied to the number of movements, mean movement acceleration, and determinism (%DET) are summarized in Table 7, and fitted lines are plotted in Fig. 7. Converging with the completion time results, we did not observe any main effect of Condition (except for %DET), but we did observe differing patterns of movement parameters over trials across the three conditions. The importance of these variables for performance in the labyrinth task is supported by significant correlations of completion time with number of movements (r = .442, p = .002), mean acceleration (r = -.577, p < .001), and %DET (r = .300, p = .041).

While there was no difference between conditions in the number of movements executed on the first trial (no significant difference in intercepts), a significantly greater linear decrease in the number of movements was identified for the Arrhythmic and Control conditions compared to the Rhythmic condition. We also observed a significant difference in the trajectory of mean movement acceleration between the first and last trial in the Arrhythmic and Control conditions, but not in the Rhythmic condition (see Fig. 7A, B).

These findings are complemented further by the results from CRQA, which show a significantly lower decrease in %DET for the Rhythmic condition compared to the Control and Arrhythmic conditions (Fig. 7C). Although participants’ movements in the Control and Arrhythmic conditions exhibited more coupling initially, they acquired more flexibility in a manner corresponding to the practice effect. No such effect was observed for the Rhythmic condition, however, where movements remained coupled across all five trials.

Together, these results demonstrate that the practice effect in the Rhythmic condition was significantly different compared to the Control and Arrhythmic conditions in measures of

- 76 -

movement flexibility and responsiveness. In summary, the ActiGraphs uncover a logical, intuitive relationship between speed of task performance and movement kinematics.

Table 7. Coefficients from mixed model regression on dependent variables from Actigraph data, measured over five trials

Intercept Variable Condition Trial Condition*Trial Trial² Condition*Trial² (Rhythmic)

Control Arrhythmic Control Arrhythmic Control Arrhythmic

Number of 74.19 3.77 5.20 -3.48 -5.39 -6.27 0.40 0.89 0.83 movements (3.69)** (5.14) (5.14) (1.78) (2.47)* (2.47)* (0.27) (0.37)* (0.37)*

Mean 16.47 -2.33 -3.25 -0.31 3.80 2.35 0.01 -0.42 -0.26 acceleration (1.69)** (2.35) (2.35) (0.63) (0.88)** (0.88)** (0.08) (0.11)** (0.11)*

23.07 8.04 10.06 -1.09 -7.48 -5.69 0.17 1.14 0.68 %DET (3.09)** (4.30) (4.30)* (1.99) (2.77)** (2.77)* (0.30) (0.42)** (0.42)

Note: Trial expresses the linear effect, Trial² the quadratic effect. To avoid over-parametrization, the Rhythmic condition was set as a reference category. The variable Mean acceleration was multiplied by 100 for easier reading. *p < 0.05; **p < 0.01.

- 77 -

Figure 6: A. First 10 seconds of hand movements from a participant in the Rhythmic condition plotted against the occurrence of beats in a 10-second segment of the rhythmic stimulus. This illustrates the difference between timing of movements required by the labyrinth task and the structure of the rhythmic beat. B. Results from a FFT analysis showing that participants in the Rhythmic condition exhibited movement frequencies similar to the beat for a higher percent of time. Error bars represent standard error. R = Rhythmic; A = Arrhythmic; C = Control.

Figure 7: Fit lines created on the basis of coefficients from the mixed-model regression applied to outcome variables from Actigraph data. Development of each trajectory is plotted against the five trials of the labyrinth task. Figures illustrate the difference between conditions in A. the decrease in number of movements; B. the increase of mean acceleration; and C. the decrease in %DET.

- 78 -

Questionnaire

Of the four attitudinal measures (closeness, liking, cooperation, and dominance), only liking had a marginally significant difference between conditions [F (2, 91) = 2.972, p = .056], pointing to a higher liking in the Rhythmic condition compared to the Arrhythmic and Control conditions. However, post hoc comparison revealed no significant differences between the Rhythmic and the other two conditions (Gabriel correction; Arrhythmic: p = .074, Control: p = .158). We also investigated correlations between liking and our measures of movement kinematics. This revealed a positive correlation between liking and %DET (r = .283; p = .054). In other words, greater motor coupling was related to greater interpersonal liking. Interestingly, our data also revealed a strong positive correlation between mean completion time of the labyrinth task and perceived liking (r = .726; p = .002) and closeness (r = .727; p = .002) in the Rhythmic condition only.

Discussion In the present study, we investigated the influence of rhythm on subsequent interpersonal behavior. Specifically, we examined how pairs of individuals performed on a joint action task after exposure to rhythmic, arrhythmic, or control auditory stimuli. This revealed three important findings: First, we observed that, having listened to a rhythmic beat, individuals’ movements become more aligned to the frequency of that beat; second, we reveal that this apparent alignment to the rhythm manifests even in the face of task demands, interfering with dynamic interpersonal coordination; and third, we show that when alignment to the rhythmic stimulus occurs in two interacting individuals, manifesting as increased motor coupling, their interpersonal attitudes toward one another become more positive.

To assess the effects of rhythm on interpersonal motor behavior, we employed a specific joint action task—the labyrinth game. Successful completion of this structured and complex task demands flexible and complementary timing of movements between two individuals; they are required to respond dynamically to each other’s movements, acting as an interactive pair rather than a synchronized unit. The strong practice (i.e., Trial) effect reveals that participants learned to coordinate their movements over repeated trials. Importantly, however, the Trial*Condition interaction shows that this improvement in completion time occurred only for pairs exposed to the arrhythmic and control stimuli. We propose that, over the course of five trials, participants’ movements in the Rhythmic condition were attracted to the frequency of the rhythmic stimuli, which served to inhibit the responsive, reactive, and complementary

- 79 -

movements achieved with practice in the other conditions. In support of this claim, we showed that the frequencies of hand movements performed by participants in the Rhythmic condition were more synchronized to the dominant frequencies of the rhythmic beat than those of participants in the other conditions.

Next, we observed a decrease in the number of movements over successive trials in the Control and Arrhythmic conditions, but not in the Rhythmic condition. Successful performance on the labyrinth game requires complementary and responsive movements between individuals to navigate the ball through the narrow corridors and turns comprising the maze. Simultaneous movements led to overshooting, which then require additional corrective adjustments. Furthermore, differences between the conditions vis-a-vis the practice effect in hand- movement acceleration indicate a lower reduction in movement variability across trials in the Rhythmic condition relative to the Arrhythmic and Control conditions. These results show that hand movements in the Rhythmic condition expressed low acceleration and lack of spatial movement freedom across successive trials. Finally, our CRQA supplemented these other measures by demonstrating similar movement patterns between the hands of individuals exposed to the rhythmic auditory stimulus. An absence of decrease in %DET for the Rhythmic condition corresponds to the decreased variability of movements at the level of dyads. Since %DET is a measure of periodicity and predictability of a signal, this finding is indicative of repetitive and predictable movement acceleration.

In this light, the absence of a Condition effect on completion time and movement kinematics might at first glance contradict this interpretation, since it may be predicted that rhythm would produce the strongest negative effect directly after it was perceived, and attenuate with time. However, we suggest that this finding reflects the very nature of human task sharing and joint- action co-representation; while task sharing concerns overall goals and intentions, action co- representation reflects the mechanics of labor division, including anticipation of our interaction partners’ movements and adapting our own movements accordingly (de Bruijn, Miedl, & Bekkering, 2011; Holländer, Jung, & Prinz, 2011; Pezzulo & Dindo, 2011; Sebanz, Bekkering, & Knoblich, 2006; Sebanz, Knoblich, & Prinz, 2005; Wenke et al., 2011). Since our participants did not interact with each other prior to the labyrinth game, they shared a task goal, but lacked co-representation of the other’s actions. Such lack of experience presumably prevented successful anticipation of the other’s movements in the first trial and initially inhibited necessary responsiveness, resulting in coupled rather than complementary movements in all conditions (reflected in our measure of %DET). In contrast, the improved

- 80 -

performance over increasing trials in those participants in the Arrhythmic and Control conditions suggests their ability to achieve better coordination with one another. It is thus likely that better alignment between partners’ co-representations of their actions (see Pezzulo & Dindo, 2011) allowed them to adapt their actions accordingly. The significant trial effect in %DET might reflect this (for similar results see Knoblich & Jordan, 2003; Newman-Norlund, Bosga, Meulenbroek, & Bekkering, 2008; van der Wel, Knoblich, & Sebanz, 2011). Such alignment was not observed in the Rhythmic condition, and we speculate that during the five trials rhythm masked sensitivity to one another’s movements by attracting them to a set frequency.

Our interpretation of the findings is in line with previous research in dynamic attention theory (DAT; Escoffier, Sheng, & Schirmer, 2010; Large & Jones, 1999), which showed that rhythm is a strong attractor of attention, even if it impairs performance (Brochard, Tassin, & Zagar, 2013). Moreover, auditory stimuli have been shown to be more powerful than visual in driving movements (Y. Chen, Repp, & Patel, 2002; B. Repp & Penel, 2002, 2004; Varlet, Marin, Issartel, Schmidt, & Bardy, 2012), which supports the interpretation that the auditory rhythm resonating in participants’ motor structures (Nozaradan, Peretz, Missal, & Mouraux, 2011) might in some instances decrease perceptual sensitivity to one another (Demos et al., 2012). In other words, perceptuomotor processes might be influenced by a rhythmic beat in a way that interferes with the complementary temporal behavior required for the labyrinth task.

We further speculate that shared listening to rhythm serves to entrain a common timing behind interpersonal motor coding. Although our measurements provide no direct evidence for such an effect, several neuroscientific findings lend support to this interpretation. For example, the interplay between motor and auditory systems has been demonstrated by several studies that identify specific brain structures (premotor cortex, supplementary motor area, and cerebellum) engaged during passive listening to rhythmic music (Baumann et al., 2007; J. Chen, Penhune, & Zatorre, 2009; Grahn & Brett, 2007; Lahav, Saltzman, & Schlaug, 2007; Schubotz, Friederici, & von Cramon, 2000). These neural systems are necessary for preparing, timing, sequencing, and coordinating movement, as well as for the production of rhythm (Dhamala et al., 2003). From this evidence we might conjecture that rhythm activates the neural motor circuits of two participants in a similar manner corresponding to the structure of the beat.

Interestingly, our data lends further empirical support to the hypothesis that motor coupling can increase positive affect toward other individuals (Miles, Nind, & Macrae, 2009; Reddish et al., 2013; Valdesolo & Desteno, 2011; Wiltermuth & Heath, 2009) In contrast with

- 81 -

the previous studies, however, our finding is based on spontaneous motor coupling induced by rhythm rather than on predefined synchronous movement patterns. While more evidence is needed before any firm interpretation is justified, our findings may go some way toward an explanation for the natural connection between music, movement, and social bonding. Such an interpretation would be supported by the positive correlation between the measure of motor coupling (%DET) and liking. Moreover, liking and closeness in the Rhythmic condition correlated positively with completion time, suggesting that increased motor coupling was associated with impaired task performance on one hand, but enhanced positive attitudes on the other (Demos et al., 2012).

It is important to acknowledge potential limitations with the present study, which may be addressed by future investigations. First, the metric structure that we used was familiar among participants due to its frequent use in popular songs in the cultural milieu of the Czech Republic, and it remains to be seen whether this meter would also work in other cultures with different musical traditions. It is necessary, therefore, to investigate metric structures cross- culturally and to assess whether alternative structures yield different results (Chen et al., 2006). For example, it might be fruitful to compare the effects of various meters and tempos on motor co-ordination; to design a rhythmically related task to observe how an extension of the beat influences interpersonal motor coordination (e.g., rowing, where precise movement timing and power are required to keep a boat on a straight trajectory); to expose each participant to a stimulus with different rhythmic pattern; or to measure the temporal extent of motor coupling between individuals (e.g., using the automatic imitation paradigm; see Shaw et al., 2013). Alternatively, neuroimaging techniques might be utilized to assess the degree to which rhythmic beats are capable of tuning two individuals’ neural motor circuits.

In summary, our study sheds new light on the potential mechanisms underlying the effects of music on social behavior. We have proposed that our findings indicate that collective listening to music impacts on subsequent interpersonal behavior through its enhancement of motor coupling between individuals. However, given that we did not observe any main effects in our measurements, more empirical data are needed to test this claim. Furthermore, our findings also suggest that listening to rhythm does not uniformly aid interpersonal interaction; such motor coupling may be detrimental to interactions that require dynamic, coordinated motor responses. Our study therefore provides preliminary empirical evidence for a more nuanced interpretation for the impact of rhythm on interpersonal behavior and its potential functions in human ceremonies, rituals, and social gatherings.

- 82 -

Appendix CRQA computation

Before submitting the ActiGraph data to CRQA, preprocessing was performed in MATLAB (MathWorks Inc., 2013). In an initial step, a single acceleration vector was calculated for each hand by collapsing across the three spatial dimensions, and then zero centering and rectifying the collapsed time series. In a second step, each acceleration vector was z-scored. This ensured that the CRQA results were based truly on the sequence of accelerations in time, and did not result simply from greater differences or commonalities between participants’ hand movement acceleration amplitudes (Shockley et al., 2002).

To conduct CRQA, the first step is to reconstruct the phase space by selecting an appropriate embedding dimension and a time delay based on Taken’s theorem (Takens, 1981; Zbilut, Zaldivar-Comenges, & Strozzi, 2002). We applied the function of average mutual information to select a suitable time delay which yielded delays from 2 to 10. A delay of five sampling points was chosen because it corresponded to a minimal time difference between movements (i.e. 6 Hz, see Movement computation). Next, we applied the function of false nearest neighbors (Kennel, Brown, & Abarbanel, 1992) and estimated embedding dimensions that ranged from 3 to 8. As over-embedding yields more reliable CRQA results than under- embedding (Marwan et al., 2002; Webber & Zbilut, 2005), we chose an embedding dimension of 6 as a compromise between the average and upper estimates of the dataset. A radius was set so as to have an average recurrence rate (%RR – ratio of points in the phase space counted as recurrent) around 3% to ensure that all subjects have nonzero %RR (Shockley, 2005).

Movement computation

To extract hand movements from the raw ActiGraph signal, we first recorded the smallest and fastest movements which served as a definition of minimal movement. This yielded minimal movement acceleration of 0.05 G and minimal distance between movements of 166 ms (movement frequency of 6 Hz). Next, we used the collapsed and rectified time series and excluded all movements smaller than 0.05 G and closer than 166 ms (for an example see Fig. 6A). Such defined time series were used to compute mean number of movements and mean acceleration.

- 83 -

EFFECTS OF RITUAL BEHAVIOR ON ANXIETY

After discussing the group-level effects of religion and ritual on prosociality, I will now proceed to review individual-level effects of ritual behavior. In line with the previous chapter I select a classical theory in religious studies, but this time on the relationship between religion and anxiety as proposed by Bronislaw Malinowski (1948/1992). Malinowski was an anthropologist who, contrary to Durkheim, emphasized field research of studied phenomena and based his theories on ethnographic observations made on the Trobriand Islands. This first- hand experience led Malinowski to criticize Durkheim for omitting the importance of personal religious inspiration (1948/1992, pp. 56-57) and to accentuate individual psychological processes operating during magical rituals. While magic was distinctively different from religion in Durkheim’s view, Malinowski argued that magical behavior is often present in the overall religious frameworks of different societies, albeit serving different functions than religion (see also Talmont-Kaminski, 2013, pp. 117-118).30

When discussing the potential functions of magical acts,31 Malinowski first noted that they are not a product of “hopelessly mystical frame of mind” as Lévy-Bruhl would have it (p. 25). Instead, magical rituals are complementary to people’s rational and practical actions. For example, Trobrianders have a suite of practical knowledge about the types of soil, cultivated plants, and the conditions that lead to an abundant harvest. Likewise, when building a canoe, they manifest knowledge and skills that surpass those of an average modern person. Yet rituals were always performed on the top of these practical actions, and that led Malinowski to suggest that they might increase the feeling of control over future uncertain outcomes of one’s labor (pp. 27-30). Rituals did not replace functional arrangements, but served to influence factors beyond people’s physical possibilities. Thus, Malinowski observed the performance of magical rituals only when people faced dangerous and uncontrollable events that involved a lot of luck and uncertainty. For instance, Trobrianders would perform magical ritual before fishing on

30 Whereas religious ceremonies have usually no practical goals other than fostering the community (e.g., celebration of a birth), magical acts are directed toward specific ends (e.g., prevent a child’s death; pp. 37 – 38). 31 The functionalist approach was widely criticized in anthropology mostly for its disregard of the development of institutions and their current functions, together with the assumption that society is in a state of functional equilibrium (Goldschmidt, 1966). However, Malinowski’s insights can be still converted into experimental designs without necessarily postulating a specific function of ritual behavior (Sosis & Handwerker, 2011; Whitehouse, 2011).

- 84 -

open sea or to secure their harvest before natural disasters (pp. 139-140). Malinowski summarized these observations as follows:

“Thus in his relation to nature and destiny, whether he tries to exploit the first or to dodge the second, primitive man recognizes both the natural and the supernatural forces and agencies, and he tries to use them both for his benefit. Whenever he has been taught by experience that effort guided by knowledge is of some avail, he never spares the one or ignores the other. He knows that a plant cannot grow by magic alone, or a canoe sail or float without being properly constructed and managed, or a fight be won without skill and daring. He never relies on magic alone, while, on the contrary, he sometimes dispenses with it completely, as in fire-making and in a number of crafts and pursuits. But he clings to it, whenever he has to recognize the impotence of his knowledge and of his rational technique.” (1948/1992, p. 32).32

One of the main functions of magical rituals, according to Malinowski, is to regain emotional balance and harmony with life (p. 79), affording to confidently pursue one’s goals without being overwhelmed by anxiety and despair (p. 140). Put simply, Malinowski hypothesized that magical acts serve to alleviate anxiety caused by uncontrollability of future threats.33

This hypothesis was further elaborated by several scholars studying ritual practices in uncertain situations such as wars (Keinan, 1994; Sosis, 2007) or sports (Felson & Gmelch, 1979). For example, Sosis & Handwerker (2011) surveyed Israeli women during the 2006 war with Lebanon, questioning them on perceived anxiety and the frequency of psalm recitation. They found that women who left the bombarded area were no less anxious than the ones who stayed, but only the second group enjoyed the beneficial effects of psalm recitation on anxiety reduction. While women who left Tzfat (the bombarded town) had to deal with mundane stressors, the ones who stayed faced uncontrollable and uncertain stressors, such as being hit by an artillery shell. In line with Malinowski’s prediction, people used pragmatic acts in order to control future threats whenever possible, but when they lacked means of controlling the future happenings, reciting psalms helped them regain the feeling of control and soothe anxiety.

However, it is unclear why magical rituals should have these effects and what the underlying mechanism are. Several researchers noted that both cultural rituals and the rituals of patients with anxiety disorders share common features: they are compulsive, redundant,

32 Malinowski’s conclusion was reinforced by Evans-Pritchard who observed that the Azande have clear ideas how the surrounding world functions mechanistically, but would use magical practices to influences hidden causes (Evans-Pritchard, 1937/1976). 33 For an alternative explanation see for instance Talmont-Kaminski (2013).

