Materialist dimension in addition to the three C’s

Although the three-sided model of the true, the beautiful, and the good is a classic account of what any social group explicitly and implicitly teaches, one aspect is still missing. To recognize the always-presupposed material reality of the people who constitute the social group, even when this reality has not been conceptualized by the group’s members, one should also include the contributions of their bodies, habits, physical culture, and social structures. To include this dimension mnemonically, one can add a “fourth C”, for community. Catherine Albanese (1981) may have been the first to propose the idea of adding this materialist dimension.

In an attempt to dislodge the focus on human subjectivity found in the three Cs, some have argued that the material dimension is the source of the others. They argue, in other words, that the cognitive, affective, and conative aspects of the members of a social group are not the causes but rather the effects of the group’s structured practices (e.g., Asad 1993: ch. 1–4; Lopez 1998).

Structure vs agency (institution vs. individual) debate

Some argue that to understand religion in terms of beliefs, or even in terms of any subjective states, reflects a Protestant bias and that scholars of religion should therefore shift attention from hidden mental states to the visible institutional structures that produce them. Although the structure/agency debate is still live in the social sciences, it is unlikely that one can give a coherent account of religion in terms of institutions or disciplinary practices without reintroducing mental states such as judgements, decisions, and dispositions (Schilbrack 2021).

Polythetic definitions of religion are illogical

Polythetic definitions of religion reject the idea that a concept or entity needs to have some essence. In doing so, polythetic definitions reject The law of identity (P is P) and therefore lead to logical contradictions and flawed thinking.

That a concept always corresponds to something possessing a defining property is a very old idea.

The traditional assumption is that every entity has some essence that makes it the thing it is, and every instance that is accurately described by a concept of that entity will have that essence. The recent argument that there is an alternative structure—that a concept need not have necessary and sufficient criteria for its application—has been called a “conceptual revolution” (Needham 1975: 351).


Some properties of a concept “crop up and disappear”

In discussions of the concept religion, this anti-essentialist approach is usually traced to Ludwig Wittgenstein (1953, posthumous). Wittgenstein argues that, in some cases, when one considers the variety of instances described with a given concept, one sees that among them there are multiple features that “crop up and disappear”, the result being “a complicated network of similarities overlapping and criss-crossing” (Wittgenstein 1953, §68). The instances falling under some concepts lack a single defining property but instead have a family resemblance to each other in that each one resembles some of the others in different ways. All polythetic approaches reject the monothetic idea that a concept requires necessary and sufficient criteria. But unappreciated is the fact that polythetic approaches come in different kinds, operating with different logics.

Polythetic definitions of religion

bounded polythetic approach prototypical examples borderline quasi

The most basic kind of polythetic approach holds that membership in a given class is not determined by the presence of a single crucial characteristic. Instead, the concept maps a cluster of characteristics and, to count as a member of that class, a particular case has to have a certain number of them, no particular one of which is required.

Because this illustration limits the number of characteristics in the properties set, I will call this first kind a “bounded” polythetic approach.

Understanding the concept religion in this polythetic way produces a graded hierarchy of instances.

Historically speaking, prototypical examples of the concept are likely to be instances to which the concept was first applied. Psychologically speaking, they are also likely to be the example that comes to mind first to those who use the concept.

If a form of life has only three, then it would be a borderline example. A form of life that has only two of these characteristics would not be included in the category, though such cases might be considered “quasi-religions” and they might be the most interesting social forms to compare to religions

A form of life that only had one of the five characteristics would be unremarkable. The forms of life that had three, four, or five of these characteristics would not be an unrelated set but rather a “family” with multiple shared features, but no one characteristic (not even belief in superempirical beings or powers) possessed by all of them. On this polythetic approach, the concept religion has no essence, and a member of this family that only lacked one of the five characteristics—no matter which one—would still clearly be a religion.

Benson Saler (1993) points out, one can use this non-essentialist approach not only for the concept religion but also for the elements within a religion (sacrifice, scripture, and so on) and to individual religions

Polythetic approaches can still be useful?

Some have claimed that, lacking an essence, polythetic approaches to religion make the concept so vague that it becomes useless (e.g., Fitzgerald 2000: 72–3; Martin 2009: 167). Given the focused example of a “bounded” approach in the previous paragraph and the widespread adoption of polythetic approaches in the biological sciences, this seems clearly false.

However, it is true that one must pay attention to the parameters at work in a polythetic approach. Using a properties set with only five elements produces a very focused class, but the properties set is simply a list of similarities among at least two of the members of a class, and since the class of religions might have hundreds of members, one could easily create a properties set that is much bigger.

there is no reason why one might not work with a properties set for religion with dozens or even hundreds of shared properties.

treating the concept religion in this way can lead to surprising discoveries of patterns within the class and the co-appearance of properties that can lead to explanatory theories.

