Theory of Religion and Historical Research. A Critical Realist Perspective on the Study of Religion as an Empirical Discipline

1.1 Separation of Historical Research and Theoretical Discourses in the Study of Religion

Ironically, the negligence of the history of religions in theoretical discussions on the aims, methods, and competences of the study of religion is paralleled by a significant increase in historical research. Historical knowledge about religions in such areas as China, Japan, South Asia, or the Middle East is much more detailed and comprehensive today than it was in the days the discipline went under the name “history of religions”. Furthermore, most historians of religion have turned away from mainly studying religious texts and pay more attention to the entanglement of religious beliefs, practices, and institutions with other aspects of culture and society. Research in the history of religions, therefore, demands training and competence in the history of particular areas. Historians, in many cases, stand aloof from discussions centring around the concept of religion and its Western genealogy or explaining religion by cognitive algorithms without paying attention to the full range of things studied in the history of religions. Conversely, theoretical discourses in the study of religion largely ignore historical studies and fail to relate theory formation to historical research.

Theoretical debates leave historians of religion in a somewhat uneasy position because what the latter do is often considered by theorists a somewhat naïve undertaking, which is criticised from various angles. From one side, historical studies are accused of being an unorganised compilation of data that resembles “butterfly collecting” due to “a lack of any sort of guiding theoretical framework” (Slingerland and Bulbulia 2011, 323). From another side, the history of religions has to face the charge of being uncritical in ignoring that “religion” is a construct of modern Western thinking and “that the production of data in books and journals about religion and religions [...] has as their most important function the maintenance and reproduction of a myth” (Fitzgerald 2007, 7).

The quoted critics come from different directions: the Cognitive Science of Religion and the so-called Critical Religion^[2] approach. Historians of religion might console themselves with the thought that both critics would equally criticise each other’s approach to the study of religion. The “critical” approaches on the one side, which include postmodern, postcolonial, and discursive studies of religion, and the cognitive science of religion on the other, completely disagree with each other about the ontological status of religion. The critical approaches usually highlight the historical contingency of the concept of religion as a modern Western construction, which is expressed in Jonathan Z. Smith’s famous but hyperbolic dictum: “Religion has no independent existence apart from the academy” (1982, XI). In this view, religion is a theoretical concept constructed by scholars. For the cognitive science of religion, on the other side, religion is not merely a concept but something that can be empirically explored and explained by using methods and findings of the natural sciences. Accordingly, religion is supposed to exist independently from the concept, as expressed by Pascal Boyer: “That people do not have a special term for religion does not mean they actually have no religion. In many places people have no word for ‘syntax’ but their language has a syntax all the same” (2001, 9). Like language, religion is considered innate equipment of the human species (Bulbulia 2005).

In recent decades, various forms of both critical studies of religion and the cognitive science of religion were among the most influential new theoretical approaches in the study of religion. Given the fundamental differences in conceptualising religion that exist between these two approaches, discussions on theory and method in the study of religion became much occupied with clarifying, defining, criticising, or defending the concept of religion. Against this backdrop, the history of religions did not receive much attention in reflections on the aims and methods of the discipline. Historical research remained a strangely blind spot in metatheoretical discourses. As Armin Geertz remarked at the end of the last century, the “study of the history of religions stood in danger of dissolving into separate philologies with no common goal and no significant theoretical reflection” (1999, 447).

1.2 The Study of History as Empirical Research

Historians of religion understandably have issues with the view that the modern origin of the concept of religion and its construction in Western discourses justifies ignoring historical research on earlier periods and excluding it from theoretical considerations in the study of religion. While it is true that religion is a modern concept and some critiques, therefore, claim that it is a “modern Western invention” (cf. Taylor 1998, 7), what historians of religion study is not invented by them. They might interpret the imaginations, institutions, practices, or social structures that are reconstructed by historical research as being “religious”, and this, of course, is an interpretation against the backdrop of modern conceptualisations. However, this does not make the objects that are interpreted a modern invention.