- 85 -

repetitive, and rigid (Dulaney & Fiske, 1994; Liénard & Boyer, 2006; Rappaport, 1999). Although not explicitly stated by Malinowski, these features resonate with his observations. For example, compulsion is tightly connected to anxiety – people often feel a need to perform a particular ritual, or else something bad might happen. Malinowski acknowledges this when writing that:

“Man, engaged in a series of practical activities, comes to a gap; the hunter is disappointed by his quarry, the sailor misses propitious winds, the canoe builder has to deal with some material of which he is never certain that it will stand the strain, or the healthy person suddenly feels his strength failing. What does man do naturally under such conditions, setting aside all magic, belief and ritual? Forsaken by his knowledge, baffled by his past experience and by his technical skill, he realizes his impotence. Yet his desire grips him only the more strongly; his anxiety, his fears and hopes, induce a tension in his organism which drives him to some sort of activity. Whether he be savage or civilized, whether in possession of magic or entirely ignorant of its existence, passive inaction, the only thing dictated by reason, is the last thing in which he can acquiesce. His nervous system and his whole organism drive him to some substitute activity” (1948/1992, p. 79).

Next, redundancy describes perhaps the most striking aspect of cultural rituals: their non- functionality. Whereas Malinowski recognized that magical rituals have specific goals (e.g., to heal someone), it is usually not clear how the actions taken might lead to that goal (e.g., put your hand on your forehead three times and recite a mantra). Furthermore, rituals often include repetitive sequences that specify how many times the same action needs to be taken. Again, citing an example from Malinowski’ observations: “When a small object has to be charmed, a leaf is folded so as to form a tub and at the narrow end of this the object is placed, while the magician chants into the broad end...He would chant his charm for about half an hour, or even longer, repeating the spell over and over again, repeating various phrases in it and various important words in a phrase” (1948/1992, p. 192). The final feature, rigidity, has to do with the invariance of rituals and the emphasis on precise performance. The same rituals cannot be performed differently each time – quite the opposite, their invariance gives them an aura of eternity (Malinowski 1948/1992, p. 141).

But why should these ritual features affect anxiety? For example, Boyer and Liénard (2006) suggest that since the ritual acts are non-functional and thus not intuitive, practitioners need to pay attention to their performance to perform them correctly. As a consequence of this

- 86 -

focus, a practitioner’s mind is taken off the stressor by “swamping” working memory. In other words, a continuous stream of information being stored in short-term memory is focused on rituals and not on the threat. This might, in turn, result in a temporal decrease of anxiety, and repeating the sequence of non-functional behaviors may further amplify this effect.

In summary, this research line suggests that when people face uncontrollable and dangerous situations, they would use any means possible (practical and magical) to safe-guard themselves against potential threats. This proposition can be further broken down into two predictions: 1) when anxious, people should perform rituals; and 2) this performance should reduce their anxiety. In other words, anxiety should provoke spontaneous ritualization manifested as redundant, repetitive, and rigid behaviors, which should in turn alleviate the experienced anxiety. The first presented study (Lang, Krátký, et al., 2015) explores the hypothesis about spontaneous responses to anxiety. Together with a team of co-authors, we conducted an experimental study where we stressed people and used motion capture technology to measure the amount of ritualization manifested in their hand movement. The results showed that increased anxiety led to an increase in hand-movement ritualization, thereby providing initial support to the first part of Malinowski’s hypothesis. In the second article that was published with the same co-authors (Krátký, Lang, et al., 2016), we discuss possible cognitive mechanisms mediating these effects and suggest ways to test them in future experiments. Finally, in the discussion chapter, I present preliminary results of a follow-up study that explored the latter part of Malinowski’s hypothesis; that is, whether ritual behavior reduces anxiety.

- 87 -

Effects of Anxiety on Spontaneous Ritualized Behavior Lang, M., Krátký, J., Shaver, J.H., Jerotijević, D., & Xygalatas, D.

Summary Environmental uncertainty and uncontrollability cause psycho-physiological distress to organisms (Foa, Zinbarg, & Rothbaum, 1992; Lazarus & Folkman, 1984; Ursin & Eriksen, 2010), often impeding normal functioning (Barrera & Norton, 2009; Olatunji, Cisler, & Tolin, 2007). A common response involves ritualization, that is, the limitation of behavioral expressions to predictable stereotypic and repetitive motor patterns (Eilam, Zor, Szechtman, & Hermesh, 2006; Langen, Durston, Kas, van Engeland, & Staal, 2011; Mason, Clubb, Latham, & Vickery, 2007). In humans, such behaviors are also symptomatic of psychopathologies like obsessive-compulsive disorder (OCD; Boyer & Liénard, 2006; Eilam et al., 2006) and autism spectrum disorders (ASDs; Leekam, Prior, & Uljarevic, 2011; Rodgers, Riby, Janes, Connolly, & McConachie, 2012). Although these reactions might be mediated by different neural pathways, they serve to regain a sense of control over an uncertain situation (Eilam, Izhar, & Mort, 2011; Gillott, Furniss, & Walter, 2001; Moulding & Kyrios, 2006; Reuven-Magril, Dar, & Liberman, 2008) by engaging in behavioral patterns characterized by redundancy (superfluous actions that exceed the functional requirements of a goal), repetitiveness (recurrent behaviors or utterances), and rigidity (emphasis on fidelity and invariance; Boyer & Liénard, 2006; Eilam et al., 2006; Liénard & Boyer, 2006; Rappaport, 1999). We examined whether ritualized behavior will manifest spontaneously as a dominant behavioral strategy in anxiogenic situations. Manipulating anxiety, we used motion-capture technology to quantify various characteristics of hand movements. We found that induced anxiety led to an increase in repetitiveness and rigidity, but not redundancy. However, examination of both psychological and physiological pathways revealed that repetitiveness and rigidity were predicted by an increase in heart rate, while self-perceived anxiety was a marginally significant predictor of redundancy. We suggest that these findings are in accordance with an entropy model of uncertainty (Hirsh, Mar, & Peterson, 2012), in which anxiety motivates organisms to return to familiar low-entropy states in order to regain a sense of control. Our results might inform a better understanding of ritual behavior and psychiatric disorders whose symptoms include over-ritualization.

- 88 -

Results Given that anxiety-related ritualization is manifested in diverse contexts, such as precautionary behavior; Boyer & Liénard, 2006; Eilam et al., 2011), human social behavior (Malinowski, 1948/1992), and specific pathologies (Eilam et al., 2006), we investigated their potential common denominators. In contrast to previous studies that examined learned, habitual, and culturally specific behavioral scripts (Keinan, 2002; Sosis & Handwerker, 2011), our study focused on gestural patterns displayed spontaneously as a distress reaction. Sixty-two undergraduate students of Masaryk University were randomly assigned to either a high-anxiety (HA; n = 31) or a low-anxiety (LA; n = 31) group. To induce anxiety in the HA condition, we used a public-speaking task, where participants were asked to prepare a speech about a decorative object (see Figure S4) and later present it in front of a panel of experts (Feldman, Cohen, Hamrick, & Lepore, 2004). After the manipulation, participants had to clean the object. We measured the time spent cleaning and hand-movement characteristics (obtained as acceleration patterns) using the GT3X ActiGraph motion sensor (D. John & Freedson, 2012) placed on participants’ wrists. We focused on cleaning because it is one of the most commonly ritualized actions both in psychopathological and ceremonial contexts (Douglas, 1970/2004; Zor et al., 2009). We operationalized ritualized behavior across three attributes: (1) redundancy (time spent cleaning the object and number of movements used), (2) repetitiveness (recurrence of hand-movement signal), and (3) rigidity (predictability of hand movements and their SD). To compute repetitiveness and rigidity, we used recurrence quantification analysis (RQA; Shockley et al., 2002) and compared each hand-movement acceleration signal against its delayed versions (Marwan et al., 2007). Whenever two signals shared the same acceleration pattern (they fell within a preselected radius; see Figure S5), a recurrence point was recorded. Hence, the percentage of recurrent points (%RR) can be understood as an indicator of repetitiveness (the percentage of movement patterns repeating over time). Furthermore, if recurrent points follow each other in time (movement patterns evolve in the same way), their predictability is high and they are said to be deterministic. Thus, the percentage of recurrent points exhibiting determinism (%DET) was used as an indicator of rigidity in movement trajectories.

We hypothesized that induced anxiety would increase gestural ritualization. Specifically, we predicted that HA participants would (1) spend more time cleaning the object and deploy more movements during cleaning, (2) display a higher percentage of recurrent movements, and (3) express more deterministic movement trajectories and yield a lower SD of

- 89 -

movement acceleration. To further explore the effects of induced anxiety, we built two additional models for every measure, looking at the effects of self-perceived anxiety during the preparation task and Z-scored heart-rate differences between the baseline and preparation periods.

Manipulation Check

Analysis of self-reported anxiety during speech preparation revealed a significant difference between conditions [t (60) = 3.431, p = 0.001], confirming that anxiety was higher for HA participants (mean = 1.986, SE = 0.184) relative to LA participants (M = 1.079, SE = 0.264; see Figure 8A). Subsequently, we analyzed changes in mean Z-scored heart rate between baseline and the preparation task. A linear mixed model (with periods nested in individuals) revealed a significant interaction between condition and time [t (45) = 6.087, p < 0.001]. Post hoc pairwise comparison with Tukey correction showed a significant increase in heart rate for the HA condition [t (50.2) = 7.839, p < 0.001], but not for the LA condition [t (50.2) = 0.654, p = 0.516]. Further analysis revealed a significant positive correlation between self-reported anxiety and heart rate (r = 0.322, p = 0.027).

Redundancy

To assess redundancy, we measured time of cleaning and number of hand movements during cleaning. We did not observe a main effect of condition on time spent cleaning (HA: mean = 69.710, SE = 8.592; LA: mean = 66.161, SE = 6.076; see Figure 8B) or number of movements (HA: mean = 188.387, SE = 20.137; LA: mean = 187.419, SE = 16.692; see Figure 8C). Likewise, we did not find heart-rate increase to be a significant predictor of redundancy measures. However, self-reported anxiety significantly predicted the number of movements and had a marginally significant positive effect on time spent cleaning (p = 0.080; see Table 8).

Repetitiveness

The repetitiveness of movements during the cleaning task was evaluated by computing the %RR of hand-movement acceleration. We found higher %RR in the HA condition (mean = 2.204, SE = 0.431) compared to the LA condition (mean = 1.084, SE = 0.155; see Figure 8D).

- 90 -

Since the %RR data were strongly heteroscedastic (showing higher variance in the HA condition), we modeled both mean and dispersion to account for this difference. We observed a significantly higher mean %RR and significantly higher dispersion of %RR in the HA condition. Furthermore, mean Z-scored heart-rate increase was significantly associated with higher mean and dispersion of %RR. We did not observe any effect of self-perceived anxiety on our measure of %RR (see Table 8).

Rigidity

To assess the rigidity of movements, we computed the percentage of recurrence points forming deterministic lines, which are indicative of a signal’s predictability. We found higher %DET in the HA condition (mean = 26.271, SE = 3.090) compared to the LA condition (mean = 16.141, SE = 2.058; see Figure 8E). Regression analyses revealed a significant difference between the HA and LA conditions and a significant positive relationship between %DET and heart-rate increase. No effects of self-perceived anxiety were observed (see Table 8). Since %DET did not display large variance between conditions, dispersion was modeled by the intercept. Contrary to our RQA results, the SD of movement acceleration was not significantly different between conditions (see Table 8; HA: mean = 0.697, SE = 0.020; LA: mean = 0.693, SE = 0.020; Figure 8F) and was not predicted by self-perceived anxiety or heart-rate increase. This suggests that the predictability of movement acceleration trajectories might be underlined by more-complex patterns unfolding over time. That is, HA participants structured their movements into sequences of predictable clusters, expressing rigidity and invariance of movement patterns inside these clusters (Figure 9; see also Figure S5).

- 91 -

Table 8. Estimated Means and SEs of Measures of Ritualized Behavior

Redundancy Repetitiveness Rigidity

No. %RR: %RR: Time (s) %DET SD Movements Mean Dispersion

Model 1 73.258 135.047 2.233 0.177 25.068 0.697 Intercept (16.639)*** (8.469)*** (0.333)*** (0.018)*** (2.128)*** (0.020)*** Condition -3.548 -1.779 -1.146 -0.080 -7.818 -0.004 (LA versus (10.523) (11.899) (0.303)*** (0.019)*** (2.664)** (0.029) HA)

Model 2 67.935 118.773 1.762 0.154 20.467 0.701 Intercept (5.133)*** (9.325)*** (0.309)*** (0.019)*** (2.308)*** (0.025)*** Reported 8.390 10.270 -0.066 -0.007 0.356 -0.004 anxiety (4.717)Ϯ (5.174)* (0.157) (0.010) (1.192) (0.013) (0–4)

Model 3 69.745 140.669 1.430 0.121 19.748 0.693 Intercept (5.588)*** (7.505)*** (0.163)*** (0.011)*** (1.739)*** (0.018)*** Heart rate -0.368 2.697 0.664 0.040 4.883 0.020 (z- scores) (6.012) (7.920) (0.200)** (0.013)** (1.64)** (0.017)

Models 1–3 describe predictors used to predict our outcome variables. In model 1, HA condition is a reference category; in model 2, self-perceived anxiety = 0 is a reference category; and in model 3, mean heart rate increase is a reference category. Ϯp < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

- 92 -

Figure 8. Mean Values with ±SEM from Measures of Ritualized Behavior for the High- and Low- Anxiety Conditions. (A) Significant difference in perceived anxiety during speech preparation. (B) Non-significant difference in cleaning times. (C) Non-significant difference for number of movements during cleaning. (D) Participants in the HA condition displayed significantly higher recurrence rate of movement acceleration. (E) Participants in the HA condition displayed significantly higher percent of deterministic movement-acceleration trajectories. (F) Non-significant difference in the SD of movement acceleration.

- 93 -

Figure 9. Recurrence Plots of Hand-Movement Acceleration. Illustrative recurrence plots built on the basis of dominant hand-movement acceleration of a participant in the HA condition (A) and a participant in the LA condition (B). Visible clusters in (A) are indicative of recurring movement patterns. Recurrence points are sparse and appear to be more evenly distributed in (B), suggesting more variable hand- movement acceleration trajectories.

- 94 -

Discussion Our results provide three-fold confirmation of the hypothesis that anxiety triggers ritualized behavior. We found that anxiety led to an increase of (1) gestural redundancy, (2) repetitiveness of movement, and (3) determinism and predictability of movement. Although redundancy did not differ significantly between conditions and was not associated with heart-rate increase, perceived anxiety during the preparation task was a significant predictor of the number of movements during cleaning and a marginally significant predictor of time spent cleaning. In other words, participants who experienced more anxiety used more movements in the cleaning task, possibly as a coping strategy (Boyer & Liénard, 2006, 2008).

Our measure of recurrent hand-movement trajectories showed a significantly higher recurrence in the HA condition. Although the object was not symmetrical and different parts required different types of movements, participants in the HA condition kept applying similar movements. These behavioral expressions were also significantly predicted by the Z-scored heart-rate increase (see Table 8). Looking at the dispersion of those data, we found a significant effect of condition and heart-rate increase, which suggests that there is individual variance in the extent to which participants are attracted to repetitive movement patterns while in a state of anxiety. The participants in the HA condition who did not display repetitive patterns of hand movements probably did not experience a sufficient amount of system destabilization. The dispersion might also be caused by other intervening variables (for example, desire for control), which we did not assess.

Finally, the measurement of %DET indicated that participants in the HA condition displayed more-predictable movement patterns. Once they returned to a familiar hand- movement acceleration pattern, they tended to follow this pattern for a longer time. Interestingly, such pattern sequencing may also help explain why we found no significant effect of treatment on mean SD of movement acceleration. Since the invariance of movements was bound to specific temporal clusters, it might be unnoticeable after averaging SDs over the temporal dimension. Importantly, we also observed a significant positive effect of heart-rate increase on %DET but no such effect on SD of hand-movement acceleration.

In summary, our findings show that ritualization may be a spontaneous response to anxiety. While the treatment significantly affected self-perceived anxiety, some participants in the LA condition also perceived the preparation task as stressful, and this cognitive appraisal appeared to be essential for an increase in redundancy. On the other hand, an increase in heart rate was more closely associated with the treatment and was a significant predictor of

- 95 -

repetitiveness and rigidity. It is possible that a cognitive appraisal of anxiety affected only behaviors directly accessible to conscious control: duration of cleaning and number of movements used during cleaning (redundancy). The other two movement characteristics, repetitiveness and rigidity, are more dynamic and finer grained (hand-acceleration changes in a matter of milliseconds) and are presumably not accessible to conscious control; however, they were affected by physiological reactions to the stressor as indicated by heart-rate increase. The threat of public speaking appears to more directly induce physiological distress reactions compared to the overall psychological discomfort that participants experienced in both conditions. This interpretation is further supported by the significant, but rather moderate, positive correlation (r = 0.322) between self-reported anxiety and mean heart-rate change, which suggests that psychological and physiological processes can affect different, albeit concurrent, cognitive mechanisms (Hardy, 1996).

We propose that these findings are in accordance with the entropy model of uncertainty (Hirsh et al., 2012). When facing a complex, uncontrollable, and unpredictable situation, an organism’s cognitive-behavioral system experiences a high-entropy state. Entropy is understood here as a reduced ability to predict successive states based on the current state (Hirsh et al., 2012). The principle of entropy minimization holds that the goal of a cognitive- behavioral system is to minimize internal entropy and increase prediction success (Clark, 2013; Friston, 2009). Although a variety of experiential possibilities is crucial for an organism’s success, a tradeoff between high-entropy risk and gains obtained from new environmental situations needs to be balanced at a manageable level – otherwise, low predictive abilities (followed by anxiety) could significantly impede functioning. Translated to the context of the present study, the possibility of public speaking might have increased participants’ psychological entropy and decreased their feeling of control (Keinan, 2002). To cope with such instability, organisms tend to return to familiar low-entropy states (Hirsh et al., 2012), often by performing repetitive and predictable actions that minimize the conflict between behavioral and perceptual affordances. Importantly, such behavioral expressions may be functionally detached from the anxiogenic situation and instead focus on increasing interoceptive predictive success (Seth, 2013), which might in turn lead to regaining a sense of control. The focus on instant interoception could also explain why low-entropy behaviors are often detached from the threatening stimuli, which from an external perspective might seem purposeless (Boyer & Liénard, 2008; Eilam et al., 2011). In our experiment, we artificially restricted participants’ behavioral expressions to the predefined task, where the readily accessible low-entropy state

- 96 -

was tied to the act of cleaning. Without such a pre-defined task, subjects might turn to different types of ritualized behaviors, either spontaneous (e.g., marching up and down) or related to available cultural scripts (such as praying).