The second key parameter for a polythetic approach is the threshold number.

each member has a majority of the properties, but this is not a requirement of polythetic approaches. The critics are right that as one increases the size of the properties set and decreases the threshold number, the resulting category becomes more and more diffuse. This can produce a class that is so sprawling that it is difficult to use for empirical study.

Scholars of religion who have used a polythetic approach have typically worked with a “bounded” approach (that is, with a properties set that is fixed), but this is not actually the view for which Wittgenstein himself argues. Wittgenstein’s goal is to draw attention to the fact that the actual use of concepts is typically not bound: “the extension of the concept is not closed by a frontier” (Wittgenstein 1953, §67). We can call this an “open” polythetic approach.

We might say that a bounded polythetic approach produces concepts that are fuzzy, and an open polythetic approach produces concepts that are fuzzy and evolving. Timothy Williamson calls this “the dynamic quality of family resemblance concepts” (1994: 86).

Why do we call something a “religion”? Well, perhaps because it has a direct relationship with several things that have hitherto been called religion; and this can be said to give an indirect relationship to other things we call the same name. (Wittgenstein 1953, §67)

Given an open polythetic approach, a concept evolves in the light of the precedents that speakers recognize, although, over time, what people come to label with the concept can become very different from the original use.

In the academic study of religions, discussions of monothetic and polythetic approaches have primarily been in service of developing a definition of the term.

If one were to offer a lexical definition (that is, a description of what the term means in common usage, as with a dictionary definition), then the definition one offers could be shown to be wrong.

On this point, some believe erroneously that one can correct a definition by pointing to some fact about the referents of the term.

One can correct a real or lexical definition in this way, but not a stipulative definition, that is, a description of the meaning that one assigns to the term. When one offers a stipulative definition, that definition cannot be wrong. Stipulative definitions are assessed not by whether they are true or false but rather by their usefulness, and that assessment will be purpose-relative

De Muckadell (2014) rejects stipulative definitions of religion for this reason, arguing that one cannot critique them and that they force scholars simply to “accept whatever definition is offered”.

However, even without knowing the real essence of religion, one can critique a stipulative definition, either for being less adequate or appropriate for a particular purpose (such as studying forms of life across cultures) or, as with the ice-skating example, for being so far from a lexical definition that it is adequate or appropriate for almost no purpose.

Polythetic definitions are increasingly popular today as people seek to avoid the claim that an evolving social category has an ahistorical essence.[14] However, the difference between these two approaches is not that monothetic definitions fasten on a single property whereas polythetic definitions recognize more. Monothetic definitions can be multifactorial, as we have seen, and they can recognize just as many properties that are “common” or even “typical” of religions, without being essential. The difference is also not that the monothetic identification of the essence of religion reflects an ethnocentrism that polythetic approaches avoid. The polythetic identification of a prototypical religion is equally ethnocentric. The difference between them, rather, is that a monothetic definition sorts instances with a Yes/No mechanism and is therefore digital, and a polythetic definition produces gradations and is therefore analog. It follows that a monothetic definition treats a set of instances that all possess the one defining property as equally religion, whereas a polythetic definition produces a gray area for instances that are more prototypical or less so. This makes a monothetic definition superior for cases (for example, legal cases) in which one seeks a Yes/No answer. Even if an open polythetic approach accurately describes how a concept operates, therefore, one might, for purposes of focus or clarity, prefer to work with a closed polythetic account that limits the properties set, or even with a monothetic approach that limits the properties set to one. That is, one might judge that it is valuable to treat the concept religion as structurally fuzzy or temporally fluid, but nevertheless place boundaries on the forms of life one will compare.

This strategy gives rise to a third kind of polythetic approach, one that stipulates that one property (or one set of properties) is required. Call this an “anchored” polythetic definition. Consistently treating concepts as tools, Wittgenstein suggests this “anchored” idea when he writes that when we look at the history of a concept,

what we see is something constantly fluctuating … [but we might nevertheless] set over against this fluctuation something more fixed, just as one paints a stationary picture of the constantly altering face of the landscape.

Given a stipulated “anchor”, a concept will then possess a necessary property, and this property reintroduces essentialism. Such a definition nevertheless still reflects a polythetic approach because the presence of the required property is not sufficient to make something a religion.

An anchored definition of religion like this would have the benefits of the other polythetic definitions. For example, it would not produce a clear line between religion and nonreligion but would instead articulate gradations between different forms of life (or between versions of one form of life at different times) that are less or more prototypically religious. However, given its anchor, it would produce a more focused range of cases.[15] In this way, the use of an anchor might both reflect the contemporary cosmological view of the concept religion and also address the criticism that polythetic approaches make a concept too vague.