As research in the history of religions has been widely neglected in theoretical discussions, a few remarks are appropriate to clarify some points. One point concerns the role of subjectivity in writing history and the possibility of historical knowledge. Historians are divided over the question of whether historiography can and does produce knowledge about the past that is more than the subjective imaginations of the historiographer.^[3] There is no doubt that in interpreting historical sources and trying to reconstruct historical facts from them, much subjectivity on the part of the researcher is involved. The skills and capacities of historians belong to them as acting subjects, and even if they succeeded in designing algorithms enabling computers to do the work of historians, there would be the subjectivity of the designers. There should, however, be no illusion: subjectivity is unavoidable, not just in historical research; it is not even confined to research in the humanities and social sciences. All perceptions and interpretations of perceptions are necessarily subjective. The natural sciences are no exception because it is human beings who design research strategies to produce empirical data, and it is human beings who finally interpret the data.

The question, therefore, is not so much whether the production of knowledge about the past involves more subjectivity than the production of knowledge about the very recent past, which is called the present time; the question instead is whether we can produce any scientific knowledge about cultural and social realities, that is, the world produced and inhabited by humans and the ways humans live in it. Since in English-speaking academia the term “science” usually is reserved for the natural sciences, historical research and, for that, research in the study of religion by definition cannot produce scientific knowledge.^[4] If it is not scientific knowledge, how else could we characterise the knowledge we aim to provide in the study of religion? To avoid futile discussions about terminology, we better ask what we do expect of the study of religion to deliver. The expectations are diverse, given the diversity of approaches in the discipline. What is more, there is no means to decide rationally what should be the aim of the study of religion because it is a normative question whose answer depends on the preferences of scholars. There might be a consensus between some of them, but since there is no authority or magisterium, there always will be different opinions. Each scholar has to make a choice between controversial approaches and, if possible, to know the reasons why she or he prefers a particular understanding of the discipline rather than another.

My own choice is to expect the study of religion to be an empirical discipline, and I will, in the following, confine my argument to this understanding and ignore different preferences that, for instance, see the study of religion as a domain of philosophy. Empirical disciplines aim to produce true knowledge about the world. The concept of truth should not be controversial, because the alternative would be aiming to provide knowledge that is not true. However, we have to be aware that there is a difference between the concept of truth and the ascertainment of truth. There are no objective and indisputable criteria for deciding whether a statement is true or not because the criteria for supporting truth claims are defined and agreed upon in a discursive community. In the present context, the discursive community is scholars who understand the study of religion and the study of history as empirical disciplines. Empirical disciplines presuppose that there exists a reality that is independent of the researchers who attempt to gain knowledge about it. They accordingly demand that the knowledge they produce in some way refers to this reality and not just to itself, that is, to academic discourses. Furthermore, they reflect on the methods of how to arrive at statements of fact, and agree, at least in principle, on the requirements for a statement of fact to be well- or ill-founded.

The last-mentioned point, in particular, is essential. In empirical disciplines, statements of fact can be criticised with empirical arguments according to the methodological criteria agreed upon in the discursive community. In historical research, this marks the difference between fact and fiction. Unlike the authors of fictional writing, historians have to observe intersubjectively accepted methodical rules to provide evidence for their statements of fact. They cannot just present a historical narrative without providing empirical and logical arguments (Lorenz 2004, 57). Correspondingly, historical accounts can be publicly assessed, doubted, and critiqued with empirical and logical arguments, which is not possible for fictional narratives.

Historical knowledge is fallible, needless to say, but this is true for any knowledge. Because scientific statements are fallible, it is imperative that they are open to critique. The possibility to critique statements with empirical arguments presupposes the idea of truth – for otherwise, all statements would be equally valid. Furthermore, empirical arguments necessarily imply referring to facts that exist independently of the academic discourses.