We hazard that the principle of entropy minimization can also be extrapolated to more- complex ritualized behaviors as those observed among gamblers and athletes and in religious rituals. Performance anxiety may result from low predictive possibilities caused by environmental complexity and uncertainty, and carrying out low-entropy stereotypical actions (rituals) may help regain a feeling of control over the situation (Evans, Gray, & Leckman, 1999; Keinan, 2002; Moulding & Kyrios, 2007; Norton & Gino, 2014; Reuven-Magril et al., 2008; Sosis & Handwerker, 2011). In turn, this regained sense of control might result in anxiety alleviation (Fam, Tan, & Waitt, 2012; Leekam et al., 2011) and, consequently, lead to better performance (Damisch, Stoberock, & Mussweiler, 2010). We note that our approach does not purport to explain all rituals. For example, recent research suggests that rituals may have a variety of positive consequences, related not only to alleviating aversive states (Norton & Gino, 2014) but also to producing positive reinforcement (Vohs, Wang, Gino, & Norton, 2013). However, the mechanisms described here might still provide additional motivations for individual participation in such rituals. Furthermore, collective rituals are social events, and as such they are complex phenomena that are not always connected to anxiogenic situations.

Our results might also be explained by the model of ritualized behavior suggested by Boyer and Liénard (Boyer & Liénard, 2006, 2008; Eilam, 2015; Keren, Boyer, Mort, & Eilam, 2013). Ritualized movements might have functioned to overload working memory, thus suppressing intrusive thoughts about the threat of public speaking (Wenzlaff & Wegner, 2000). Differences in redundancy predicted by self-perceived anxiety could be also interpreted as an attempt to delay subsequent unpleasant tasks. Alternatively, participants could have engaged in ritualization as a form of adjunctive behavior, resulting from a process of reducing corticosteroid levels as a response to a stressful situation (Mason, 1991). However, this matter can only be resolved by further empirical studies focusing on the cognitive processes involved in ritualization and controlling for the role of cognitive load. Future research may, for example, investigate how environmental affordances impact the level and dynamics of ritualization. In addition, allowing participants to move more freely or choose actions by themselves may reveal yet unknown links between personality types and ritualization. Likewise, a better understanding of the role of personality (for example, desire for control) may reveal mediating variables between anxiety and ritualized behavior. Most importantly, the current findings must

- 97 -

be supplemented by the investigation of the effects of ritualized behavior on assuaging anxiety. Overall, our study offers a novel insight into the link between anxiety and ritualized behavior. Although this link had long been theorized, our methodology allowed us to operationalize and evaluate these theories in a quantitative way. Our findings may shed light on the persistence of ritual behavior throughout the animal kingdom and can potentially help us gain a better understanding of psychiatric disorders like obsessive-compulsive disorder or autism spectrum disorders, whose symptoms include over-ritualization.

Experimental Procedures We recruited 32 female and 30 male Masaryk University students (mean age = 23.851, SD = 1.839), who received course credit for participation. The study was approved by the ethical committee of the Faculty of Arts, Masaryk University, and informed consent was obtained from all subjects. Participants were randomly assigned to either an HA (n = 31) or LA condition (n = 31) defined by the presence/absence of an anxiety-inducing task. Prior to the experiment, we told participants that we would be collecting physiological measurements during the experiment and fitted them with the equipment. Each participant wore a heart-rate monitor around their chest and one accelerometer on each wrist. Baseline heart rates were obtained at a resting state. Subsequently, participants were seated at a table with a decorative object placed in front of them (a round shiny metal object on a ceramic stand; 250 X 240 X 80 mm; Figure S4). In the HA condition, we used a modified version of the public-speaking task (Feldman et al., 2004) to induce anxiety: participants were given 3 min to prepare a 5-min-long speech about the decorative object to be delivered in front of an art expert. They were also provided with a set of seven questions about the object they were required to answer during the speech (see the Supplemental Information). We informed HA participants that the experts were waiting in an adjacent room and would be rating their performance. Participants in the LA condition were instructed to think about the same object for 3 min and to try find answers to the same seven questions, but public speaking was not mentioned. During the preparation period, participants in both conditions were not allowed to touch the decorative object, which was clean. Before participants were to present the speech (HA) or end the task and leave (LA), they were asked to hold the object with both hands, facing its horizontal plane, and clean it with a wet cloth until they considered it to be clean. We did not provide a particular reason for cleaning so as to make the cleaning task appear redundant; however, since the object had a metallic shine, it could always be more clean or polished. Once participants decided that the

- 98 -

object was clean, those in the HA condition were told that they would not have to make their presentation due to a momentary absence of the art expert. All participants subsequently filled out a final questionnaire concerning their feelings during the manipulation period (five items). All participants were debriefed after the end of data collection.

Data Analysis

Heart-rate data were Z-scored to control for natural differences between participants. Due to a malfunctioning device, we lost heart-rate data from 15 participants; however, the lost data were almost equally distributed across conditions (eight and seven). Hand-movement acceleration data collected by ActiGraph motion sensors were preprocessed and filtered to extract individual movement characteristics. Recurrence and determinism of hand-movement acceleration were computed with MATLAB CRP Toolbox 5.17 (Marwan, Wessel, Meyerfeldt, Schirdewan, & Kurths, 2002), using RQA (Shockley et al., 2002).

The relationships between independent and dependent variables were analyzed in R version 3.0.3 (R Core Team, 2014). We fitted two models with a normal distribution for the manipulation check (heart-rate increase and self-reported anxiety) and three linear models for each of the five measures: models with a normal distribution for time of cleaning, mixed models with a normal distribution for SD of movement acceleration, mixed models with a negative binomial distribution for number of movements to account for the distribution of count data, and mixed models with a beta distribution for percentage of recurrence and determinism to account for the distribution of proportions with lower and upper bounds of 0 and 1 (Smithson & Verkuilen, 2006).

- 99 -

Anxiety and ritualization: Can attention discriminate compulsion from routine? Krátký, J., Lang, M., Shaver, J.H., Jerotijević, D., & Xygalatas, D.

Humans and other animals engage in ritual behaviors, yet their evolutionary functions are unknown. Marked by limitation of behavioral variability, greater repetitiveness, and stereotypy (Izhar & Eilam, 2010; Serruya & Eilam, 1996), ritual expressions manifest on multiple levels of behavior (Keren, Boyer, Mort, & Eilam, 2010). Ritualization occurs in gestures, but also in complex behaviors, and is typically characterized by the presence of redundant or unnecessary steps in behavioral patterns that are not functionally related to a pragmatic goal (Nielbo & Sørensen, 2011; Zacks, Tversky, & Iyer, 2001).

It has been suggested that rituals can function as a type of coping strategy, an automatic response to novelty, unpredictability and uncontrollability; in other words, to environmental features that cause anxiety and psycho-somatic stress (Conrad, 2011; Foa et al., 1992). In addition, human rituals can be understood as culturally evolved behavioral responses to ecological or social threats (Malinowski, 1948/1992; Sosis & Handwerker, 2011). However, in its excessive form, ritual behavior is symptomatic of certain human pathologies, such as obsessive-compulsive disorder and autism spectrum disorder (Zor et al., 2009).

Despite the fact that rituals are found in a wide range of domains, explanations for their pervasiveness have been inconclusive. Recently, two partially contradictory explanations of ritual behavior as a response to anxiogenic situations have been suggested. First, Boyer & Liénard (2006) presented a model of ritualized behavior (RB) that describes ritualization as a scripted sequence of redundant, goal-demoted behaviors. The precise execution of such a sequence demands diligent focus on the task and results in higher cognitive effort, precluding the practitioner from conscious preoccupation with the stressor. However, there is evidence that ritual behaviors are often performed in an automated way (automated behavior – AB; Eilam, 2006; Serruya & Eilam, 1996) that requires little cognitive effort. By simplifying action via repetition, stereotypy, and routinization, individuals can allocate more cognitive resources to threatening external stimuli (Fentress, 1976), thus increasing their chances of survival. Since both RB and AB manifest as behavioral stereotypy and repetition, they might be indistinguishable from each other by mere observation. In what follows, we propose a way to

- 100 -

distinguish between these competing models in order to gain a better understanding of the cognitive mechanisms that underlie behavioral ritualization.

In a recent study, we documented a link between anxiety and spontaneous gestural motor ritualization (Lang, Krátký, et al., 2015). Study subjects underwent a treatment based on the public speech paradigm (Feldman et al., 2004) resulting in two levels of stress (high anxiety – HA; low anxiety – LA). Subsequently, each individual was asked to perform a motor task consisting of cleaning the object with their hands during which levels of spontaneous gestural ritualization were measured. We hypothesized that differences in stress levels would manifest in gestural dynamics measured by GT3X ActiGraph motion sensors. Specifically, we predicted that HA subjects would display a higher level of motor ritualization. The results showed that subjects exhibited a shift toward higher redundancy, repetitiveness and rigidity of hand movements in the HA compared to the LA condition. We considered this increased ritualization as the manifestation of a psycho-physiological state of anxiety, and interpreted our findings in the light of an entropy model of uncertainty (Hirsh et al., 2012). According to this model, anxiety acts as a destabilizing factor increasing overall systemic entropy, and associated spontaneous ritualization acts as a coping strategy that decreases overall perceived entropy and prediction error (Clark, 2013; Friston, 2009).

However, it is not clear whether such ritualization stems from RB or AB, because both manifest as high levels of repetitiveness and rigidity. In other words, did HA participants’ behavior become more ritualized because they focused on the cleaning, or on the threat? By investigating participants’ hand-movement trajectories more thoroughly, we can get a better insight into their locus of attention. In the case of AB, a focus on external anxiogenic stimuli has been shown to have adverse effects on attention and processing efficiency (Eysenck, Derakshan, Santos, & Calvo, 2007; Wenzlaff & Wegner, 2000) that may manifest as limited motor control. Such a limitation of the degree of motor control should affect overall task execution (Englert & Oudejans, 2014; Kass, Cole, & Stanny, 2007; Nieuwenhuys & Oudejans, 2012) and possibly lead to less detailed and/or less complex behavioral patterns. On the other hand, RB should manifest as elaborate patterns that require conscious attention and longer execution, thereby distracting participants from the stressor.

A degree of structural organization of hand-movement trajectories in 3-dimensional space should discriminate between those with the locus of attention directed at the task (RB) and those with the locus of attention directed at the stressor (AB). Given that RB and AB may in principle involve typologically and quantitatively similar movements, we would need to

- 101 -

investigate at least three movement characteristics to be able to distinguish between the two. First, it is important to examine the hierarchy of movements, that is, the degree of structuration of behaviors in a given space. RB should exhibit a higher degree of movement nesting (a ratio between longer and shorter moves) with a preference toward shorter (nested) movements within the space demarcated by longer movements. If participants focus on executed motor task, their motor patterns will display inner structure and hierarchy as revealed by movement nesting. Second, RB should exhibit a higher recurrence of movement patterns across space, thus it is important to compare the similarity of trajectories in sub-regions of the cleaning space. If participants focus on cleaning, their nested movement patterns will be similar across the surface of an object. Finally, the proportion of surface cleaned by participants is critical for distinguishing between the RB and AB models. If subjects visit the whole space in a structured manner, this would suggest a consciously followed overall task plan executed over the surface of the entire object.

To test such hypotheses, we will need to map movement trajectories that go beyond the one-dimensional measures of acceleration used in our previous study (Lang, Krátký, et al., 2015). For instance, videotaping participants’ movements or more complex measures like 3D motion trackers could provide a richer picture of movement trajectories, thus allowing a more fine-grained distinction between RB and AB. Further research should also examine whether a higher degree of movement organization in RB leads to a higher degree of predictive success compared to AB, and in what ways this interacts with resource depletion related to the focus of attention.

To conclude, examining the above hypotheses will help us gain a better understanding of human ritualization and its distinction from routinization. Such a distinction might prove important for investigating the ways in which rituals are associated with anxiety across various contexts.

- 102 -

DISCUSSION

In the two previous chapters, I attempted to illustrate how religious phenomena as observed and interpreted by humanists/social scientists can be disassembled into particular sub- phenomena and further into mechanisms mediating these phenomena. By showing that religious people behave morally upon hearing sacred music, I supported Durkheim’s assertion that symbols embody society’s rules of moral conduct. The fact that learned associations between symbols and norms evoke certain behavioral schemas suggests that symbols may be powerful tools keeping society together. Furthermore, demonstrating that people are susceptive to rhythms and cannot help themselves but to synchronize with these rhythms, thereby increasing mutual liking, supports Durkheim’s idea of collective effervescence reached through unified motion. Indeed, verbalizing and moving together creates the feeling of group unity and entitativity (Reddish et al., 2013). Next, looking at the individual level, I presented original research inspired by Malinowski’s insights into the relationship between magical rituals and anxiety. By demonstrating that anxious people spontaneously engage in ritual-like behavior, I lent empirical support to a mechanism underlying Malinowski’s observations that Trobrianders perform magical rituals only before uncontrollable and dangerous events, which are highly anxiogenic.

These initial findings can be extended in multiple ways, whereby the presented theories will be further tested, refined, and furnished with a more complex understanding. For example, we are preparing a manuscript documenting mechanistic operations responsible for the positive effects of synchrony on pro-group behavior (Bahna, Lang, Reddish, Shaver, & Xygalatas, forthcoming). Inspired by the findings from our first experiment on rhythm (Lang, Shaw, et al., 2015), we assumed that rhythm drives motor synchrony through the entrainment of people’s motor cortices. Due to this entrainment, people easily couple their movements what may give them a feeling of a successful joint action, which can, in turn, lead to increased trust and higher cooperation in subsequent tasks. In this follow-up experiment, we found that the actual degree of motor coupling as measured by hand-movement acceleration, that is, how much two participants synchronized their movements to the beat, predicted inter-personal liking and cooperation in an economic game. In addition, we included a measure of endorphin release and found that endorphin levels were significantly higher for the synchronized participants, and that this increase mediated their higher cooperation in the economic game. Together with the

- 103 -

first experiment, these results suggest that rhythmic beat synchronizes participants’ motor cortices, thereby helping synchronize participants’ actual movements. Perception of this synchronization further affects satisfaction with the joint action, as indicated by the increased endorphin levels, and leads to higher willingness to cooperate in the future.

As an extension of my anxiety research, I am currently engaged in several laboratory and field projects exploring the second part of Malinowski’s hypothesis, which predicts that rituals will reduce anxiety. For instance, together with Jan Krátký and Dimitris Xygalatas, we plan to recruit Hindu women in Mauritius and ask them to prepare a speech about precautious behaviors during floods that should be later recorded on a microphone and evaluated by a government expert. After the speech preparation period, half of the women based in a local Hindu temple will perform their usual prayers, while the other half based in an empty apartment will sit in silence. After the prayer/silence period, we will tell them that they were not chosen to give the speech and then measure their anxiety, hypothesizing that the women who prayed will exhibit lower stress levels than the women who sat in silence. Preliminary data collected in a pilot study suggest that women who prayed indeed reported feeling less anxious after prayer compared to women sitting in silence. Interestingly, the praying women also reported lower anxiety compared to their baseline anxiety measured before the experiment, suggesting that prayer may help not only with acute, induced anxiety, but also with more long-term stress. In the full experiment, we plan to use physiological measurements in order to support this initial observation by elucidating physiological mechanisms mediating this effect.

Together, these extensions show how experimental studies done both in the laboratory and in the field can test some aspects of humanistic theories. Having said that, I am aware of the grandiosity of the proposal put forward in this thesis. My aim was not to argue for causal completeness or total unity of sciences, however (Dupré, 1996). Indeed, that would be a naïve consilience (Slingerland & Bulbulia, 2011). I rather aimed to illustrate how scientists and humanists may find a common ground in studying religion by bridging gaps between disciplines through mechanistic analyses as proposed by CSR. Arguably, this endeavor cannot lead to total consilience; be that due to insufficient methodologies, or ontological incompatibilities between different disciplines. For example, Durkheim’s thesis on the religious aspect of sociality enhancement was chosen as an illustration of humanistic theories that embrace the scientific method, whereby they afford translations into the lower-level sciences. But this theory has also a number of ontological claims that cannot be scientifically verified and do not lend themselves to mechanistic decompositions. I criticized Durkheim’s

- 104 -

terms, such as social facts or collective consciousness, which might be good metaphors, but poor analytical concepts. Yet Durkheim seemed to assign these immaterial concepts critical, sui generis importance in his theory (Gisbert, 1959; Hagens, 2006), thus precluding modern scholars from fully testing some of his hypotheses. So the compatibility of humanistic and scientific theories is to a large extent determined by willingness to moderate ontological assumptions on both sides.

Another problematic aspect of the consilient approach is a contemporary vagueness of proposed multi-level theories. Carefully investigating a phenomenon with its sub-phenomena and operating mechanisms is a monumental project that needs either extreme interdisciplinary cooperation with an enormous volume of pages, or extensive simplifications that might not satisfy individual specialists. This dissertation is a good example of the latter case, simplifying both Durkheim’s and Malinowski’s work and choosing only the aspects of their theories fitting to the multi-level approach to the study of religious phenomena. While I attempted to incorporate some later reflections of their work, these reflections had to necessarily be very limited. Furthermore, I only presented studies of particular mechanisms, but as emphasized in the mechanistic approach, such partial mechanisms do not constitute the whole phenomenon. Indeed, many more empirical studies would be needed to reach this goal, although I believe that even such miniscule empirical contribution as presented in this thesis is a valuable one. Therefore, I would like to invite scholars to further elaborate the consilient CSR approach with the goal of having the possibility to incorporate knowledge from other disciplines into one’s theories, finding inspiring methodologies, and offering a complex understanding of religious phenomena. In the final sub-chapter of this dissertation, I offer suggestions how the CSR multi- level proposition may be developed into an even more complex approach.

Future Directions

First, one might wonder where “religion” disappeared in the mechanistic multi-level study of religious phenomena. For example, our study of rhythm and motor coordination (Lang, Shaw, et al., 2015) neither mentions religion, nor provides a direct insight into religious phenomena. Nevertheless, it helps understand how the phenomenon of collective effervescence arises, and how this phenomenon operates over several different contexts, religion being one of them. The question remains, however, whether the religious context adds something that the other contexts do not, something that modifies the mechanism’s composition and operations?

- 105 -

While mechanisms and their parts can be studied in isolation (e.g., effects of auditory stimuli in a soundproof room), those mechanisms always work in a specific environment. That is, the workings of a mechanism established under particular conditions might not hold in other conditions (Bechtel, 2009). The specificity of the environment’s influences is an important part of mechanistic explanations (Bechtel & Hamilton, 2007). Any complex theory of religious phenomena should therefore move along two axes: vertical, describing a mechanism’s composition and organization; and horizontal, describing the specificity of environmental influences, moving from coarse-grain environmental features to specific spatio-historical contexts.

Returning back to the example of ritual behavior and anxiety alleviation, someone who performs a ritual in front of a god statue might direct attention both to the ritual performance and the deity. The purported moderating effect of working memory overload on anxiety (Boyer & Liénard, 2008; Krátký, Lang, et al., 2016) could be thus diminished by feelings of certainty and safety evoked by the presence of the god statue (Kirkpatrick, 2012). Or, ritualized behavior and the feeling of safety may interact with each other and amplify the inhibitory effect of ritual on anxiety. While the observed phenomenon of anxiety alleviation would not change, the underlying mechanisms could be different based on the environment. Additionally, if the statue would be dedicated to a merciless god demanding human sacrifices, ritual performed in front of such a statue might actually increase anxiety. The activation of mechanistic operations exhibits context biases that are crucial for an understanding of studied phenomena (Gervais, Willard, Norenzayan, & Henrich, 2011; Paden, 2013). This is not to say that we must resign from any attempts of identifying universal mechanisms. General patterns in contextualized behavior can be identified. Rather, this example illustrates that the evolved cognitive mechanisms studied by CSR are not context independent, and a thorough understanding of a specific context in which mechanisms work is a crucial layer affecting mechanisms’ emergent properties. Thus, contextualized mechanistic analysis can connect the universality of the sciences with the specificity of the humanities and explore what religious context adds to and how it modifies mechanistic operations.