Reflexive scholars have argued that the fact that what counts as religion shifts according to one’s definition reflects an arbitrariness in the use of the term. They argue that the fact that religion is not a concept found in all cultures but rather a tool invented at a certain time and place, by certain people for their own purposes, and then imposed on others, reveals its political character. The perception that religion is a politically motivated conceptual invention has therefore led some to skepticism about whether the concept picks out something real in the world. As with instrumentalism in philosophy of science, then, reflection on religion has raised doubts about the ontological status of the referent of one’s technical term.

A watershed text for the reflexive turn regarding the concept religion is Jonathan Z. Smith’s Imagining Religion (1982).

A central theme of his essays is that the concept religion (and subcategories such as world religions, Abrahamic faiths, or nonliterate traditions) are not scientific terms but often reflect the unrecognized biases of those who use these concepts to sort their world into those who are or are not “like us”.

Smith shows that, again and again, the concept religion was shaped by implicit Protestant assumptions, if not explicit Protestant apologetics.

[T]here is no data for religion. Religion is solely the creation of the scholar’s study. It is created for the scholar’s analytic purposes by his imaginative acts of comparison and generalization. Religion has no independent existence apart from the academy. (1982: xi, italics in original)

Smith is not a nonrealist about religion. In the first place, he did not think that the constructed nature of religion was something particular to this concept: any judgement that two things were similar or different in some respect presupposed a process of selection, juxtaposition, and categorization by the observer.

His slogan is that one’s (conceptual) map is not the same as and should be tested and rectified by the (non-conceptual) territory

Smith did not think that scholars should cease to use religion as a redescriptive or second-order category to study people in history who lacked a comparable concept. On the contrary, he chastised scholars of religion for resting within tradition-specific studies, avoiding cross-cultural comparisons, and not defending the coherence of the generic concept.

Smith did not think that the fact that concepts were human products undermined the possibility that they successfully corresponded to entities in the world: an invented concept for social structures can help one discover religion—not “invent” it—even in societies whose members did not know the concept.

The second most influential book in the reflexive turn in religious studies is Talal Asad’s Genealogies of Religion (1993).

Smith and Asad are therefore examples of scholars who critique the concept religion without denying that it can still refer to something in the world, something that exists even before it is named.

Other critics have gone farther. They build upon the claims that the concept religion is an invented category and that its modern semantic expansion went hand in hand with European colonialism, and they argue that people should cease treating religion as if it corresponds to something that exists outside the sphere of modern European influence. It is common today to hear the slogan that there is no such “thing” as religion. In some cases, the point of rejecting thing-hood is to deny that religion names a category, all the instances of which focus on belief in the same kind of object—that is, the slogan is a rejection of substantive definitions of the concept (e.g., Possamai 2018: ch. 5). In this case, the objection bolsters a functional definition and does not deny that religion corresponds to a functionally distinct kind of form of life. Here, the “no such thing” claim reflects the unsettled question, mentioned above, about the grounds of substantive definitions of “religion”. In other cases, the point of this objection is to deny that religion names a defining characteristic of any kind—that is, the slogan is a rejection of all monothetic definitions of the concept. Perhaps religion (or a religion, like Judaism) should always be referred to in the plural (“Judaisms”) rather than the singular. In this case, the objection bolsters a polythetic definition and does not deny that religion corresponds to a distinct family of forms of life. Here, the “no such thing” claim rejects the assumption that religion has an essence. Despite their negativity, these two objections to the concept are still realist in that they do not deny that the phrase “a religion” can correspond to a form of life operating in the world.

More radically, one sees a denial of this realism, for example, in the critique offered by Wilfred Cantwell Smith (1962).

Smith’s thesis is that in many different cultures, people developed a concept for the individuals they considered pious, but they did not develop a concept for a generic social entity, a system of beliefs and practices related to superempirical realities. Before modernity, “there is no such entity [as religion and] … the use of a plural, or with an article, is false” (1962: 326, 194; cf. 144). Smith recommends dropping religion. Not only did those so described lack the concept, but the use of the concept also treats people’s behavior as if the phrase “a religion” names something in addition to that behavior. A methodological individualist, Smith denies that groups have any reality not explained by the individuals who constitute them. What one finds in history, then, is religious people, and so the adjective is useful, but there are no religious entities above and beyond those people, and so the noun reifies an abstraction.