At this point, it suffices to remark that even the postmodern critique of historiography, and the history of religions in particular, presupposes the concept of truth. To criticise the history of religions for constructing distorted accounts of social realities implies that these accounts are considered false and that it is possible to arrive at a less distorted view. If we abandoned the idea of truth, we would have to accept all historical narratives, which consequently would make it self-contradictory to criticise the narratives given by historians of religions.

1.3 Historical Research and the Concept of Religion

As has been remarked, while the interpretation of pre-modern realities as being religious is doubtlessly a modern interpretation, this does not affect the existence of historical facts that are interpreted in this way. However, it can be argued that applying the concept of religion to pre-modern realities distorts our perception by imposing modern classifications that do not reflect the way things were understood in these societies (Barton and Boyarin 2016; Nongbri 2008, 2013).

To understand how social reality is understood in a given society certainly is a legitimate aim in the study of religion, though not the only one. Besides this “emic” perspective, it also is legitimate to adopt an “etic” perspective,^[5] that is, interpreting social realities in the context of our theories and conceptions. Theory formation in the study of religion is more than reconstructing and understanding emic conceptions. Nevertheless, for historians, it is an intriguing question whether it is only in modern societies that a distinction is made between religious and non-religious things. It is an empirical question that can only be answered by historical research.

We certainly cannot expect to find religion as a concept in historical contexts where it was unknown. However, it is possible to scrutinise pre-modern classifications and terminologies to find out how they deal with things that, from a modern perspective, might be classified as being religions or religious. Some historians of religion have worked in this direction. Their findings indicate that in many pre-modern societies, distinctions similar to the modern one between religious and non-religious affairs were not entirely unknown.^[6]

In the case of Europe, the word religio was common since antiquity, although before early modern times it usually was not used as a generic term.^[7] However, as Giovanni Casadio has shown based on literary evidence, the word religio in Roman antiquity also was used in the plural, religiones, in contrasting Roman religion with the religions of other peoples (Casadio 2010, 315–319). Relating to medieval Europe, Peter Biller has drawn attention to the fact that in the 13^th century, Roger Bacon put forward a classification of and comparisons between various great secte of the world (Biller 1985, 368). Both examples show that well before modern times and colonial encounters there was an awareness of different religions even if it might not have been the subject of everyday discourses.

Concerning pre-modern Asian societies, several studies come to similar results. Based on textual research, Robert Campany has argued that in medieval China there was a tendency “to refer nominally to entities that seem to correspond roughly to the ones named ‘religions’ in Western discourse” (Campany 2003, 311). Also dealing with medieval China is Max Deeg’s (2013) research on the identity construction of Buddhism, Daoism, Confucianism, and Nestorianism. He highlights that apologetic discourses not only indicate that there was an awareness of these traditions as being different from one another but also show that they mutually conceived each other as belonging to the same class. Christoph Kleine (2010, 2013 a, 2013b) has extensively discussed the differentiation of religious and political functions, and their conceptual distinction in medieval Japan, arguing that the functional differentiation of religion is certainly not a purely modern Western phenomenon. Karénina Kollmar-Paulenz (2013) has demonstrated that the Mongolian terms signifying religion – in particular the word šasin – underwent semantic changes similar to the word “religion” in early modern Europe: while it first referred only to Mongolian Buddhism, it later on came to be used as a comparative concept that also embraced shamanism, Islam, and finally Christianity.

A common observation of these historical studies is the fact that conceptualisations roughly corresponding to what modern scholars classify as religions can be found in Asian societies well before the impact of Western thought and classifications. It appears that a crucial factor promoting this process is apologetics and the awareness of multiple religious actors competing with each other in the same field. This is also supported by Oliver Freiberger’s (2013) analysis of the juxtaposition of Buddhist and other religious institutions in early Buddhist texts. Of course, the semantics of the terms applied do not precisely correspond to the modern term “religion”, which is outright impossible because this term has no unambiguous meaning. Equally important is the observation that emic discourses on what nowadays are called “religions” were as diverse, manifold, and unsettled as are modern discourses on religion. We must be aware that not only modern European concepts and classifications have a history and do change, but so does the order of knowledge in other societies, a point particularly stressed by Kollmar-Paulenz (2013, 185–188).