- 106 -

In addition to context, humans have evolved in certain environments and physiological and behavioral mechanisms are thus adapted to those environments. That is to say, a third dimension should be added to the aforementioned horizontal and vertical dimensions. To paraphrase Bechtel (2009, p. 559): the mechanistic approach should look up and down, around, and back in time. An understanding of mechanistic operations should be thus posited in Euclidean rather than Cartesian space by moving along three axes of contextual diversity, mechanistic complexity, and evolutionary history. Rather than using vertical and horizontal axes, we need to talk about contextual width, mechanistic height, and diachronic depth (see Figure 10).

Figure 10. A three-dimensional representation of research strategies accentuating different levels of studied phenomena. The height axis represents phenomenal and mechanistic decomposition. The width axis concerns moving from explanations of general cultural patterns to explanations of specific cultural phenomena. The depth axis, additionally, regards evolutionary and developmental analyses with the assessment of biological fitness of observed phenomena.

- 107 -

While mechanisms may have different organizational properties in various environments (moving along the width axis), scholars can understand these differences by examining the current and past adaptive value of mechanisms in particular environments (the depth axis). Specifically, they can look at A) the evolution and ontogenetic development of mechanisms in different environments, thereby explicating how the currently observed mechanism came into being and what is its current impact on biological fitness; and B) the evolution of environments themselves in order to explain how evolutionary niche-construction interacted with a mechanism’s evolution and ontogenetic development, thereby affecting biological fitness. In other words, this dimension affords to study how environments (including culture) evolve and affect a mechanism’s evolution (A. W. Geertz, 2014).

First, the evolutionary history of a mechanism’s parts and their organization can be specified (moving on the height-depth plane), whereby another layer of corroboration would be added. Hypothesized relations between the mechanism’s components should conform to evolutionary logic, what in other words means that the components and their particular organization had to be selected for during the course of evolution.34 For example, when hypothesizing that ritual behavior positively affects social bonding, researchers would need to 1) demonstrate the existence of such a phenomenon (Xygalatas, Mitkidis, et al., 2013); 2) find lower-level phenomena mediating this effect, such as synchronous movement (Reddish et al., 2014); 3) decompose these phenomena into mechanisms like motor coupling (Hasson & Frith, 2016; Lang, Shaw, et al., 2015); 4) show their neural underpinning, for instance, the mirror- neuron system (Gallese, Gernsbacher, Heyes, Hickok, & Iacoboni, 2011; Rizzolatti, Fadiga, Fogassi, & Gallese, 1999); and, finally, 5) trace the evolutionary history of motor coupling and mirror neuron network while showing their adaptive value (Huber et al., 2009), including the investigation of similar mechanisms in phylogenetically related species (Hecht et al., 2013; Kaneko & Tomonaga, 2012).

Second, ecological niches (Laland, Kendal, & Brown, 2007)35 have a diachronic perspective as well (the depth-width plane). While mechanisms evolve during the course of millennia through genetic transmission and are realized de novo during one’s ontogeny, ecological niches are developing without the need to be re-created anew for every single human

34 I am aware that not all mechanisms will have a clear adaptive value that would explain their presence in the studied phenomena, as documented by the concept of evolutionary spandrel (Gould & Lewontin, 1979). However, scholars can still investigate a spandrel’s current function and the possibility of its exaptation (Gould & Vrba, 1982). 35 That is, environments surrounding people, which include also cultural phenomena and institutions.

- 108 -

being. That is, cultures are cumulative and exhibit a “ratchet effect” that locks cultural inventions firmly in place and turns the wheel of development forward (Tennie, Call, & Tomasello, 2009; Tomasello, 2009). Analyzing how the current and historical ecological niches were constructed (Laland et al., 2007; Laland, Matthews, & Feldman, 2016) in order to enhance biological fitness of its constructors (i.e., living organisms) adds another explanatory layer to the contextual analysis of religious phenomena. For example, Ara Norenzayan and his colleagues (Norenzayan, 2013; Norenzayan et al., 2016; Purzycki et al., 2016) argue that the belief in big moralizing gods evolved as a response to the growing complexity of human societies. That is, a new niche was evolving due to the increased size of human societies, and belief in moralizing gods became an important part of this niche because it provided adaptive value to people “playing” this evolutionary strategy (Maynard Smith, 1982). The belief in moralizing gods, it is argued, serves to curtail negative social phenomena, such as free-riding, cheating, and other behaviors that hamper cooperation, by threatening people with supernatural punishment. The history of monotheistic religions, for example, can be understood in evolutionary terms as a niche-constructing activity that gave their proponents adaptive advantage by securing cooperative exchange.36

Although the addition of an evolutionary layer to the contextual mechanistic approach might seem like a step back toward universalistic explanations, I believe that it is compatible with the multi-level approach. Rather than emphasizing universalism, the evolutionary level adds another perspective on studied phenomena that is complementary with other-level explanations. This situation is aptly captured by the so called proximate and ultimate levels of analysis used in evolutionary biology and ethology (Mayr, 1961; Tinbergen, 1963). In a seminal paper on cause and effect in biology, Ernst Mayr (1961) posed the following question: “Why did the warbler on my summer place in New Hampshire start his southward migration on the night of the 25th of August?” Mayr offered four possible causes of the warbler’s migration: 1) the warbler would not have sufficient food sources during winter; 2) migration is a genetically encoded trait of the warbler species; 3) the warbler detected a change in the length of daylight; 4) weather conditions were optimal for taking off that day. The reader can probably recognize that there are two sets of explanations: one dealing with evolved mechanisms (1 and 2) and one dealing with immediate causes (3 and 4). Mayr’s point is that all four causes can be valid at the same time, they just address different layers of analysis: ultimate, dealing with

36 However, see Baumard, Hyafil, Morris, & Boyer (2015) for a critique of this theory and Kundt (2015) for a critique of cultural evolutionary approaches.

- 109 -

evolutionary causes, and proximate, dealing with immediate intrinsic and extrinsic causes. The warbler’s behavior is multi-layered, thus multi-causal.

Whereas Niko Tinbergen (1963) also recognized that there are short-term and long- term levels of analysis, he distinguished among four analytical levels: 1) causation (how are traits activated); 2) survival value (what are the effects of traits on biological fitness); 3) ontogeny (how traits develop during lifetime); and 4) evolution (why traits evolved). While the first and third level concern the proximate causes, the second and fourth level address ultimate effects and causes respectively. Applying these four levels on the case of the warbler’s migration, it can be argued that migratory behavior evolved to secure the warblers’ survival;37 and that the association between daylight length and migration, together with the warbler’s sensitivity to light developed during ontogeny, are the immediate causes influencing the warbler’s decision in a particular year. Abstracting from Mayr’s and Tinbergen’s approaches, I believe that the mechanistic analysis should ask proximate and ultimate questions and try to integrate them into a complex intertwined theories of religious phenomena. Distinguishing between the proximate and ultimate levels can shed light on seemingly competing explanations and show that they are dedicated to different analytical layers. Consider, for instance, the Biblical parable of the Good Samaritan (Luke 10: 25-37, New International Version), in which the Good Samaritan helps an injured man on his way to Jericho. When looking for possible causes of the Samaritan’s behavior, one can make at least five arguments. First, the Samaritan felt sorry for the robbed victim and after weighing all the pros and cons of helping, the Samaritan decided that he has enough time before the evening, and thus can carry the victim to the closest inn (Darley & Bateson, 1973). Second, the Samaritan was raised in a family that emphasized helping behaviors, even toward members of other religious groups (House et al., 2013). Upon encountering the victim, the Samaritan felt obliged to help. Third, seeing a suffering human displaying distress signals triggered a cascade of neurophysiological processes in the Samaritan together with compulsion to help the victim (Hein, Silani, Preuschoff, Batson, & Singer, 2010; Singer et al., 2006). Fourth, helping other people in distress is an evolved mechanism securing parent-offspring bonding, which facilitates parents’ motivation to care for and raise their offspring, and spills over to other interpersonal relationships (Preston & de Waal, 2002). Thus, the Samaritan acted according to his genetic programing. Fifth, helping close but also distant kin displaying distress signals increases

37 Modern evolutionary theory uses the term reproductive success rather than survival value to emphasize the importance of successful genetic transmission (P. Bateson & Laland, 2013; Kaplan & Hill, 1985; Strassmann & Gillespie, 2003).

- 110 -

biological fitness over the long run, thereby facilitating transmission of this behavior through the genetic code. Alternatively or in addition to this, such behavior could be part of the ecological legacy of the Samaritan society maximizing fitness of its members.38

In summary, all five explanations are valid at the same time and can be combined into a coherent model of the Samaritan’s behavior. The first answer comprises external proximate causes and can be regarded as the top contextual layer. Deciding whether the Samaritan has enough time before the dark, or enough oil and wine to help the sufferer, or even whether his act will have some legal consequences are all important contextual factors that can help explain his decision. Indeed, this is the layer most familiar to humanists. It can address what cultural systems embed the actor’s behavior, what is the role of institutions in these systems, what are the behavioral norms, or how people rationalize their decisions. In other words, this layer can analyze contextualized motivations, intentions, and values. Methods used on this level can range from ethnographic observation and comparison, through historical textual analysis, survey research, to psychological experiments.39 Furthermore, since institutional practices and discourses can be regarded as an indispensable part of one’s social and cognitive niche (Whiten & Erdal, 2012), social construction theory (Berger & Luckmann, 1991) and discourse analysis (Taira, 2013; von Stuckrad, 2013) can analyze how people navigate through these ecological niches and use or misuse their power relations to maximize their biological fitness.

The second answer may be of high relevance to humanists as well, because it employs similar methods as the first answer. The Samaritan’s individual development in a specific cultural niche will play an important role in his decisions. For example, if the Samaritan would be raised to help only Samaritans and to despise other religious groups, he might have a strong reluctance toward helping the suffering Jew. Socialization into cultural norms will be enormously influential (Berger & Luckmann, 1991). However, the socialization process can be also understood as an interactive exchange between the subject and her or his environment. This exchange facilitates the calibration of mental mechanisms’ sensitivity and triggers gene expressions, thereby influencing the future workings of internal proximate mechanisms (Champagne & Mashoodh, 2009; Müller, 2007). While some behavioral traits are innate, they always develop in an interaction with environment (Tinbergen, 1968). If, for instance, the Good

38 While I emphasize genetic transmission in this example, in theory, both the genetic and cultural inheritance can work together (Richerson & Boyd, 2005). 39 Admittedly, scholars working only with text might not have enough data to include other levels than the external proximate causes, but they can leave the possibility of other-level explanations open for researchers willing to bring additional evidence from different disciplines. For a good example see Chalupa (2014).

- 111 -

Samaritan would be desensitized to human suffering by engaging in mortal combats during his childhood, his empathic concern for the robbed victim might not be strong enough to trigger the cascade of neurophysiological responses to distress signals. Understanding the socialization forces together with developmental sensitivities may be crucial for elucidating a mechanism’s functioning in a specific context.40 Indeed, humans are very special in having an extremely long ontogenetic period that is needed for the development of extremely complex human bodies and minds as well as for socialization into enormously complex cultural niches (Kaplan, Hill, Lancaster, & Hurtado, 2000).

The third answer concerning specific neurophysiological mechanisms triggered by the perception of a suffering fellow human can be regarded as an internal proximate mechanism. This analytical level is the crux of the mechanistic approach, and will be of highest interest to cognitive and behavioral scientists, together with psychophysiologists and neuroscientists. Going from visual, olfactory, and auditory perception, through specific neuronal and neuroendocrine activations to an initiation of motor sequences with the goal of helping the robbed victim will require a complex model of the inner mechanistic composition and organization. While I focused on the mechanism of empathic concern in my example, there might be other important mechanisms acting simultaneously that should be considered. For example, the flight or fight response may be an important mechanism that needs to be regulated in the helping context, else people could flee the dangerous situation as illustrated by priest and Levite in the Good Samaritan parable. As emphasized before, hypothesizing about active mechanisms mediating the observed behavior will be crucially dependent on context.

Likewise, the fourth and fifth ultimate explanations can be potentially altered by reputational concerns. The Samaritan’s decision to help the robbed traveler can be explained as stemming from an evolved trait that favors cooperation as a strategy maximizing one’s biological fitness (Axelrod & Hamilton, 1981). Although costly for the Samaritan, such helping behavior could increase his positive reputation, thereby heightening the future likelihood of being helped and, consequentially, enhancing the fitness of his and related genomes (Hamilton, 1964; Trivers, 1971). Complementary and alternative ultimate explanations can be offered by the niche construction theory, for instance, emphasizing the adaptive value of socially

40 However, it would be difficult to postulate the developmental influences as exclusively proximate or ultimate causes (Laland, Sterelny, Odling-Smee, Hoppitt, & Uller, 2011). While the distinction between proximate and ultimate is useful, it is a simplification nonetheless. I positioned development onto the depth axis what is consonant with the diachronic perspective of this axis, yet ontogeny does not answer ultimate questions. Rather, it translates ultimate principles into proximate ones.

- 112 -

constructed norms affecting people’s decision making. These questions will be of the foremost interest to evolutionary biologist and evolutionary psychologists, together with human behavioral ecologists.

Moving along the three Euclidean coordinates of contextual specificity, mechanistic composition, and evolutionary history should allow scholars to mutually corroborate their theories by projecting them against all three axes. Well specified models should be able to move smoothly along the range of values on all those axes, combining external and internal proximate explanations with ultimate explanations. Of course, by working in their own fields, different scholars will build models that will include the whole range of, for example, mechanistic composition, but will be anchored only in a very narrow lower segment of cultural specificity. But scholars educated in contextual disciplines can enrich the suggested mechanisms with external proximate causes, thereby widening the contextual width axis and stimulating the redefinition of a mechanism’s components and organization.

Nevertheless, some scholars might stress only the contextual specificity axis and disregard the other two. Indeed, some phenomena can have low explanatory power on the depth and height axes, especially when dealing with idiosyncratic influences of an individual’s history on specific micro-contextual acts. But evolved embodied mechanisms will still be present during such acts, and can be, therefore, analyzed alongside the micro-context. Whether such an analysis would be feasible is to be decided by individual scholars on the basis of theoretical assumptions and expectations. Likewise, it would not be practical to specify the workings of a mechanism under each possible condition. Rather, prior knowledge and hypotheses should drive suggestions of important conditions influencing the mechanism’s behavior. Identifying such important environments requires contextual knowledge and intuition, skills that are well developed by the humanities (Slingerland & Collard, 2011).

In conclusion, the contextualized, proximate, and ultimate mechanistic analyses afford scholars to move freely within the 3-dimensional space when exploring religious phenomena, drawing from various interdisciplinary methodologies and approaches. Nonetheless, granting such “intellectual omnivory” also brings out the hardest problem yet to be solved by the mechanistic analyses: how to study the joint workings of different mechanisms. By fragmenting religion to investigate its sub-phenomena and mechanisms on lower levels, the observed high-level phenomenon is “destroyed” (Bechtel & Hamilton, 2007). Only by assembling the fragments of mechanistic knowledge back together, one can catch a glimpse of

- 113 -

the emerging phenomena of religion (Sosis, 2009). Nevertheless, such a synthesis will have to deal with the exponentially increasing complexity of higher explanatory levels. While we might be able to discover mechanisms that govern certain behaviors, synthetizing the workings of these mechanisms in order to predict human behavior turns out to be hardly possible. E. O. Wilson (1998) aptly describes this situation as if one would go through a branched labyrinth: it is relatively easy to go backwards to the laws of physics (reduction), while going forward through multiple branching corridors would usually lead to different end points (synthesis). In other words, when looking from bottom-up, it would be difficult to predict where one would end (1998, pp. 73 – 74). Mathematical models using coupled non-linear equations may be able to capture the basic elements and laws of a particular behavior (Haken, Kelso, & Bunz, 1985; M. Richardson, Dale, & Marsh, 2014), yet the sheer number of influences that enter these equations only as stochastic noise can significantly hinder successful predictions (Strogatz, 1994). Synthetizing mechanistic explanations in order to make predictions is thus another important step in the understanding of human behavior. Arguably, this step is even more challenging than bridging the humanities-science divide.

- 114 -

REFERENCES

Adamczyk, A. (2009). Understanding the effects of personal and school religiosity on the decision to abort a premarital pregnancy. Journal of Health and Social Behavior, 50(2), 180–195.

Adler, D. S. (2009). Archeology: The earliest musical tradition. Nature, 460(7256), 695–696.

Ahmed, A., & Salas, O. (2013). Religious context and prosociality: An experimental study from Valparaíso, Chile. Journal for the Scientific Study of Religion, 52(3), 627–637.

Albright, C. R. (2000). The “God Module” and the complexifying brain. Zygon®, 35(4), 735–744.

Alcorta, C. S., & Sosis, R. (2005). Ritual, emotion, and sacred symbols: The evolution of religion as an adaptive complex. Human Nature, 16(4), 323–359.

Aron, A., Aron, E., & Smollan, D. (1992). Inclusion of other in the self scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology, 63(4), 596–612.

Asprem, E. (2015). Reverse-engineering ’esotericism ’: how to prepare a complex cultural concept for the cognitive science of religion. Religion, 46(2), 158–185.

Aveyard, M. E. (2014). A call to honesty: Extending religious priming of moral behavior to Middle Eastern Muslims. PLoS ONE, 9(7), e9947.

Axelrod, R., & Hamilton, W. (1981). The evolution of cooperation. Science, 211(4489), 1390–1396.

Bahna, V., Lang, M., Reddish, P., Shaver, J. H., & Xygalatas, D. (Forthcoming). Effects of behavioral synchrony on pain threshold and cooperation.

Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: direct effects of trait construct and stereotype-activation on action. Journal of Personality and Social Psychology, 71(2), 230–244.

Bargh, J. A., & Morsella, E. (2008). The unconscious mind. Perspectives on Psychological Science, 3(1), 73–79.

- 115 -

Bargh, J. A., Schwader, K. L., Hailey, S. E., Dyer, R. L., & Boothby, E. J. (2012). Automaticity in social-cognitive processes. Trends in Cognitive Sciences, 16(12), 593– 605.

Barrera, T. L., & Norton, P. J. (2009). Quality of life impairment in generalized anxiety disorder, social phobia, and panic disorder. Journal of Anxiety Disorders, 23(8), 1086– 90.

Barrett, J. (2011). Cognitive science of religion: Looking back, looking forward. Journal for the Scientific Study of Religion, 50(2), 229–239.

Bastian, B., Jetten, J., & Ferris, L. J. (2014). Pain as social Glue: Shared pain increases cooperation. Psychological Science, 25(11), 2079–2085.

Bateson, M., Nettle, D., & Roberts, G. (2006). Cues of being watched enhance cooperation in a real-world setting. Biology Letters, 2(3), 412–4.

Bateson, P., & Laland, K. N. (2013). Tinbergen’s four questions: An appreciation and an update. Trends in Ecology and Evolution, 28(12), 712–718.

Baumann, S., Koeneke, S., Schmidt, C. F., Meyer, M., Lutz, K., & Jancke, L. (2007). A network for audio-motor coordination in skilled pianists and non-musicians. Brain Research, 1161, 65–78.

Baumard, N., Hyafil, A., Morris, I., & Boyer, P. (2015). Increased affluence explains the emergence of ascetic wisdoms and moralizing religions. Current Biology, 25(1), 10–15.

Bechtel, W. (2008). Mental Mechanisms: Philosophical Perspectives on Cognitive Neuroscience. New York, London: Routledge; Tylor and Francis Group.

Bechtel, W. (2009). Looking down, around, and up: Mechanistic explanation in psychology. Philosophical Psychology, 22(5), 543–564.

Bechtel, W. (2011). Mechanism and biological Explanation. Philosophy of Science, 78(4), 533–557.

Bechtel, W., & Abrahamsen, A. (2010). Dynamic mechanistic explanation: Computational modeling of circadian rhythms as an exemplar for cognitive science. Studies in History and Philosophy of Science Part A, 41(3), 321–333.

Bechtel, W., & Hamilton, A. (2007). Reduction, integration, and the unity of science:

- 116 -

Natural, behavioral, and social sciences and the humanities. In T. Kuipers (Ed.), Philosophy of science: Focal issues (pp. 377–430). New York: Elsevier.

Bechtel, W., & Richardson, R. C. (2010). Discovering Complexity: Decomposition and Localization as Strategies in Sceintific Research (Vol. 1). Cambridge, MA; London: The MIT Press.

Berger, P. L. (1967). The Sacred Canopy: Elements of a Sociological Theory of Religion. Garden city, NY: Doubleday.

Berger, P. L., & Luckmann, T. (1991). The Social Construction of Reality: A Treatise in the Sociology of Knowledge. London: Penguin Books.

Bering, J. M., McLeod, K., & Shackelford, T. K. (2005). Reasoning about dead agents reveals possible adaptive trends. Human Nature, 16(4), 360–381.

Bickle, J. (2003). Philosophy and Neuroscience: A Ruthlessly Reductive Account. Dordrecht: Kluwer Academic.

Blaydes, L., & Gillum, R. M. (2013). Religiosity-of-interviewer effects: Assessing the impact of veiled enumerators on survey response in Egypt. Politics and Religion, 6(3), 459– 482.

Bond, A., & Lader, M. (1974). The use of analogue scales in rating subjective feelings. British Journal of Medical Psychology, 47(3), 211–218.

Boyer, P. (1994). The Naturalness of Religious Ideas: A Cognitive Theory of Religion. London, Berkeley, Los Angeles: University of California Press.

Boyer, P. (2000). Evolutionary psychology and cultural transmission. American Behavioral Scientist, 43(6), 987–1000.

Boyer, P. (2001). Religion Explained: The Evolutionary Origins of Religious Thought. New York: Basic Books.

Boyer, P. (2003). Religious thought and behaviour as by-products of brain function. Trends in Cognitive Sciences, 7(3), 119–124.

Boyer, P. (2011). From studious irrelevancy to consilient knowledge: Modes of scholarship and cultural anthropology. In E. Slingerland & M. Collard (Eds.), Creating Consilience: Integrating the Sciences and the Humanities. Oxford: Oxford University Press.

- 117 -

Boyer, P., & Liénard, P. (2006). Why ritualized behavior? Precaution systems and action parsing in developmental, pathological and cultural rituals. Behavioral and Brain Sciences, 29, 1–56.

Boyer, P., & Liénard, P. (2008). Ritual behavior in obsessive and normal individuals: Moderating anxiety and reorganizing the flow of action. Current Directions in Psychological Science, 17(4), 291–294.

Brenner, P. S. (2011). Exceptional behavior or exceptional identity? Public Opinion Quarterly, 75(1), 19–41.

Brochard, R., Tassin, M., & Zagar, D. (2013). Got rhythm…for better and for worse. Cross- modal effects of auditory rhythm on visual word recognition. Cognition, 127(2), 214–9.

Brown, S. (2000). The “Musilanguage” model of music. In N. L. Wallin, B. Merker, & S. Brown (Eds.), The Origins of Music. Cambridge: MIT Press.

Bruyn, L. De, Leman, M., Moelants, D., & Demey, M. (2009). Does social interaction activate music listeners? In R. Kronland-Martinet, S. Ystad, & K. Jensen (Eds.), Computer Music, Modeling and Retrieval. Genesis of Meaning of Sound and Music. (pp. 93–106). Berlin, Heidelberg: Springer-Verlag.

Bulbulia, J., Xygalatas, D., Schjoedt, U., Fondevila, S., Sibley, C. G., & Konvalinka, I. (2013). Images from a jointly-arousing collective ritual reveal affective polarization. Frontiers in Psychology, 4( 960), 1-11.

Bulbulia, J., & Slingerland, E. (2012). Religious Studies as a Life Science. Numen, 59, 564– 613.

Bulbulia, J., Wilson, M. S., & Sibley, C. G. (2014). Thin and thinner: Hypothesis-driven research and the study of humans. Numen, 61(2–3), 166–181.

Burlein, A. (2012). Knowledge is made for cutting: Foucault, cognitive science, and intellectual taste. Method & Theory in the Study of Religion, 24(2), 118–142.

Butler, M. (2006). Unlocking the Groove: Rhythm, Meter, and Musical Design in Electronic Dance Music. Bloomington: Indiana University Press.

Button, K. S., Ioannidis, J. P. a, Mokrysz, C., Nosek, B. a, Flint, J., Robinson, E. S. J., & Munafò, M. R. (2013). Power failure: why small sample size undermines the reliability

- 118 -

of neuroscience. Nature Reviews. Neuroscience, 14(5), 365–76.

Campbell, D. T. (1958). Common fate, similarity, and other indices of the status of aggregates of persons as social entities. Behavioral Science, 3(1), 14–25. article.

Chalupa, A. (2014). Pythiai and inspired divination in the Delphic Oracle: Can cognitive sciences provide us with an access to “dead minds”? Journal of Cognitive Historiography, 1, 24–51.

Champagne, F. A., & Mashoodh, R. (2009). Genes in Context: Gene-Environment Interplay and the Origins of Individual Differences in Behavior. Current Directions in Psychological Science, 18(3), 127–131.

Chartrand, T., & Bargh, J. (1999). The Chameleon Effect: The perception-behavior link and social interaction. Journal of Personality and Social Psychology, 76(6), 893–910.

Chaves, M. (2010). Rain fances in the fry deason: Overcoming the religious congruence fallacy. Journal for the Scientific Study of Religion, 49(1), 1–14.

Chen, J., Penhune, V., & Zatorre, R. (2009). The role of auditory and premotor cortex in sensorimotor transformations. Annals of the New York Academy of Sciences, 1169, 15– 34.

Chen, J., Zatorre, R., & Penhune, V. (2006). Interactions between auditory and dorsal premotor cortex during synchronization to musical rhythms. NeuroImage, 32(4), 1771– 81.

Chen, Y., Repp, B., & Patel, A. (2002). Spectral decomposition of variability in synchronization and continuation tapping: Comparisons between auditory and visual pacing and feedback conditions. Human Movement Science, 21, 515–532.

Cialdini, R. B., Reno, R. R., & Kallgren, C. a. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology.

Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. The Behavioral and Brain Sciences, 36(3), 181–204.

Conard, N. J., Malina, M., & Münzel, S. C. (2009). New flutes document the earliest musical tradition in southwestern Germany. Nature, 460(7256), 737–40.

- 119 -

Conrad, C. (2011). The Handbook of Stress: Neurophysiological Effects on the Brain. Wiley Online Library 9781118083222: Wiley-Blackwell.

Cribari-Neto, F., & Zeileis, A. (2010). Beta regression in R. Journal of Statistical Software, 34(2), 1–24.

Cross, I., & Morley, I. (2008). The evolution of music: Theories, definitions and the nature of the Evidence. In I. Cross & I. Morley (Eds.), Communicative Musicality (pp. 61–82). Oxford: Oxford University Press.

Damisch, L., Stoberock, B., & Mussweiler, T. (2010). Keep your fingers crossed! How superstition improves performance. Psychological Science, 21(7), 1014–1020.

Darley, J. M., & Batson, C. D. (1973). “From Jerusalem to Jericho”: A study of situational and dispositional variables in helping behavior. Journal of Personality and Social Psychology, 27(1), 100–108. de Bruijn, E. R., Miedl, S. F., & Bekkering, H. (2011). How a co-actor’s task affects monitoring of own errors: evidence from a social event-related potential study. Experimental Brain Research, 211(3–4), 397–404. de Jong, L. H. (2002). Levels of explanation in biological psychology. Philosophical Psychology, 15(4), 441–462. de Jong, L. H., & Schouten, M. K. D. (2005). Ruthless reductionism: A review essay of John Bickle’s Philosophy and neuroscience: A ruthlessly reductive account. Philosophical Psychology, 18(4), 473–486.

Demos, A., Chaffin, R., Begosh, K., Daniels, J., & Marsh, L. (2012). Rocking to the beat: effects of music and partner’s movements on spontaneous interpersonal coordination. Journal of Experimental Psychology. General, 141(1), 49–53.

Desmet, F., Leman, M., & Lesaffre, M. (2010). Statistical analysis of human body movement and group interactions in response to music. In A. Fink, B. Lausen, W. Seidel, & A. Ultsch (Eds.), Advances in Data Analysis, Data Handling, and Business Inteligence. Berlin, Heidelberg: Springer-Verlag.

Dhamala, M., Pagnoni, G., Wiesenfeld, K., Zink, C. F., Martin, M., & Berns, G. S. (2003). Neural correlates of the complexity of rhythmic finger tapping. NeuroImage, 20(2), 918–26.

- 120 -

Dissanayake, E. (2006). Ritual and ritualization: Musical means of conveying and shaping emotion in humans and other animals. In S. Brown & U. Voglsten (Eds.), Music and Manipulation: On the Soical Uses and Soical Control of Music (pp. 31–56). Oxford, New York: Berghahn Books.

Douglas, M. (1970/2004). Natural symbols: Explorations in cosmology. London, New York: Routledge.

Dulaney, S., & Fiske, A. (1994). Cultural rituals and Obsessive-Compulsive Disorder: Is there a common psychological mechanism? Ethos, 22(3), 243–283.

Dunbar, R. I. M., Kaskatis, K., MacDonald, I., & Barra, V. (2012). Performance of music elevates pain threshold and positive affect: implications for the evolutionary function of music. Evolutionary Psychology, 10(4), 688–702.

Dupré, J. (1996). Metaphysical disorder and scientific disunity. In P. Galison & D. Stump (Eds.), The Disunity of Sciecne: Boundaries, Context, and Power (pp. 101–17). Standford, California: Standford University Press.

Durkheim, E. (1912/1964). The Elementary Forms of the Religious Life. London: George Allen & Unwin LTD.

Durkheim, E. (1895/1982). The Rules of Sociological Method. New York: The Free Press.

Eilam, D. (2006). Ritualized behavior in animals and humans: Time, space, and attention. The Behavioral and Brain Sciences, 29(6), 22–23.

Eilam, D. (2015). The cognitive roles of behavioral variability: Idiosyncratic acts as the foundation of identity and as transitional, preparatory, and confirmatory phases. Neuroscience & Biobehavioral Reviews, 49, 55–70.

Eilam, D., Izhar, R., & Mort, J. (2011). Threat detection: behavioral practices in animals and humans. Neuroscience and Biobehavioral Reviews, 35(4), 999–1006.

Eilam, D., Zor, R., Szechtman, H., & Hermesh, H. (2006). Rituals, stereotypy and compulsive behavior in animals and humans. Neuroscience & Biobehavioral Reviews, 30(4), 456–471.

Engler, S., & Gardiner, M. Q. (2010). Ten implications of semantic holism for theories of religion. Method & Theory in the Study of Religion, 22, 283–292.

- 121 -

Englert, C., & Oudejans, R. R. D. (2014). Is choking under pressure a consequence of skill- focus or increased distractibility ? Results from a tennis serve task. Psychology, 5, 1035– 1043.

Escoffier, N., Sheng, D. Y. J., & Schirmer, A. (2010). Unattended musical beats enhance visual processing. Acta Psychologica, 135(1), 12–6.

Eskelson, B., & Madsen, L. (2011). Estimating Riparian understory vegetation cover with Beta regression and copula models. Forest Science, 57(3), 212–221.

Evans-Pritchard, E. (1937/1976). Witchcraft, Oracles, and Magic among the Azande. Oxford: Clarendon Press.

Evans, D., Gray, F., & Leckman, J. (1999). The rituals, fears and phobias of young children: Insights from development, psychopathology and neurobiology. Child Psychiatry and Human Development, 29(4), 261–276.

Eysenck, M., Derakshan, N., Santos, R., & Calvo, M. (2007). Anxiety and cognitive performance: attentional control theory. Emotion, 7(2), 336–53.

Fam, S. D., Tan, Y. S., & Waitt, C. (2012). Stereotypies in captive primates and the use of Inositol: Lessons from Obsessive-Compulsive Disorder in humans. International Journal of Primatology, 33, 830–844.

Feldman, P. J., Cohen, S., Hamrick, N., & Lepore, S. J. (2004). Psychological stress, appraisal, emotion and Cardiovascular response in a public speaking task. Psychology & Health, 19(3), 353–368.

Felson, R., & Gmelch, G. (1979). Uncertainty and the use of magic. Current Anthropology, 20(3), 587–589.

Fentress, J. (1976). Dynamic boundaries of patterned behavior: Interaction and self- organization. In P. Bateson & R. A. Hinde (Eds.), Growing Points in Ethology (pp. 135– 67). Cambridge, MA: Cambridge University Press.

Feyerabend, P. K. (1975). Against Method: Outline of an Anarchist Theory of Knowledge. New York: Verso Books.

Fischer, R., Callander, R., Reddish, P., & Bulbulia, J. (2013). How do rituals affect cooperation? An experimental field study comparing nine ritual types. Human Nature,

- 122 -

24(2), 115–25.

Fischer, R., Xygalatas, D., Mitkidis, P., Reddish, P., Tok, P., Konvalinka, I., & Bulbulia, J. (2014). The fire-walker’s high: Affect and physiological responses in an extreme collective ritual. PLoS ONE, 9(2), e88355.

Fisher, R. J. (1993). Social desirability indirect questioning. Journal of Consumer Research, 20, 303–315.

Fitch, W. T. (2006). The biology and evolution of music: a comparative perspective. Cognition, 100(1), 173–215.

Fitch, W. T., & Rosenfeld, A. (2007). Perception and production of syncopated rhythms. Music Perception, 25(1), 43–58.

Foa, E. B., Zinbarg, R., & Rothbaum, B. O. (1992). Uncontrollability and unpredictability in post-traumatic stress disorder: an animal model. Psychological Bulletin, 112(2), 218– 238.

Franek, J. (2014). Has the cognitive science of religion (re)defined “religion”? Religio, 22(1), 3–27.

Frazer, J. G. (1894). The Golden Bough. A Study in Comparative Religion. New York, London: MacMillan and Co.

Freud, S. (1961). The Future of an Illusion. New York: WW Norton & Company Inc.

Friston, K. (2009). The free-energy principle: a rough guide to the brain? Trends in Cognitive Sciences, 13, 293–301.

Furrow, J. L., King, P. E., & White, K. (2004). Religion and positive youth development: Identity, meaning, and prosocial concerns. Applied Developmental Science, 8(1), 17–26.

Galen, L. W. (2012). Does religious belief promote prosociality? A critical examination. Psychological Bulletin, 138(5), 876–906.

Gallese, V., Gernsbacher, M. a., Heyes, C., Hickok, G., & Iacoboni, M. (2011). Mirror Neuron Forum. Perspectives on Psychological Science, 6(4), 369–407.

Geertz, A. W. (2008). How Not to Do the Cognitive Science of Religion Today. Method & Theory in the Study of Religion, 20(1), 7–21.

- 123 -

Geertz, A. W. (2010). Brain, body and culture: A biocultural theory of religion. Method & Theory in the Study of Religion, 22(4), 304–321.

Geertz, A. W. (2014). Whence religion? How the brain constructs the world and what this might tell us about the origins of religion, cognition and culture. In A. W. Geertz (Ed.), Origins of Religion, Cognition, and Culture (pp. 17–70). New York: Routledge.

Geertz, C. (1973). The Interpretation of Cultures. New York: Basic Books.

Geertz, C. (1983). Local Knowledge: Further Essays in Interpretative Anthropology. London: Fontana Press.

Gervais, W. M., & Norenzayan, A. (2012). Like a camera in the sky? Thinking about God increases public self-awareness and socially desirable responding. Journal of Experimental Social Psychology, 48(1), 298–302.

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), 389–410.

Gervais, W. M., Xygalatas, D., & McKay, R. T. (Under review). Intuitive moral distrust of Atheists Across 13 Countries.

Gillott, A., Furniss, F., & Walter, A. (2001). Anxiety in high-functioning children with autism. Autism : The International Journal of Research and Practice, 5(1994), 277–286.

Gino, F., Ayal, S., & Ariely, D. (2009). Contagion and differentiation in unethical behavior: the effect of one bad apple on the barrel. Psychological Science, 20(3), 393–8.

Gisbert, P. (1959). Social facts in Durkheim’s system. Anthropos, 3(4), 353–369.

Goffman, E. (1956). The Presentation of Self in Everyday Life. New York, NY: Doubleday.

Goldschmidt, W. (1966). Comparative Functionalism: An Essay in Anthropological Theory. Berkley, Los Angeles: University of California Press.

Gomes, C. M., & McCullough, M. E. (2015). The effects of implicit religious primes on dictator game allocations: A preregistered replication experiment. Journal of Experimental Psychology: General, 144(6), e94–e104.

Gould, S. J., & Lewontin, R. C. (1979). The spandrels of San Marco and the panglossian paradigm: A critique of the adaptationist programme. Proceedings of the Royal Society

- 124 -

of London. Series B, Biological Sciences, 295(1161), 581–598.

Gould, S. J., & Vrba, E. S. (1982). Exaptation - A missing term in the science of form. Paleobiology, 8(1), 4–15.

Graham, J., Meindl, P., & Beall, E. (2012). Integrating the streams of morality research: The case of political ideology. Current Directions in Psychological Science, 21(6), 373–377.

Grahn, J. A., & Brett, M. (2007). Rhythm and beat perception in motor areas of the brain. Journal of Cognitive Neuroscience, 19(5), 893–906.

Hagens, T. G. (2006). Conscience collective or false consciousness?: Adorno’s critique of Durkheim’s sociology of morals. Journal of Classical Sociology, 6(2), 215–237.

Haken, H., Kelso, J., & Bunz, H. (1985). A theoretical model of phase transition in human hand movements. Biological Cybernetics, 51, 347–356.

Hamer, D. (2005). The God Gene: How Faith is Hardwired into our Genes. New York: Anchor Books.