This brief review of some research made by historians of religion sufficiently shows that what modern scholars classify as religions in many pre-modern societies had been conceptualised as discernible units of social life, which usually went along with a tendency to reify these concepts as things that exist, having properties and histories. These findings debunk the thesis that it only is the “dust-covered scholars of religion” (Masuzawa 2015, 148) who have constructed such objects as “Buddhism” and other religions, and that all critical scholarship can do is to study the Western genealogy of these “ideological constructions” (Masuzawa 2015, 147). Modern Western thinkers were neither the only ones nor the first who have interpreted social realities. Historical ignorance certainly helps believe in the exceptionality of modern understandings of the world, but it is a faulty foundation of far-reaching theoretical claims. It engenders blindness for the realities beyond the scope of theoretical discourses that content themselves with the critique of other academic theories.

As an empirical discipline, the study of religion does not confine historical research to the analysis of discourses, be they modern or pre-modern; it also investigates things that discourses are about or that are not even conceptualised in the discourses under consideration. It is possible to disagree on the question of whether Buddhism or Christianity are theoretically constructed objects or not, but it can hardly be disputed that Buddhist and Christian monasteries, textual traditions, or rituals exist and existed in pre-modern times. They exist as social realities and, as such, can not only be the objects of empirical research but also the subject matter of theories.

In this part, I have shown that historical research is an indispensable element of the study of religion if the latter is understood as an empirical discipline. This is true even if it is argued that the concept of religion should not be applied to the interpretation of pre-modern social realities because it emerged only in modern times, for the argument evidently cannot be made without recourse to historical research. Given the unquestionable importance of historical research in the study of religion, it is remarkable that history has received little attention in recent theoretical and metatheoretical discussions. I have addressed some of the points that deserve consideration: the questions of whether the history of religions can be considered empirical research and whether it is justified to extend it to societies unaffected by modern discourses on religion. In the following part, I will discuss some ontological and epistemological preconditions of empirical research that are relevant to the study of religion.

2.1 Theory, Metatheory, and Methodology

Theoretical discourses in the study of religion have a double face: they include both theories of religion and theories about the study of religion. While theories of religion are attempts at explaining religion, theories about the study of religion are metatheoretical propositions and reflections on the aims, conditions, methods, and restrictions of studying religion. Metatheory cannot be separated from methodology where the methods of forming theories, conducting empirical research, and connecting theory formation with empirical research are discussed. Metatheory and methodology of an empirical discipline need to be founded on – or at least be informed by – principles discussed in the philosophy of science. Hence, there is a cascade of theoretical levels, with the philosophy of science as the most general theoretical foundation of empirical research, over more specific metatheories of particular disciplines and their methodological implications, down to theories about the subject matter of these disciplines and, finally, theories explaining social realities.

The reason why theoretical discourses often do not strictly separate different levels of theory in the study of religion probably is the fact that religion as the discipline’s subject matter is a highly disputed concept. As discussing methods of studying religion demands some kind of theoretical pre-understanding of what is being talked about, methodological arguments merge with theories of religion and attempts at defining it. The intricate entanglement of methodology and metatheory with theories of religion is quite apparent when we consider the different understandings of religion mentioned above: if religion is taken as a construct of scholars, the study of religion naturally demands other methods than when religion is regarded as a universal phenomenon grounded in the genetic evolution of the human species.