Hamilton, W. D. (1964). The genetical evolution of social behaviour. Journal of Theoretical Biology, 7(1), 1–16.

Hardy, L. (1996). Testing the predictions of the cusp catastrophe model of anxiety and performance. The Sport Psychologist, 10, 140–156.

Hasson, U., & Frith, C. D. (2016). Mirroring and beyond: coupled dynamics as a generalized framework for modelling social interactions. Philosophical Transactions of the Royal Society B, 371, 20150366.

Hecht, E. E., Murphy, L. E., Gutman, D. A., Votaw, J. R., Schuster, D. M., Preuss, T. M., … Parr, L. A. (2013). Differences in neural activation for object-directed grasping in chimpanzees and humans. Journal of Neuroscience, 33(35), 14117–14134.

Hein, G., Silani, G., Preuschoff, K., Batson, C. D., & Singer, T. (2010). Neural responses to ingroup and outgroup members’ suffering predict individual differences in costly helping. Neuron, 68(1), 149–60.

Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? The Behavioral and Brain Sciences, 33(2–3), 61-83-135.

Hirsh, J., Galinsky, A., & Zhong, C. (2011). Drunk, powerful, and in the dark: How general

- 125 -

processes of disinhibition produce both prosocial and antisocial behavior. Perspectives on Psychological Science, 6(5), 415–427.

Hirsh, J., Mar, R. A., & Peterson, J. B. (2012). Psychological entropy: a framework for understanding uncertainty-related anxiety. Psychological Review, 119(2), 304–20.

Holländer, A., Jung, C., & Prinz, W. (2011). Covert motor activity on NoGo trials in a task sharing paradigm: evidence from the lateralized readiness potential. Experimental Brain Research. Experimentelle Hirnforschung. Expérimentation Cérébrale, 211(3–4), 345– 56.

House, B. R., Silk, J. B., Henrich, J., Barrett, H. C., Scelza, B. a, Boyette, A. H., … Laurence, S. (2013). Ontogeny of prosocial behavior across diverse societies. Proceedings of the National Academy of Sciences of the United States of America, 110(36), 14586–91.

Hove, M., & Risen, J. L. (2009). It’s all in the timing: Interpersonal synchrony increases affiliation. Social Cognition, 27(6), 949–960.

Huber, L., Range, F., Voelkl, B., Szucsich, A., Viranyi, Z., & Miklosi, A. (2009). The evolution of imitation: what do the capacities of non-human animals tell us about the mechanisms of imitation? Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1528), 2299–2309.

Huron, D. (2001). Is music an evolutionary adaptation? Annals of the New York Academy of Sciences, 930, 43–61.

Inzlicht, M., McGregor, I., Hirsh, J. B., & Nash, K. (2009). Neural markers of religious conviction. Psychological Science, 20(3), 385–92.

Ip, G. W., Chiu, C., & Wan, C. (2006). Birds of a feather and birds flocking together: Physical versus behavioral cues may lead to trait- versus goal-based group perception. Journal of Personality and Social Psychology, 90(3), 368–381.

Izhar, R., & Eilam, D. (2010). Together they stand: A life-threatening event reduces individual behavioral variability in groups of voles. Behavioural Brain Research, 208(1), 282–285.

John, D., & Freedson, P. (2012). ActiGraph and Actical physical activity monitors: a peek under the hood. Medicine and Science in Sports and Exercise, 44, S86–S89.

- 126 -

John, L. K., Loewenstein, G., & Rick, S. I. (2014). Cheating more for less: Upward social comparisons motivate the poorly compensated to cheat. Organizational Behavior and Human Decision Processes, 123(2), 101–109.

Kahneman, D. (2011). Thinking, fast and slow. New York: Farrar, Straus and Giroux.

Kaneko, T., & Tomonaga, M. (2012). Relative contributions of goal representation and kinematic information to self-monitoring by chimpanzees and humans. Cognition, 125(2), 168–178.

Kaplan, H., & Hill, K. (1985). Hunting ability and reproductive success among male ache foragers: Preliminary results. Current Anthropology, 26(2), 223-246.

Kaplan, H., Hill, K., Lancaster, J., & Hurtado, A. M. (2000). A theory of human life history evolution: Diet, intelligence, and longevity. Evolutionary Anthropology, 9, 156–185.

Kass, S. J., Cole, K. S., & Stanny, C. J. (2007). Effects of distraction and experience on situation awareness and simulated driving. Transportation Research Part F: Traffic Psychology and Behaviour, 10(4), 321–329.

Kay, P., & Kempton, W. (1984). What is the Sapir-Whorf hypothesis? American Anthropologist, 86(1), 65–79.

Keinan, G. (1994). Effects of stress and tolerance of ambiguity on magical thinking. Journal of Personality and Social Psychology, 67(1), 48–55.

Keinan, G. (2002). The effects of stress and desire for control on superstitious behavior. Personality and Social Psychology Bulletin, 28(1), 102–108.

Kelemen, D., Rottman, J., & Seston, R. (2012). Professional Physical Scientists Display Tenacious Teleological Tendencies: Purpose-Based Reasoning as a Cognitive Default. Journal of Experimental Psychology: General, 142(4), 1074.

Kennel, M., Brown, R., & Abarbanel, H. (1992). Determining embedding dimension for phase-space reconstruction using a geometrical construction. Physical Review A, 45(6), 3403.

Keren, H., Boyer, P., Mort, J., & Eilam, D. (2010). Pragmatic and idiosyncratic acts in human everyday routines: The counterpart of compulsive rituals. Behavioural Brain Research, 212(704), 90–5.

- 127 -

Keren, H., Boyer, P., Mort, J., & Eilam, D. (2013). The impact of precaution and practice on the performance of a risky motor task. Behavioral Sciences, 3(3), 316–329.

Kirkpatrick, L. (2011). The role of evolutionary psychology within an interdisciplinary science of religion. Religion, 41(3), 329–339.

Kirkpatrick, L. (2012). Attachment theory and the evolutionary psychology of religion. International Journal for the Psychology of Religion, 22(3), 231–241.

Kirschner, S., & Tomasello, M. (2009). Joint drumming: social context facilitates synchronization in preschool children. Journal of Experimental Child Psychology, 102(3), 299–314.

Kirschner, S., & Tomasello, M. (2010). Joint music making promotes prosocial behavior in 4- year-old children. Evolution and Human Behavior, 31(5), 354–364.

Knoblich, G., & Jordan, J. S. (2003). Action coordination in groups and individuals: learning anticipatory control. Journal of Experimental Psychology. Learning, Memory, and Cognition, 29(5), 1006–16.

Koelsch, S. (2010). Towards a neural basis of music-evoked emotions. Trends in Cognitive Sciences, 14(3), 131–7.

Koelsch, S. (2011). Toward a neural basis of music perception - a review and updated model. Frontiers in Psychology, 2(110), 1-20.

Koelsch, S., Kasper, E., Sammler, D., Schulze, K., Gunter, T., & Friederici, A. D. (2004). Music, language and meaning: brain signatures of semantic processing. Nature Neuroscience, 7(3), 302–7.

Konvalinka, I., Vuust, P., Roepstorff, A., & Frith, C. D. (2010). Follow you, follow me: continuous mutual prediction and adaptation in joint tapping. Quarterly Journal of Experimental Psychology (2006), 63(11), 2220–30.

Konvalinka, I., Xygalatas, D., Bulbulia, J., Schjoedt, U., Jegindö, E.-M., Wallot, S., … Roepstorff, A. (2011). Synchronized arousal between performers and related spectators in a fire-walking ritual. PNAS, 108(20), 8514–8519.

Krátký, J., Lang, M., Shaver, J. H., Jerotijević, D., & Xygalatas, D. (2016). Anxiety and ritualization: Can attention discriminate compulsion from routine? Communicative &

- 128 -

Integrative Biology, 9(3), e1174799.

Krátký, J., McGraw, J. J., Xygalatas, D., Mitkidis, P., & Reddish, P. (2016). It depends who is watching you: 3-D agent cues increase fairness. PLoS ONE, 11(2), 1–11.

Kuhn, T. (1962). The Structure of Scientific Revolutions (1996 ed.). Chicago, London: The University of Chicago Press.

Kundt, R. (2015). Contemporary evolutionary theories of culture and the study of religion. London, New Delhi: Blackwell Scientific.

LaBouff, J. P., Rowatt, W. C., Johnson, M. K., & Finkle, C. (2012). Differences in attitudes toward outgroups in religious and nonreligious contexts in a multinational sample: A situational context priming study. International Journal for the Psychology of Religion, 22(1), 1–9.

Lahav, A., Saltzman, E., & Schlaug, G. (2007). Action representation of sound: audiomotor recognition network while listening to newly acquired actions. The Journal of Neuroscience : The Official Journal of the Society for Neuroscience, 27(2), 308–14.

Lakens, D. (2010). Movement synchrony and perceived entitativity. Journal of Experimental Social Psychology, 46(5), 701–708.

Laland, K., Kendal, J. R., & Brown, G. R. (2007). The niche construction perspective. Journal of Evolutionary Psychology, 5(1–4), 51–66.

Laland, K., Matthews, B., & Feldman, M. W. (2016). An introduction to niche construction theory. Evolutionary Ecology, 30(2), 191–202.

Laland, K., Sterelny, K., Odling-Smee, J., Hoppitt, W., & Uller, T. (2011). Cause and effect in biology revisited: is Mayr’s proximate-ultimate dichotomy still useful? Science, 334(6062), 1512–6.

Lang, M., Krátký, J., Shaver, J. H., Jerotijević, D., & Xygalatas, D. (2015). Effects of anxiety on spontaneous ritualized behavior. Current Biology, 25(14), 1892–1897.

Lang, M., Mitkidis, P., Kundt, R., Nichols, A., Krajčíková, L., & Xygalatas, D. (2016). Music as a sacred cue? Effects of religious music on moral Behavior. Frontiers in Psychology, 7(814), 1–13.

Lang, M., Shaw, D. J., Reddish, P., Wallot, S., Mitkidis, P., & Xygalatas, D. (2015). Lost in

- 129 -

the rhythm: Effects of rhythm on subsequent interpersonal coordination. Cognitive Science. E-print before publiation.

Langen, M., Durston, S., Kas, M. J. H., van Engeland, H., & Staal, W. G. (2011). The neurobiology of repetitive behavior: …and men. Neuroscience and Biobehavioral Reviews, 35(3), 356–65.

Large, E. W., & Jones, M. (1999). The dynamics of attending: How people track time- varying events. Psychological Review, 106(1), 119–159.

Large, E. W., & Palmer, C. (2002). Perceiving temporal regularity in music. Cognitive Science, 26(1), 1–37.

Lawson, E. T., & McCauley, R. N. (1990). Rethinking Religion: Connecting Cognition and Culture. Cambridge: Cambridge University Press.

Lazarus, R., & Folkman, S. (1984). Stress, Appraisal, and Coping. New York: Springer.

Leekam, S. R., Prior, M. R., & Uljarevic, M. (2011). Restricted and repetitive behaviors in autism spectrum disorders: a review of research in the last decade. Psychological Bulletin, 137(4), 562–93.

Liénard, P., & Boyer, P. (2006). Whence collective rituals? A cultural selection model of ritualized behavior. American Anthropologist, 108(4), 814–827.

London, J. (1995). Some examples of complex meters and their implications for models of metric perception. Music Perception, 13(1), 59–77.

Longuet-Higgins, H., & Lee, C. (1984). The rhythmic interpretation of monophonic music. Music Perception, 1(4), 424–441.

Malhotra, D. (2008). (When) are religious people nicer? Religious salience and the’sunday effect’on pro-social behavior. Judgement and Decision Making, 5(2), 138–143.

Malinowski, B. (1948/1992). Magic, Science and Religion and Other Essays. Long Grove, Il: Waveland Press Inc.

Marwan, N., & Kurths, J. (2002). Nonlinear analysis of bivariate data with cross recurrence plots. Physics Letters A, 302(5), 299-307.

Marwan, N., Romano, M. C., Thiel, M., & Kurths, J. (2007). Recurrence plots for the analysis of complex systems. Physics Reports, 438(5–6), 237–329.

- 130 -

Marwan, N., Wessel, N., Meyerfeldt, U., Schirdewan, A., & Kurths, J. (2002). Recurrence plot based measures of complexity and its application to heart rate variability data. Phys. Rev. E., 66(2), 26702.

Mason, G. (1991). Stereotypies : a critical review. Animal Behaviour, 41, 1015–1037.

Mason, G., Clubb, R., Latham, N., & Vickery, S. (2007). Why and how should we use environmental enrichment to tackle stereotypic behaviour? Applied Animal Behaviour Science, 102, 163–188.

MathWorks Inc. (2013). MATLAB. Version 2013a. Natick, MA.

Maynard Smith, J. (1982). Evolution and the Theory of Games. Cambridge, London: Cambridge University Press.

Mayr, E. (1961). Cause and effect in biology. Science, 134(3489), 1501–1506.

Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: A theory of self- concept maintenance. Journal of Marketing Research, 45(6), 633–644.

Mazar, N., & Zhong, C.-B. (2010). Do green products make us better people? Psychological Science, 21(4), 494–8.

McCauley, R. N. (1999). Levels of explanation and cognitive architectures. In W. Bechtel & G. Graham (Eds.), A Companion to Cognitive Science. Oxford: Blackwell.

McCauley, R. N. (2007). Reduction: Models of cross-scientific relations and their implications for the psychology-neuroscience interface. In P. Thagard (Ed.), Philosophy of psychology and cognitive science. Amsterodam: Elsevier.

McCauley, R. N. (2009). Time is of the essence: Explanatory pluralism and accommodating theories about long-term processes. Philosophical Psychology, 22, 611–635.

McCauley, R. N., & Bechtel, W. (2001). Explanatory pluralism and heuristic identity theory. Theory & Psychology, 11(6), 736–760.

McCauley, R. N., & Cohen, E. (2010). Cognitive science and the naturalness of religion. Philosophy Compass, 5(9), 779–792.

Mccullough, M. E., & Worthington, Everett L., J. (1999). Religion and the forgiving personality. Journal of Personality, 67(6), 1141–1164.

- 131 -

McNamara, P. (2009). The Neuroscience of Religious Experience. Cambridge: Cambridge University Press.

McNeill, W. H. (1995). Keeping Together in Time: Dance and Drill in Human History. Cambridge, MA: Harvard University Press.

Mead, N. L., Baumeister, R. F., Gino, F., Schweitzer, M. E., & Ariely, D. (2009). Too tired to tell the truth: Self-control resource depletion and dishonesty. Journal of Experimental Social Psychology, 45(3), 594–597.

Merker, B. H., Madison, G. S., & Eckerdal, P. (2009). On the role and origin of isochrony in human rhythmic entrainment. Cortex, 45(1), 4–17.

Miles, L. K., Nind, L. K., & Macrae, C. N. (2009). The rhythm of rapport: Interpersonal synchrony and social perception. Journal of Experimental Social Psychology, 45(3), 585–589.

Miller, G. (2000). Evolution of human music through sexual selection. In N. Wallin, B. Merker, & S. Brown (Eds.), The Origins of Music (pp. 329–360). Cambridge, MA: The MIT Press.

Mitchell, S. D. (2009). Unsimple Truths: Science, Complexity, and Policy. Chicago, London: The University of Chicago Press.

Mitkidis, P., Lienard, P., Nielbo, K. L., & Sørensen, J. (2014). Does goal-demotion enhance cooperation? Journal of Cognition and Culture, 14(3–4), 263–272.

Miura, A., Kudo, K., Ohtsuki, T., & Kanehisa, H. (2011). Coordination modes in sensorimotor synchronization of whole-body movement: a study of street dancers and non-dancers. Human Movement Science, 30(6), 1260–71.

Moulding, R., & Kyrios, M. (2006). Anxiety disorders and control related beliefs: the exemplar of Obsessive-Compulsive Disorder (OCD). Clinical Psychology Review, 26(5), 573–83.

Moulding, R., & Kyrios, M. (2007). Desire for control, sense of control and obsessive- compulsive symptoms. Cognitive Therapy and Research, 31, 759–772.

Mukherjee, S., Srinivasan, N., & Manjaly, J. A. (2014). Global processing fosters donations toward charity appeals framed in an approach orientation. Cognitive Processing, 15(3),

- 132 -

391–396.

Müller, G. B. (2007). Evo-devo: extending the evolutionary synthesis. Nature Reviews. Genetics, 8(12), 943–9.

Nagel, E. (1961). The Structure of Science. New York: Harcourt, Brace & World.

Nagel, T. (1974). What is it like to be a bat? The Philosophical Review, 83(4), 435–450.

Nagel, T. (1986). The View from Nowhere. New York: Oxford University Press.

Newberg, A. B. (2001). Putting the mystical mind together. Zygon, 36(3), 501–507.

Newell, B. R., & Shanks, D. R. (2014). Unconscious influences on decision making: A critical review. Behavioral and Brain Sciences, 37(1), 1–19.

Newman-Norlund, R. D., Bosga, J., Meulenbroek, R. G. J., & Bekkering, H. (2008). Anatomical substrates of cooperative joint-action in a continuous motor task: virtual lifting and balancing. NeuroImage, 41(1), 169–77.

Nielbo, K. L., & Sørensen, J. (2011). Spontaneous processing of functional and non- functional action sequences. Religion, Brain & Behavior, 1(1), 18–30.

Nieuwenhuys, A., & Oudejans, R. R. D. (2012). Anxiety and perceptual-motor performance: Toward an integrated model of concepts, mechanisms, and processes. Psychological Research, 76(6), 747–759.

Norenzayan, A. (2013). Big Gods: How Religion Transformed Cooperation and Conflict. Princeton, Oxford: Princeton University Press.

Norenzayan, A., & Shariff, A. F. (2008). The origin and evolution of religious prosociality. Science, 322(5898), 58–62.

Norenzayan, A., Shariff, A. F., Gervais, W. M., Willard, A. K., McNamara, R. A., Slingerland, E., & Henrich, J. (2016). The cultural evolution of prosocial eeligions. Behavioral and Brain Sciences, 39(e1), 1-65.

North, A., Mackenzie, L., Law, R., & Hargreaves, D. (2004). The effects of musical and voice “fit” on responses to advertisements. Journal of Applied Social Psychology, 34(8), 1675–1708.

Norton, M. I., & Gino, F. (2014). Rituals alleviate grieving for loved ones, lovers, and

- 133 -

lotteries. Journal of Experimental Psychology: General, 143(1), 266–72.

Nozaradan, S., Peretz, I., Missal, M., & Mouraux, A. (2011). Tagging the neuronal entrainment to beat and meter. Journal of Neuroscience, 31(28), 10234–10240.

Olatunji, B. O., Cisler, J. M., & Tolin, D. F. (2007). Quality of life in the anxiety disorders: a meta-analytic review. Clinical Psychology Review, 27(5), 572–81.

Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716-aac4716.

Paden, W. E. (1992). Interpreting the Sacred: Ways of Viewing Religion. Boston: Beacon Press.

Paden, W. E. (2013). Tracks and themes in a shifting landscape: reflections on 50 years of the study of religion. Religion, 43(1), 89–101.

Pearce, E., Launay, J., & Dunbar, R. I. M. (2015). The ice-breaker effect: singing mediates fast social bonding. Royal Society Open Science, 2, 150221.