Discussing the relation between historical research and theory formation in the study of religion therefore cannot avoid addressing religion as the subject matter of theories as well. However, I will propose neither a theory nor a definition of religion but a methodological approach that circumvents the issue of defining religion. Such an approach goes against the predominant view that theories in the study of religion cannot avoid specifying what religion is. It is, therefore, necessary to explicate the methodological presuppositions on which my argument is based, which will be done in the next section.

2.2 Ontological and Epistemological Presuppositions

As has been shown above, the assumption that classifications distinguishing between religious and non-religious affairs took shape only in modern European discourses cannot be maintained because historians have found similar classifications in other societies. However, the fact that distinctions are made in discourses – whether in modern Europe or elsewhere – does not imply that they describe differences existing in the real world. That we distinguish between elves and imps does not allow for the conclusion that elves and imps exist as real entities that are different from each other. Applied to the study of religion, Timothy Fitzgerald makes a similar point: “My argument is that there is no essential difference between religious and non-religious domains, but they are imagined and represented as if they are essentially different, along the axis of binary either-or alternatives” (Fitzgerald 2013, 104, italics mine). This argument is a response to Kevin Schilbrack, who maintains that the social construction of the concept “religion” does not exclude “that ‘religion’ is descriptively and analytically useful, and it is useful because there really are religions that exist ‘out there’ in the world” (Schilbrack 2012, 98, italics mine).

The two positions reflect in a nutshell a fundamental ontological problem lurking in theoretical disputes in the study of religion: do religions and does religion exist, and if so, in which sense? This question necessitates clarifying one’s ontological presuppositions: what does it mean that something “exists” or is “real”?

Let me start with Schilbrack’s formulation that “religions exist ‘out there’ in the world”. Schilbrack explains his ontological assumptions in a later article, drawing on Searle’s social ontology and Bhaskar’s Critical Realism (Schilbrack 2017, 166–168): “out there” means “outside our theories”. The implication is that there exists a reality that is outside and independent of our research. This is a realist ontology.^[8] However, it is a critical realism that is fully aware of the fact that supposing the reality of the external world is not the same as claiming that this reality is accurately caught by our descriptions and explanations.

2.3 Descriptions, Data, and Theories

According to a critical realist ontology, the intransitive objects of empirical research exist independently of being observed, but they can only be talked about if they are described. Descriptions are data, that is, linguistically represented information about the external world. Descriptions are the first step in any empirical research and the precondition of explanations and theories that claim to contribute to knowledge about the external world. Only what has been described can be explained (Archer 1995, 20).

Because data are descriptions providing information, they are not “out there” to be collected like mushrooms but are produced by researchers. Thus, they belong to the transitive dimension of research. They depend on the researchers who provide descriptions, that is, on their interests, ontological assumptions, research questions, and the methods used to produce the data. As data are linguistic entities,^[15] they also depend on the language used, which provides a reservoir of terms and concepts that restrict and channel the way something is represented linguistically. Data, therefore, are profoundly affected by subjective factors pertaining to the researchers, the academic discourses they participate in, and the social environment in which their research takes place.

In this sense, “[d]ata are always already theory-laden” (Stausberg and Engler 2016, 67). However, we can analytically distinguish between data or descriptions and the interpretation of data. It is true that scholars of religion usually choose to work with data they have already interpreted as describing religious phenomena, which presupposes a theoretical concept of religion (Stausberg and Engler 2016, 67), but the production of data does not necessarily depend on such a theoretical interpretation. One can, for instance, describe the grand state rituals in early modern China and produce an enormous amount of data on them without interpreting these rituals as being religious.^[16] This does not preclude other scholars from using the data against the backdrop of their understanding of religion and asking questions or providing arguments that relate to religion. However, it only is when religion as a theoretical perspective comes in that particular data can be considered data “for religion”. The reliability of the data does not depend on how they are interpreted theoretically.