Penner, H. H. (2000). Interpretation. In W. Braun & R. McCutcheon (Eds.), Guide to the Study of Religion. London: Casell.

Pezzulo, G., & Dindo, H. (2011). What should I do next? Using shared representations to solve interaction problems. Experimental Brain Research, 211(3–4), 613–30.

Phillips-Silver, J., & Trainor, L. J. (2005). Feeling the beat: movement influences infant rhythm perception. Science, 308(5727), 1430.

Piazza, J., Bering, J. M., & Ingram, G. (2011). “ Princess Alice is watching you”: Children’s belief in an invisible person inhibits cheating. Journal of Experimental Child Psychology, 109(3), 311–320.

Pinker, S. (1998). How the Mind Works. Penguin Books.

Pinker, S. (2003). The Blank Slate: The Modern Denial of Human Nature. New York: Penguin Books.

Popper, K. R. (1935/2005). The Logic of Scientific Discovery. London, New York: Routledge.

Preston, S. D., & de Waal, F. B. M. (2002). Empathy: Its ultimate and proximate bases.

- 134 -

Behavioral and Brain Sciences, 25(1), 1-20.

Purzycki, B. G., Apicella, C., Atkinson, Q. D., Cohen, E., McNamara, R. A., Willard, A. K., … Henrich, J. (2016). Moralistic gods, supernatural punishment and the expansion of human sociality. Nature, 530(7590), 327–330.

Putnam, R. D., Campbell, D. E., & Garrett, S. R. (2010). American Grace: How Religion Divides and Unites Us. New York, London: Simon & Schuster.

R Core Team. (2014). R: A Language and Environment for Statistical Computing. Version 3.2.3. Vienna: R Foundation for Statistical Computing.

Randolph-Seng, B., & Nielsen, M. E. (2007). Honesty: One effect of primed religious representations. The International Journal for the Psychology of Religion, 17(4), 303– 315.

Rappaport, R. (1999). Ritual and Religion in the Making of Humanity. Cambridge: Cambridge University Press.

Reddish, P., Bulbulia, J., & Fischer, R. (2014). Does synchrony promote generalized prosociality? Religion, Brain & Behavior, 4(1), 3–19.

Reddish, P., Fischer, R., & Bulbulia, J. (2013). Let’s dance together: Synchrony, shared intentionality and cooperation. PloS One, 8(8), e71182.

Repp, B. H., & Su, Y.-H. (2013). Sensorimotor synchronization: a review of recent research (2006-2012). Psychonomic Bulletin & Review, 20(3), 403–52.

Repp, B. H., & Penel, A. (2002). Auditory dominance in temporal processing: new evidence from synchronization with simultaneous visual and auditory sequences. Journal of Experimental Psychology: Human Perception and Performance, 28(5), 1085–1099.

Repp, B. H., & Penel, A. (2004). Rhythmic movement is attracted more strongly to auditory than to visual rhythms. Psychological Research, 68, 252–270.

Reuven-Magril, O., Dar, R., & Liberman, N. (2008). Illusion of control and behavioral control attempts in obsessive-compulsive disorder. Journal of Abnormal Psychology, 117(2), 334–41.

Richardson, M. J., Dale, R., & Marsh, K. (2014). Complex dynamical systems in social and personality psychology: Theory, modeling, and analysis. In H. Reis & C. Judd (Eds.),

- 135 -

Handbook of Research Methods in Social and Personality Psychology (pp. 253–282). New York: Cambridge University Press.

Richardson, M. J., Marsh, K. L., Isenhower, R. W., Goodman, J. R. L., & Schmidt, R. C. (2007). Rocking together: dynamics of intentional and unintentional interpersonal coordination. Human Movement Science, 26(6), 867–91.

Richerson, P. J., & Boyd, R. (2005). Not by Genes Alone: How Culture Transformed Human Evolution. Chicago, London: The University of Chicago Press.

Ridley, M. (2003). Nature via Nurture: Genes, Experience, and What Makes Us Human. New Jersey: Harper Collins Publishers.

Rizzolatti, G., Fadiga, L., Fogassi, L., & Gallese, V. (1999). Resonance behaviors and mirror neurons. Archives Italiennes de Biologie, 137, 85-100.

Rodgers, J., Riby, D. M., Janes, E., Connolly, B., & McConachie, H. (2012). Anxiety and repetitive behaviours in autism spectrum disorders and williams syndrome: a cross- syndrome comparison. Journal of Autism and Developmental Disorders, 42(2), 175–80.

Rudski, J. M., & Edwards, A. (2007). Malinowski goes to college: factors influencing students’ use of ritual and superstition. The Journal of General Psychology, 134(4), 389–403.

Saroglou, V., Pichon, I., Trompette, L., Verschueren, M., & Dernelle, R. (2005). Prosocial behavior and religion: New evidence based on projective measures and peer ratings. Journal for the Scientific Study of ReligionStudy of Religion, 44(3), 323–348.

Schaffner, K. (1967). Approaches to reduction. Philosophy of Science, 34(2), 137–147.

Schauer, P. (2011). Quantifying the importance of motifs on attic figure-painted pottery. In E. Slingerland & M. Collard (Eds.), Creating Consilience: Integrating the Sciences and the Humanities. Oxford: Oxford University Press.

Schellenberg, E. G. (2005). Music and cognitive abilities. Current Directions in Psychological Science, 14(6), 317–320.

Schellenberg, E. G., Nakata, T., Hunter, P. G., & Tamoto, S. (2007). Exposure to music and cognitive performance: tests of children and adults. Psychology of Music, 35(1), 5–19.

Schjoedt, U. (2009). The religious brain: A general introduction to the experimental

- 136 -

neuroscience of religion. Method & Theory in the Study of Religion, 21(3), 310–339.

Schjoedt, U., Stødkilde-Jørgensen, 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), 199–207.

Schlenker, B., & Pontari, B. (2000). The strategic control of information: Impression management and self-presentation in daily life. In Perspectives on Self and Identity (pp. 199–232).

Schmidt, R. C., Richardson, M. J., Arsenault, C., & Galantucci, B. (2007). Visual tracking and entrainment to an environmental rhythm. Journal of Experimental Psychology. Human Perception and Performance, 33(4), 860–70.

Schubotz, R. I., Friederici, a D., & von Cramon, D. Y. (2000). Time perception and motor timing: a common cortical and subcortical basis revealed by fMRI. NeuroImage, 11(1), 1–12.

Sebanz, N., Bekkering, H., & Knoblich, G. (2006). Joint action: bodies and minds moving together. Trends in Cognitive Sciences, 10(2), 70–6.

Sebanz, N., Knoblich, G., & Prinz, W. (2005). How two share a task: corepresenting stimulus-response mappings. Journal of Experimental Psychology: Human Perception and Performance, 31(6), 1234–1246.

Seidel, A., & Prinz, J. (2013). Sound morality: Irritating and icky noises amplify judgments in divergent moral domains. Cognition, 127(1), 1–5.

Serruya, D., & Eilam, D. (1996). Stereotypies, compulsions, and normal behavior in the context of motor routines in the rock hyrax ( Procavia capensis ). Psychobiology, 24(3), 235–246.

Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences, 17(11), 565–573.

Shariff, A. F., & Norenzayan, A. (2007). God is watching you: Priming god concepts increases prosocial behavior in an anonymous economic game. Psychological Science, 18(9), 803–809.

Shariff, A. F., Willard, A. K., Andersen, T., & Norenzayan, A. (2016). Religious priming: A

- 137 -

meta-analysis with a focus on prosociality. Personality and Social Psychology Review, 20(1), 27–48.

Shaw, D. J., Czekóová, K., Chromec, J., Mareček, R., & Brázdil, M. (2013). Copying you copying me: interpersonal motor co-ordination influences automatic imitation. PloS One, 8(12), e84820.

Shockley, K., Butwill, M., Zbilut, J. P., & Webber, C. L. (2002). Cross recurrence quantification of coupled oscillators. Physics Letters A, 305, 59–69.

Shweder, R. (2011). The metaphysical realities of the unphysical sciences: Or why vertical integration seems unrealistic to ontological pluralists. In E. Slingerland & M. Collard (Eds.), Creating Consilience: Integrating the Sciences and the Humanities. Oxford: Oxford University Press.

Shweder, R., & Sullivan, M. A. (1993). Cultural psychology: Who needs it? Annual Review of Psychology, 44(1), 497–523.

Sinclair, R., Lovsin, T., & Moore, S. (2007). Mood state, issue involvement, and argument strength on responses to persuasive appeals. Psychological Reports, 101, 739–753.

Singer, T., Seymour, B., O’Doherty, J. P., Stephan, K. E., Dolan, R. J., & Frith, C. D. (2006). Empathic neural responses are modulated by the perceived fairness of others. Nature, 439(7075), 466–469.

Slingerland, E. (2008a). What Science Offers the Humanities: Inegrating Body and Culture. Cambridge: Cambridge University Press.

Slingerland, E. (2008b). Who’s afraid of reductionism? The study of religion in the age of cognitive science. Journal of the American Academy of Religion, 76(2), 375–411.

Slingerland, E. (2011). Mind-body dualism and the two cultures. In E. Slingerland & M. Collard (Eds.), Creating Consilience: Integrating the Sciences and the Humanities. Oxford: Oxford University Press.

Slingerland, E., & Bulbulia, J. (2011). Evolutionary science and the study of religion. Religion, 41(3), 307–328.

Slingerland, E., & Collard, M. (2011). Creating consilience: Toward a second wave. In E. Slingerland & M. Collard (Eds.), Creating Consilience: Integrating the Sciences and the

- 138 -

Humanities. Oxford: Oxford University Press.

Smith, W. R. (1898/1995). Lectures on the Religion of the Semites. Sheffield: Sheffield Academic Press Ltd.

Smithson, M., & Verkuilen, J. (2006). A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychological Methods, 11(1), 54– 71.

Snow, C. P. (1961). The Two Cultures and the Scientific Revolution. New York: Cambridge University Press.

Sorensen, J. (2005). Religion in mind: A eeview article of the cognitive science of religion. Numen, 52(4), 465–494.

Sosis, R. (2007). Psalms for safety. Current Anthropology, 48(6), 903–911.

Sosis, R. (2009). The adaptationist-byproduct sebate on the evolution of religion: Five misunderstandings of the adaptationist program. Journal of Cognition and Culture, 9(3), 315–332.

Sosis, R., & Bulbulia, J. (2011). The behavioral ecology of religion: the benefits and costs of one evolutionary approach. Religion, 41(3), 341–362.

Sosis, R., & Handwerker, W. P. (2011). Psalms and coping with uncertainty: Religious Israeli women’s responses to the 2006 Lebanon war. American Anthropologist, 113(1), 40–55.

Sosis, R., & Ruffle, B. J. (2003). Religious ritual and cooperation: Testing for a relationship on Israeli religious and secular Kibbutzim. Current Anthropology, 44(5), 713–722.

Srull, T. K., & Wyer, R. S. (1979). The role of category accessibility in the interpretation of information about persons: Some determinants and implications. Journal of Personality and Social Psychology, 37(10), 1660–1672.

Stark, R. (1997). Bringing Theory Back in. In L. Young (Ed.), Rational Choice Theory and Religion. New York: Routledge.

Stasinopoulos, D., & Rigby, R. (2007). Generalized additive models for location scale and shape (GAMLSS) in R. Journal of Statistical Software, 23(7), 1–46.

Stegmueller, D., Scheepers, P., Roßteutscher, S., & De Jong, E. (2012). Support for redistribution in western Europe: Assessing the role of religion. European Sociological

- 139 -

Review, 28(4), 482–497.

Stenner, W. (1960). On Aboriginal religion: III. Symbolism in the higher rites. Oceaniea Publications, 31(2), 100–120.

Stokes, C. E., & Regnerus, M. D. (2009). When faith divides family: Religious discord and adolescent reports of parent-child relations. Social Science Research, 38(1), 155–167.

Strassmann, B. I., & Gillespie, B. (2003). How to measure reproductive success? American Journal of Human Biology, 15(3), 361–369.

Strogatz, S. H. (1994). Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering. Reading, Massachusetts: Perseus Book Program.

Styns, F., van Noorden, L., Moelants, D., & Leman, M. (2007). Walking on music. Human Movement Science, 26(5), 769–85.

Taira, T. (2013). Making space for discursive study in religious studies. Religion, 43(1, SI), 26–45.

Takens, F. (1981). Detecting strange attractors in turbulence. In Dynamical Systems and Turbulence, Warwick 1980 (Vol. 1980, pp. 366–381). Berlin, Heidelberg: Springer.

Talmont-Kaminski, K. (2013). Religion as Magical Ideology: How the Supernatural Reflects Rationality. Durham: Acumen.

Taves, A. (2010). No field ss an island: Fostering collaboration between the academic study of religion and the sciences. Method & Theory in the Study of Religion, 22(2), 170–188.

Taves, A. (2011). Religious Experience Reconsidered: A Building-Block Approach to the Study of Religion and Other Special Things. Princeton, Oxford: Princeton University Press.

Tennie, C., Call, J., & Tomasello, M. (2009). Ratcheting up the ratchet: on the evolution of cumulative culture. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(1528), 2405–2415.

Thompson, W. F., Schellenberg, E. G., & Husain, G. (2001). Arousal, mood, and the Mozart effect. Psychological Science, 12(3), 248–251.

Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift Für Tierpsychologie, 20, 410–433.

- 140 -

Tinbergen, N. (1968). On war and peace in animals and man. Science, 160(3835), 1411– 1418.

Tomasello, M. (2009). Why We Cooperate. Cambridge: MA: MIT Press.

Tooby, J., & Cosmides, L. (1992). The psychological foundations of culture. In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The Adapted Mind : Evolutionary Psychology and the Generation of Culture (pp. 19–136). Oxford: Oxford University Press.

Trivers, R. L. (1971). The evolution of reciprocal altruism. Quarterly Review of Biology, 46(1), 35–57.

Tversky, A., & Kahneman, D. (1982). Judgments of and by representativeness. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment Under Uncertainty: Heuristics and Biases. Cambridge, UK: Cambridge University Press.

Tylor, E. (1871). Primitive Culture: Researches into the Development of Mythology, Philosophy, Religion, Art, and Custom. London: John Murray.

Ursin, H., & Eriksen, H. R. (2010). Cognitive activation theory of stress (CATS). Neuroscience and Biobehavioral Reviews, 34, 877–881.

US Federal Bureau of Prisons. (2015). Freedom of Information Act request number 2015- 06498.

Valdesolo, P., & Desteno, D. (2011). Synchrony and the social tuning of compassion. Emotion, 11(2), 262–6.

Valdesolo, P., Ouyang, J., & DeSteno, D. (2010). The rhythm of joint action: Synchrony promotes cooperative ability. Journal of Experimental Social Psychology, 46(4), 693– 695. van der Wel, R. P. R. D., Knoblich, G., & Sebanz, N. (2011). Let the force be with us: dyads exploit haptic coupling for coordination. Journal of Experimental Psychology. Human Perception and Performance, 37(5), 1420–31.

Van Dyck, E., Moelants, D., Demey, M., Deweppe, A., Coussement, P., & Leman, M. (2013). The impact of the bass drum on human dance movement. Music Perception, 30(4), 349–360. van Elk, M., Matzke, D., Gronau, Q. F., Guan, M., Vandekerckhove, J., & Wagenmakers, E.-

- 141 -

J. (2015). Meta-analyses are no substitute for registered replications: a skeptical perspective on religious priming. Frontiers in Psychology, 6(September), 1365.

Varlet, M., Marin, L., Issartel, J., Schmidt, R. C., & Bardy, B. G. (2012). Continuity of visual and auditory rhythms influences sensorimotor coordination. PloS One, 7(9), e44082.

Vohs, K. D., Wang, Y., Gino, F., & Norton, M. I. (2013). Rituals enhance consumption. Psychological Science, 24(9), 1714–1721. von Stuckrad, K. (2013). Discursive study of religion: Approaches, definitions, implications. Method & Theory in the Study of Religion, 25(1), 5–25.

Webber, C. L., & Zbilut, J. P. (2005). Recurrence quantification analysis of nonlinear dynamical systems. In M. Riley & G. Van Orden (Eds.), Tutorials in contemporary nonlinear methods for the behavioral sciences (pp. 26–94).

Weber, M. (1920/2002). The Protestant Ethic and the Spirit of Capitalism. Max Weber: Essays in sociology. Los Angeles: Roxbury Publishing Company.

Weber, M. (1919/2008). Max Weber’s Complete Writings on Academic and Political Vocations. (J. Dreijmnis, Ed.). New York: Algora Publishing.

Webster, G. D., & Weir, C. G. (2005). Emotional responses to music: Interactive effects of mode, texture, and tempo. Motivation and Emotion, 29(1), 19–39.

Weinstein, D., Launay, J., Pearce, E., Dunbar, R. I. M., & Stewart, L. (2015). Singing and social bonding: Changes in connectivity and pain threshold as a function of group size. Evolution and Human Behavior, 37(2), 152–158.

Wenke, D., Atmaca, S., Holländer, A., Liepelt, R., Baess, P., & Prinz, W. (2011). What is shared in joint action? Issues of co-representation, response conflict, and agent identification. Review of Philosophy and Psychology, 2(2), 147–172.

Wenzlaff, R., & Wegner, D. (2000). Thought suppression. Annual Review of Psychology, 51, 59–91.

West, B., Welch, K., & Galecki, A. (2007). Linear Mixed Models: A Practical Guide Using Statistical Software. Boca Razon London New York: Chapman & Hall/CRC.

Whitehouse, H. (2005). The cognitive foundations of religiosity. In R. N. McCauley (ed.) Mind and Religion: Psychological and Cognitive Foundations of Religiosity. (pp. 207–

- 142 -

232). Walnut Creek, CA: AltaMira Press.

Whitehouse, H. (2008). Cognitive evolution and religion: Explaining recurrent geatures of religion. Issues in Ethnology and Anthropology, 3(3), 35–47.

Whitehouse, H. (2011). Whence and whither sociocultural anthropology. In E. Slingerland & M. Collard (Eds.), Creating Consilience: Integrating the Sciences and the Humanities. Oxford: Oxford University Press.

Whiten, A., & Erdal, D. (2012). The human socio-cognitive niche and its evolutionary origins. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 367(1599), 2119–29.

Wildman, W. J., & McNamara, P. (2008). Challenges facing the neurological study of religious behavior, belief, and rxperience. Method & Theory in the Study of Religion, 20(3), 212–242.

Wildman, W. J., Sosis, R., & Mcnamara, P. (2012). Reductionism in the scientific study of religion. Religion, Brain & Behavior, 1(3), 169–172.

Wilson, E. O. (1998). Conscielience: The Unity of Knowledge. New York: Vintage Books.

Wiltermuth, S. S., & Heath, C. (2009). Synchrony and cooperation. Psychological Science, 20(1), 1–5.

Winkler, I., Háden, G. P., Ladinig, O., Sziller, I., & Honing, H. (2009). Newborn infants detect the beat in music. Proceedings of the National Academy of Sciences of the United States of America, 106(7), 2468–71.