Though in practice, data are often presented in a way that already includes their interpretation, there is a difference between descriptions and interpretations. They differ, as it were, in the degree of subjectivity. It is possible that scholars fully agree on the reliability of certain data, let us say the number of people who in a survey have declared to believe in the existence of ghosts, but the same scholars might completely disagree on whether the data indicate that these interviewees and their belief in ghosts are religious. In principle, data can be confirmed or disproved by empirical evidence, whereas their interpretation rests on theoretical arguments.

This is not to say that data represent the reality objectively: far from it. Data are descriptions and, as such, belong to the transitive dimension of research, which, as has been made clear, depends on many subjective factors. However, in empirical research, at least if it is taken seriously, data are not purely subjective. They not only depend on the scholars who produce them; they also depend on what is being described. Data are the hinge that connects the transitive dimension of research with its intransitive dimension. They are theory-dependent as well as reality-dependent. Otherwise, they would be mere fictions, and there would be no difference between reliable and faked data.

Certainly, historical research, like research in all sciences, delivers data that describe reality in a very incomplete and selective way. However, there is no other way. The alternative would be to deny the possibility of empirical research. All we could do in the study of religion would be “thinking about religion”^[17], that is, create theories without reference to data. However sophisticated such theorising may be, it would just be “casual conversation”.^[18]

3.1 Empty Theories

The main cause of empty theories is the conflation of the transitive and intransitive aspects of knowledge or, to put it differently, the confusion between the subjects of theoretical discourses and the objects of empirical research. In the study of religion, the central subject of theoretical discourses is religion. The subject matter is defined, discussed, and explained in theories. It belongs to the transitive dimension of research; it is a linguistic entity, that is, a term and concept whose meaning depends on the explanations given by researchers. The pitfall of empty theories occurs when “religion” as the transitive subject of discourses is treated as an intransitive object of research, that is, if theories of religion pretend to explain something that exists independently of our theories (cf. Jensen 2009, 332). This results in an aporia because we cannot specify what the theories are about, except by referring to religion as their discursively constructed subject. It is not possible to escape the tautology of identifying religion as that which has before been defined as the meaning of the concept. If, however, the goal is to clarify “What kind of thing is religion?” (Stausberg and Engler 2016, 69), the answer cannot consist of explicating a term, but it demands to specify properties or conditions “for identifying something as ‘religion’” (Stausberg 2009, 3). As the identification of something as being religion presupposes “religion” as a pre-conceived category, it will only explain how this category is understood and how something corresponds to this understanding.

However, the type of theory considered here does not aim to explain a concept but to explain religion as an extra-linguistic phenomenon. As Gardiner and Engler (2010, 287) put it:

“Theories are necessarily theories of some phenomena. A theory is a theory of something to the extent that, at least [in] part, it aims at explaining that thing. A theory explains, at least in part, to the extent that it allows us to understand that phenomenon.”

The thing to be explained is religion or a religious phenomenon. Referring to the requirements of a theory of religion that have been postulated by Stausberg (2009, 3–6), the authors further explain:

“Insofar as a theory interprets and explains some phenomenon, demarcating or identifying that phenomenon comes first. Looking for the origins, functions, or structures of religion(s) presupposes prior criteria for identifying something as religious.”

As the criteria for identifying are prior to any “looking for” something, they must be based on a pre-conceived definition. According to Gardiner and Engler, on the basis of a definition “the theorist of religion selects the ‘data’ from which to theorize or delimits the proper ‘object’ of theoretical investigation” (Gardiner and Engler 2010, 287). There seems to be a certain uneasiness with the terms “data” and “object”, which is indicated by the scare quotes. In fact, the objects of theoretical investigation are the data, which belong to the transitive dimension of research because they are linguistic entities produced by the researchers. “Demarcating a phenomenon” thus amounts to the same as applying a concept or category to the interpretation of data.

It appears, therefore, that this form of theory cannot avoid the circle of first defining a concept, based on which certain data are selected to demarcate an object of theorising that necessarily meets the conditions defined by the theory. Because in this way religion cannot be found to be something other than or different from what it has been defined as before, this kind of theorising will never arrive at answering the question “What kind of thing is religion?” (Stausberg and Engler 2016, 69) but only illustrate a pre-conceived category by selecting data fitting the definition.

3.2 Avoiding Empty Theories: Religion as a Theoretical Perspective

While the method proposed by Gardiner and Engler cannot lead to a theory that explains religion as an extra-linguistic entity that can be identified by empirical research, it can be a step on the way to substantial explanations, provided we change the explanandum. Instead of explaining or illustrating the category of religion, we have to explain the data. The data are descriptions providing information about intransitive objects of the external reality, and it is the external world that explanations in empirical disciplines aim at.

As has been made clear above, data belong to the transitive dimension of knowledge; they have been produced by researchers to supply information about the reality that represents the intransitive dimension of knowledge. In the study of religion, the production and use of data are guided by theoretical discourses in the discipline, whose main subject is religion. Instead of being the explanandum, religion is a theoretical concept that shapes the perspective taken to narrow down the range of objects that theories in the study of religion attempt to explain.

Gardiner and Engler’s suggestion that explanatory theories start with selecting data based on a pre-conceived definition of religion is reasonable. However, if these data are intended to be used to explain religion, the result will be disenchanting. The approach admittedly produces a plethora of theories of religion, but they cannot be empirically validated; they fail to answer any of the questions that, according to Stausberg, theories have to address. Theorists who, for instance, subscribe to definitions that focus on the belief in supernatural agents will pay attention to other data than those who adopt Durkheim’s definition of religion. This engenders a quandary if their theories claim to explain religion, because different definitions will result in selecting different kinds of data fitting the divergent understandings of the concept. The theories would use the same word – “religion” – but give it different meanings. The ensuing problem is that the theories explain different things while both claim to explain religion.

The problem can be avoided if we keep in mind that the aim of theories in empirical disciplines is not to explain concepts but the intransitive reality. Religion is the theoretical perspective guiding the researcher to choose objects of research that are to be explained. The explanandum, accordingly, is the selected data. Take, for example, state rituals in contemporary China in which mythic ancestors are venerated (Seiwert 2016 a, 2016b). By taking religion as a theoretical perspective, we may select such rituals as objects of research. Some scholars might qualify the rituals as religious practices because they are related to superempirical realities,^[19] which the venerated ancestors doubtlessly are. Now, the value of a theory explaining the origins, functions, and structure of the practices would not be diminished by the fact that from the view of other researchers the state rituals concerned are not at all religious. They could argue that the Communist state ideology without any doubt is atheist, that is, secular, and the rituals are not religious because there is no reference to belief in the existence of supernatural agents.^[20] These researchers, therefore, would not select such rituals as objects of research. Nevertheless, this would not invalidate theories explaining them.

To summarise the argument: from a critical realist perspective, the aim of theories is to contribute to our understanding of the intransitive reality. To do this, we have to explain data that describe real things. Accordingly, the explanandum is the data, and not the category we have in mind when selecting data that demand an explanation. In the study of religion, the category of religion provides theoretical perspectives to address particular data. Thus, there necessarily are some theoretical assumptions before selecting the data, but these theoretical preconceptions are devoid of explanatory power except for explaining themselves. It is only after the objects to be explained have been described that theories explaining them can be conceived. This is why data produced by historical and other empirical research are the fundament without which substantial theories in the study of religion cannot be built.

3.3 Blind Theories

As has been noted above, theoretical blindness is a pitfall that cannot be wholly avoided because our perception of the intransitive reality is always shaped by subjective factors, and theoretical assumptions are one of them. Margaret Archer has discussed in detail the effects of ontological assumptions in the social sciences on the production of data and accordingly on explanations (Archer 1995, 16–30). Instead of discussing the problem theoretically, I will briefly illustrate the consequences of different ontologies by the examples of naturalist individualism and social holism.