Xygalatas, D. (2012). The Burning Saints : Cognition and Culture in the Fire-walking Rituals of the Anastenaria. London: Equinox.

Xygalatas, D. (2013). Effects of religious setting on cooperative behavior: a case study from Mauritius. Religion, Brain & Behavior, 3(2), 91–102.

Xygalatas, D., Kotherová, S., Maňo, P., Kundt, R., Cigán, J., Klocová, K., & Lang, M. (In press). The Random Allocation Game in Mauritius. Religion, Brain & Behavior.

Xygalatas, D., Kundtová Klocová, E., Cigán, J., Kundt, R., Maňo, P., Kotherová, S., … Kanovsky, M. (2015). Location, location, location: Effects of cross- religious primes on prosocial behaviour. International Journal for the Psychology of Religion. E-print before

- 143 -

publication.

Xygalatas, D., & Lang, M. (2017). Religion and prosociality. In N. Clements (Ed.), Macmillan Interdisciplinary Handbooks. Religion: Mental Religion. (pp. 119–133). Farmington Hills, MI: Macmillan Reference USA.

Xygalatas, D., Mitkidis, P., Fischer, R., Reddish, P., Skewes, J., Geertz, A. W., … Bulbulia, J. (2013). Extreme rituals promote prosociality. Psychological Science, 24(8), 1602–5.

Xygalatas, D., Schjødt, U., Konvalinka, I., Jegindø, E.-M. E., Roepstorff, A., & Bulbulia, J. (2013). Autobiographical memory in a fire-walking ritual. Journal of Cognition and Culture, 13, 1–16.

Zacks, J. M., Tversky, B., & Iyer, G. (2001). Perceiving, remembering, and communicating structure in events. Journal of Experimental Psychology. General, 130(1), 29–58.

Zbilut, J. P., Zaldivar-Comenges, J.-M., & Strozzi, F. (2002). Recurrence quantification based Liapunov exponents for monitoring divergence in experimental data. Physics Letters A, 297(3–4), 173–181.

Zentner, M., & Eerola, T. (2010). Rhythmic engagement with music in infancy. Proceedings of the National Academy of Sciences of the United States of America, 107(13), 5768–73.

Ziv, N. (2015). Music and compliance: Can good music make us do bad things? Psychology of Music.

Ziv, N., Hoftman, M., & Geyer, M. (2012). Music and moral judgment: The effect of background music on the evaluation of ads promoting unethical behavior. Psychology of Music, 40, 738–760.

Zor, R., Keren, H., Hermesh, H., Szechtman, H., Mort, J., & Eilam, D. (2009). Obsessive- compulsive disorder: a disorder of pessimal (non-functional) motor behavior. Acta Psychiatrica Scandinavica, 120(4), 288–98.

Zuckerman, P. (2007). Atheism: Contemporary eates and patterns. In M. Martin (Ed.), Cambridge Companion to Atheism. Cambridge: Cambridge University Press.

- 144 -

SUPPLEMENTARY MATERIAL

Music as a Sacred Cue? Effects of Religious Music on Moral Behavior

Lang, M., Mitkidis, P., Kundt, R., Nichols, A., Krajčíková, L., & Xygalatas, D.

Supplementary Material

1.1 List of Musical Stimuli Used in Pre-screening

Religious stimuli:

J. S. Bach - Ave Maria (Gounod’s interpretation)

Jan Zwart - Toccata Psalm 146

J. S. Bach - BWV 147 Jesu joy of man's desiring

J.S. Bach - BWV 29 We thank thee, God

Secular stimuli:

Max Richter - H In New England

P. I. Tchaikovsky - Romance for piano in F Minor, Op. 5

Yann Tiersen - Comptine d'Un Autre Été

J. S. Bach - BWV 140 Sleepers Awake

1.2 Characteristics of Musical Excerpts Rated Pre-experiment:

We adapted the Positive Affect Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) to measure positive and negative emotional characteristics of our musical stimuli (see section 1.4). We used Principal Component Analysis (PCA) with oblique rotation (“oblimin”) to test if the selected characteristics load on positive (five items) and negative (four items) scales. The Bartlett’s test of sphericity indicated that all items were sufficiently inter-

- 145 -

correlated [χ2 (36) = 3489.81, p < .001] and the Kaiser-Meyer-Olkin (KMO) test revealed sampling adequacy (MSA = .73). The number of factors was set to two because we expected to find one positive and one negative factor. Table S1 shows factor loadings. Both scales had sufficient Cronbach’s α (positivity = .79, negativity = .68).

In addition to the positive and negative elements, we also included ratings of stimuli’s holiness, tempo, and impact. The measure of holiness served to select a stimulus that would be associated with religion. The questions on the stimuli’s tempo (How fast/slow was the stimulus?) were highly negatively correlated (r = -.62, p < .001), so the question “Slow” was reversed and both questions were combined into a measure of tempo. Similarly, we combined the measures of stimuli’s deepness and strength (r = .51, p < .001) to assess stimuli’s impact on participants. This measure served as a control for possible musical aspects that might prime participants similarly as religion but without the moralizing aspect.

1.3 Characteristics of Musical Excerpts Rated Post-experiment:

The same procedure as in the pre-experiment ratings was used to construct the measures of positivity, negativity, holiness, tempo, and impact for the post-experiment ratings. To construct the positivity and negativity measures, we used PCA with oblique rotation (“oblimin”) and set the number of factors to two. The Bratlett’s test was significant [χ2 (36) = 687.91, p < .001] and the KMO score was adequate (MSA = .75). In contrast to the pre-experiment ratings, the item “Exciting” loaded highly on both factors. The items “Boring” and “Irritating” were strongly negatively loaded on the positivity factor and only weakly positively loaded on the negativity factor (see Table S2, Model 1). These results suggest that the scales obtained in our pre-experiment ratings would not be replicated in the post-experiment ratings. We also combined the ratings of “Fast” and “Slow” into a measure of tempo, however, their correlation was not significant (r = -.10, p = .125). The only scale that replicated the findings from the pre- experiment screening was the impact scale consisting of items “Deep” and “Strong” (r = .38, p < .001).

In order to better understand why the post-experimental results differed from our pre- screening, we conducted a second analysis without the white-noise stimulus. Since rating white noise on tempo and emotional characteristics is rather ambivalent, we expected that scales for the religious and secular stimuli should better approximate the pre-screening scales. The factor loadings and Cronbach’s α are displayed in Table S2, Model 2. Indeed, loadings for the

- 146 -

religious and secular stimuli resemble those of pre-screening with one exception: the item “Exciting” highly loaded on both factors. The correlation between “Slow” and “Fast” was significant (r = -.18, p = .026), as was the “Deep” and “Strong” correlation (r = .31, p < .001). Since these findings replicated the pre-screening findings, we created the measures of positivity (excluding the “Exciting” item; Cronbach’s α = .78), negativity (Cronbach’s α = .62), tempo, and impact. However, comparisons of the white-noise stimulus’ ratings with the religious and secular stimuli need to be interpreted with caution.

Table S1. Factor loadings of emotional characteristics used in pre-experimental ratings.

Factor Loadings

Positivity Negativity Interesting .85 Pleasant .84 Exciting .68 Relaxing .68 Happy .54 Distressing .85 Irritating .68 Boring .64 Sad .63

Cronbach’s α .79 .68

- 147 -

Table S2. Factor loadings of emotional characteristics used in post- experimental ratings. Model 1 shows loadings for all conditions, Model 2 includes only the religious and secular conditions.

Factor Loadings Factor Loadings Model 1 Model 2 Positivity Negativity Positivity Negativity Interesting .77 .79 Pleasant .76 .74 Exciting .46 .68 .54 .59 Relaxing .71 .71 Happy .68 .34 Distressing .78 .83 Irritating -.64 .27 .41 Boring -.65 .27 .63 Sad .73 .66

Cronbach’s α .78 .62 .78 .64

- 148 -

1.4 Post-experiment Questionnaire

1) Did you recognize the musical excerpt?

a. Yes b. No c. I am not sure

2) Who was the author?

3) Did you perceive the sound as… (6-point scale)

Profane o o o o o Holy

4) Please rate how much this song was…

Not at all/A little/Moderately/Quite a bit/Extremely

a. Sad b. Fast c. Boring d. Pleasant e. Happy f. Irritating g. Slow h. Exciting i. Deep j. Interesting k. Distressing l. Strong m. Relaxing

5) Are you a...

a. Very religious person b. Religious person c. Neither religious/nor antireligious d. Rather secular person e. Not religious at all

6) Are you part of a church/religious organization?

- 149 -

a. Yes b. No

7) If yes, what is your religion?

a. Christian b. Jewish c. Muslim d. Buddhist e. Hindu f. Other

8) How often do you usually attend religious services/ceremonies?

a. More than once per week b. Once per week c. Once per month d. Several times a year e. Once per year f. Not often at all g. Never

9) Are you

a. Female b. Male

10) Please specify the year when you were born

- 150 -

2. Supplementary Figures

Figure S1. A histogram depicting the distribution of correctly solved matrices in the pre- experiment testing. The significant “jump” from five to six marks the border of ethical behavior in our experiment.

Figure S2. A. The distribution of the number of claimed matrices across our sites. B. The distribution of the percentage of unethically claimed matrices after collapsing the data for participants that solved between two and five matrices into 0% of claimed matrices.

- 151 -

Lost in the Rhythm: Effects of Rhythm on Subsequent Interpersonal Coordination

Lang, M., Shaw, D.J., Reddish, P., Wallot, S., Mitkidis, P., & Xygalatas, D.

Supplementary Material

Figure S3. Illustration of the labyrinth task and main hand interactions. Green = dominant hands interaction; Yellow = non-dominant hands interaction. The close caption illustrates how interactions between these hand types capture most effectively the degree of interpersonal co-ordination.

- 152 -

Equation 1.

2 Variablet|i|j= β0 + β1 (Trial) + β2 (Trial ) + β3 (Condition = [Rhythmic vs. Control]) + β4

(Condition = [Rhythmic vs. Arrhythmic]) + β5 (Condition-by-Trial = [Rhythmic vs. Control])

2 + β6 (Condition-by-Trial = [Rhythmic vs. Arrhythmic]) + β7 (Condition-by-Trial = [Rhythmic

2 vs. Control]) + β8 (Condition-by-Trial = [Rhythmic vs. Arrhythmic]) + u0j + u1j (Trial) + εt|i|j

In the specification above, Variable represents the value of %DET for trial t and interaction of two hands (as mentioned above) i nested within dyad j. Trial represents the linear effect of learning and Trial2 the quadratic effect. To examine differences in movement kinematics between conditions, the Rhythmic condition was set as a reference category against which all other conditions were compared (β3-β8). uoj represents a random intercept and u1j a random slope for a dyad j; εt|i|j represents the unstructured variance of residuals across repeated trials.

Equation 2.

2 Variablet|i|j|k= β0 + β1 (Trial) + β2 (Trial ) + β3 (Condition = [Rhythmic vs. Control]) + β4

(Condition = [Rhythmic vs. Arrhythmic]) + β5 (Condition-by-Trial = [Rhythmic vs. Control])

2 + β6 (Condition-by-Trial = [Rhythmic vs. Arrhythmic]) + β7 (Condition-by-Trial = [Rhythmic

2 vs. Control]) + β8 (Condition-by-Trial = [Rhythmic vs. Arrhythmic]) + u0j + u0k + u1k (Trial)

+ εt|i|j|k

In the model above, Variable represents the value of the dependent variable for trial t and hand i in person j, nested within dyad k. β0 is intercept (i.e. the Rhythmic condition), Trial represents

2 the linear practice effect, and Trial the quadratic effect. β3-β8 are comparisons between

- 153 -

conditions and condition-by-trial interactions. u0j represents a random intercept for a person, u0k represents a random intercept for a dyad k, and u1k is a random linear slope. Again, εt|i|j|k represents the unstructured variance of residuals across repeated trials.

- 154 -

Effects of Anxiety on Spontaneous Ritualized Behavior

Lang, M., Krátký, J., Shaver, J.H., Jerotijević, D., & Xygalatas, D.

Supplementary Information

Figure S4. Photography of the decorative object used for the cleaning task. Related to Figure 8.

- 155 -

Questions about the decorative object What was the artist’s intention? How old do you think the object is? What aesthetic value does this object carry? What would you say about the shape of this object? What art genre does this object belong to? What materials were used to make this object? Do you know any other artists working in a similar genre?

Movement computation The raw hand-movement acceleration data recorded by ActiGraph sensors were collapsed across the three spatial dimensions, zero-centered, and rectified so as to obtain all movements independent of direction. Subsequently, the preprocessed data were band-pass filtered at minimal acceleration of 0.02 G and maximal acceleration of 2 G, and minimal distance between movements was set at 166 ms (movement frequency of 6 Hz). The number of movements was assessed as the number of acceleration peaks, and standard deviation of movement was computed as the squared variance of movement amplitudes corrected for individual mean movement amplitude.

RQA computation To compute RQA, the phase space of the analyzed signal was reconstructed by selecting an appropriate embedding dimension and a time-delay based on Takens’ theorem (Takens, 1981; Zbilut, Zaldivar-Comenges, & Strozzi, 2002). In the first step, time-delay of 4 sampling points was chosen with the help of average mutual information function. Next, we used the function of false nearest neighbors (Kennel, Brown, & Abarbanel, 1992) to estimate the embedding dimension which yielded dimensions from 3 to 6. We chose 6 embedding dimensions as a conservative estimate to ensure sufficient embedding for all participants (Webber & Zbilut, 2005). Finally, a minimal distance between two signals (radius) was chosen so as to obtain mean %RR < 2% and a minimal value of %RR > 0.1%, in order to eliminate possible noise from the data (Webber & Zbilut, 2005). Although the selection of the appropriate radius is based on the researcher’s decision, our simulation of different radius values displayed in Figure S5 documents that the difference in %RR between the HA and LA conditions remains stable even as the radius changes.

- 156 -

Figure S5. A simulation of different radius values on recurrence rate influencing density of a recurrence plot (shaded areas are SE). The dashed line designates the chosen value of radius. The exponential increase of %RR corresponds to a typical sigmoid distribution of percentage data. Related to Figure 9.

Data modeling To model dependent variables related to ritualized behavior, we used a bottom-up approach (West et al., 2007), were we first added random effects and then tested if fixed effects significantly improved the model (Akaike information criterion with ChiSq test (p < .05)). Random intercepts for each participant were included due to the clustered nature of our hand- movement acceleration data (hands were nested within participants). To fit models with negative binomial and beta distributions, we used the function gamlss (gamlss package; Stasinopoulos & Rigby, 2007). Gamlss allows to model the mean and dispersion of proportional data which are typically heteroscedastic and skewed (Cribari-Neto & Zeileis, 2010; Stasinopoulos & Rigby, 2007).

- 157 -

In the model specifications, y represents dependent variables; g is logit link function; h represents hand nested within an individual; u represents random intercept for each individual; ε represents error term; 휙 represents dispersion parameter.

The final models were specified as follows:

1. Self-reported anxiety 2 푦 = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) + 휀 ~푁표푟푚푎푙(휇, 휎 )

2. Heart rate increase

푦 = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦)

+ 훽2(푃푒푟𝑖표푑 = 퐵푎푠푒푙𝑖푛푒 푣푠. 푃푟푒푝푎푟푎푡𝑖표푛) + 훽3(퐶표푛푑𝑖푡𝑖표푛 ∗ 푃푒푟𝑖표푑) + 휀 ~푁표푟푚푎푙(휇, 휎2)

3. Time spent cleaning

Model 1: 푦 = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) + 휀 ~푁표푟푚푎푙(휇, 휎2) 2 Model 2: 푦 = 훽0 + 훽1(푆푒푙푓 − 푟푒푝표푟푡푒푑 푎푛푥𝑖푒푡푦) + 휀 ~푁표푟푚푎푙(휇, 휎 ) 2 Model 3: 푦 = 훽0 + 훽1(푀푒푎푛 ℎ푒푎푟푡 푟푎푡푒 𝑖푛푐푟푒푎푠푒) + 휀 ~푁표푟푚푎푙(휇, 휎 )

4. Number of movements

Model 1: 푔(푦) = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) + 푢0 + 휀 ~푁푒푔퐵𝑖푛(푟, 푝)

Model 2: 푔(푦) = 훽0 + 훽1(푆푒푙푓 − 푟푒푝표푟푡푒푑 푎푛푥𝑖푒푡푦) + 푢0 + 휀 ~푁푒푔퐵𝑖푛(푟, 푝)

Model 3: 푔(푦) = 훽0 + 훽1(푀푒푎푛 ℎ푒푎푟푡 푟푎푡푒 𝑖푛푐푟푒푎푠푒) + 푢0 + 휀 ~푁푒푔퐵𝑖푛(푟, 푝)

5. Recurrence rate

Model 1: 푔(푦) = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) +

휙1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) + 푢0 + 휀 ~퐵푒푡푎 (휇푖, 휙푖)

Model 2: 푔(푦) = 훽0 + 훽1(푆푒푙푓 − 푟푒푝표푟푡푒푑 푎푛푥𝑖푒푡푦) + 휙1(푆푒푙푓 − 푟푒푝표푟푡푒푑 푎푛푥𝑖푒푡푦) +

푢0 + 휀 ~퐵푒푡푎 (휇푖, 휙푖)

- 158 -

Model 3: 푔(푦) = 훽0 + 훽1(푀푒푎푛 ℎ푒푎푟푡 푟푎푡푒 𝑖푛푐푟푒푎푠푒) +

휙1(푀푒푎푛 ℎ푒푎푟푡 푟푎푡푒 𝑖푛푐푟푒푎푠푒) + 푢0 + 휀 ~퐵푒푡푎 (휇푖, 휙푖)

6. Determinism

Model 1: 푔(푦) = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) + 푢0 +

휀 ~퐵푒푡푎 (휇푖, 휙푖)

Model 2: 푔(푦) = 훽0 + 훽1(푆푒푙푓 − 푟푒푝표푟푡푒푑 푎푛푥𝑖푒푡푦) + 푢0 + 휀 ~퐵푒푡푎 (휇푖, 휙푖)

Model 3: 푔(푦) = 훽0 + 훽1(푀푒푎푛 ℎ푒푎푟푡 푟푎푡푒 𝑖푛푐푟푒푎푠푒) + 푢0 + 휀 ~퐵푒푡푎 (휇푖, 휙푖)

7. Standard deviation of movement

Model 1: 푦 = 훽0 + 훽1(퐶표푛푑𝑖푡𝑖표푛 = 퐿표푤 퐴푛푥𝑖푒푡푦 푣푠. 퐻𝑖푔ℎ 퐴푛푥𝑖푒푡푦) + 푢0 + 휀 ~푁표푟푚푎푙(휇, 휎2) 2 Model 2: 푦 = 훽0 + 훽1(푆푒푙푓 − 푟푒푝표푟푡푒푑 푎푛푥𝑖푒푡푦) + 푢0 + 휀 ~푁표푟푚푎푙(휇, 휎 ) 2 Model 3: 푦 = 훽0 + 훽1(푀푒푎푛 ℎ푒푎푟푡 푟푎푡푒 𝑖푛푐푟푒푎푠푒) + 푢0 + 휀 ~푁표푟푚푎푙(휇, 휎 )

- 159 -