If one subscribes to a narrow naturalist ontology that assumes that only material objects can have material effects, “[i]t makes no sense to assume that a nonmaterial object, such as a ‘culture‘ or a ‘cultural scheme,’ could have material effects” (Boyer 1994, 87). Accordingly, the idea of culture as something existing independently of individual cognitions – which are thought to be physical processes in the brains of individuals – is refuted (Boyer 2001, 35). On such suppositions, possible explanations of material effects, for instance, performing or participating in rituals, are restricted to cognitions of individuals and are blind for immaterial and social factors such as obligations, rights, relations of power, and social structures.

On the other hand, a holistic social ontology that in the Durkheimian sense maintains the existence of social facts with coercive power over individual behaviour will tend to be blind for the agency of individuals and their power to change the social reality. Accordingly, descriptions of social and cultural processes, as a rule, will ignore individual decisions and creativity. Furthermore, explanations will be restricted to social factors and exclude psychological or biological causes.

A favourable characterisation of theoretical blindness would be to admit that a theory cannot explain everything. There are always some things that remain ignored or unexplained by one theory but might be explained by another. Theories necessarily reduce the complexity of social reality. They do ignore facts that are irrelevant. Obviously, what is relevant or not is theory-dependent. However, if we only describe and select what according to our theories is worth being considered and explained, we arrive at a dead end. If we produce only data that are relevant to our theory, we will end up with a very restricted description of the social world. Moreover, the factors taken into account to provide explanations likewise will be restricted to those that have been reckoned on before.

To ask questions that demand explanations admittedly is contingent upon theoretical perspectives. However, it is equally conditioned by observations and descriptions. Their importance in the process of theory formation in the study of religion can be explained by comparing it with the development of the theory of evolution in biology, which is the paradigm of a powerful theory in the life sciences. Preceding Darwin’s theory of natural selection were decades and centuries of collecting and describing plants, animals, and other natural objects. The collections of the naturalists were not much concerned with theoretical questions, but they were indispensable for developing classifications of and theories about living organisms. Of course, observations and descriptions were not sufficient conditions for the emergence of the theory of common descent of different species; but they were necessary conditions. Without describing the variety of existing species, the question of their origin could not have been asked.

What I want to show is that theory formation without empirical data is not possible, and often the produced data are prior to theories that make them relevant. New observations and their descriptions can put existing theories into question, or they can be a reason to ask new questions. In the study of religion, descriptions are mostly provided by historical research. If one compares them derogatorily to “butterfly collecting” (Slingerland and Bulbulia 2011, 323), one ignores that without observing, collecting, and describing insects there would be no scientific knowledge at all about them, let alone theories about the evolution of their metamorphosis.

Since unexpected observations can trigger new questions, it is not a detriment to the study of religion if historians provide descriptions that at first sight seem to be irrelevant to any guiding theory. To illustrate this point, I refer to Johannes Prip-Møller’s description of Chinese Buddhist monasteries, a work utterly devoid of any theoretical ambition but full of descriptions, sketches, and photographs. Among others, the author describes details of an ordination ceremony in which a number of marks are branded into the top of the heads of the novices (Prip-Møller 1937, 317–320). This practice of inflicting pain on the novices is hardly mentioned elsewhere before and difficult to explain. However, once noticed, it can guide us to look more closely at occurrences of infliction of pain in rituals and attempt to understand and explain such behaviour. Moreover, it might be a cause for paying more attention to bodily pain when theorising about religion.^[21]

My point here is not that this case would be particularly important but rather the fact that new theoretical questions can be obtained from observations and descriptions that are not guided by theory; and that theories considering the effects of and causes for inflicting pain in religious contexts could not be conceived without previous descriptions of existing practices. Without data produced by the history of religions, theories in the study of religion cannot advance but only be self-referential. They would be blind to the complexity of human behaviour, including those forms that are characterised as religious.