[n]either religion in general nor any one of the religions … is in itself an intelligible entity, a valid object of inquiry or of concern either for the scholar or for the [person] of faith. (1962: 12)

More radical still are the nonrealists who argue that the concepts religion, religions, and religious are all chimerical. Often drawing on post-structuralist arguments, these critics propose that the notion that religions exist is simply an illusion generated by the discourse about them

Advocates of this position sometimes call their approach the “Critical Study of Religion” or simply “Critical Religion”, a name that signals their shift away from the pre-critical assumption that religion names entities in the world and to a focus on who invented the concept, the shifting contrast terms it has had, and the uses to which it has been put.

Inspired by the post-structuralist critiques that religion does not apply to cultures that lack the concept, some historians have argued that the term should no longer be used to describe any premodern societies, even in Europe.

though it is common to speak of religions existing in the past, human history until the concept emerged in modernity is more accurately understood as a time “before religion”. His aim is “to dispel the commonly held idea that there is such a thing as ‘ancient religion’” (2013: 8).

Citing Nongbri, Carlin Barton and Daniel Boyarin (2016) argue that the Latin religio and the Greek thrēskeia do not correspond to the modern understanding of religion and those studying antiquity should cease translating them with that concept. There was no “Roman religious reality”, they say (2016: 19). These historians suggest that if a culture does not have the concept of X, then the reality of X does not exist for that culture. Boyarin calls this position “nominalism”, arguing that religion is

not in any possible way a “real” object, an object that is historical or ontological, before the term comes to be used. (2017: 25)

These critics are right to draw attention to the fact that in the mind of most contemporary people, the concept religion does imply features that did not exist in ancient societies, but the argument that religion did not exist in antiquity involves a sleight of hand. None of these historians argues that people in antiquity did not believe in gods or other spiritual beings, did not seek to interact with them with sacrifices and other rituals, did not create temples or scriptures, and so on. If one uses Tylor’s definition of religion as belief in spiritual beings or James’s definition of religion as adjusting one’s life to an unseen order—or any of the other definitions considered in this entry—then religion did exist in antiquity. What these historians are pointing out is that ancient practices related to the gods permeated their cultures.

To be sure, ancient people had words to describe proper reverence of the gods, but … [t]he very idea of “being religious” requires a companion notion of what it would mean to be “not religious” and this dichotomy was not part of the ancient world; (2013: 4)

there was no “discrete sphere of religion existing prior to the modern period”

The point is not … that there weren’t practices with respect to “gods” (of whatever sort) but that these practices were not divided off into separate spheres

Steve Mason also argues that religion did not exist in antiquity since religion is “a voluntary sphere of activity, separate in principle” from politics, work, entertainment, and military service (2019: 29). In short, what people later came to conceptualize as religion was in antiquity not a freestanding entity. The nominalist argument, in other words, adds to the definition of the concept religion a distinctively modern feature (usually some version of “the separation of church and state”), and then argues that the referent of this now-circumscribed concept did not exist in antiquity. Their argument is not that religion did not exist outside modernity, but that modern religion did not exist outside modernity.

These post-structuralist and nominalist arguments that deny that religion is “out there” have a realist alternative. According to this alternative, there is a world independent of human conceptualization, and something can be real and it can even affect one’s life, whether or not any human beings have identified it. This is true of things whose existence does not depend on collective agreement, like biochemical signaling cascades or radioactive beta particles, and it is equally true of things whose existence does depend on collective agreement, like kinship structures, linguistic rules, and religious commitments.

This realist claim that social structures have existed without being conceptualized raises the question: if human beings had different ways of practicing religion since prehistoric times, why and when did people “finally” create the taxon? Almost every scholar involved in the reflexive turn says that religion is a modern invention.

Nevertheless, it is not plausible that modern Europeans were the first to want a generic concept for different ways of interacting with gods. It is easy to imagine that if the way that a people worship their gods permeates their work, art, and politics, and they do not know of alternative ways, then it would not be likely that they would have created a concept for it. There is little need for a generic concept that abstracts a particular aspect of one’s culture as one option out of many until one is in a sustained pluralistic situation. The actions that today are categorized as religious practices—burial rites, the making of offerings, the imitation of divinized ancestors—may have existed for tens of thousands of years without the practitioners experiencing that diversity or caring to name it. Nevertheless, it is likely that a desire to compare the rules by which different people live in relation to their gods would have emerged in many parts of the world long before modernity. One would expect to find people developing such social abstractions as cities and then empires emerged and their cultures came into contact with each other. From this realist perspective, it is no surprise that, according to the detailed and example-filled argument of Barton and Boyarin (2016), the first use of religion as a generic social category, distinct from the concept of politics, for the ways that people interact with gods is not a product of the Renaissance, the Reformation, or modern colonialism at all, but can be found in the writings of Josephus (37–c. 100 CE) and Tertullian (c. 155–c. 220 CE).

From the realist perspective, it is no surprise to see the development of analogous terms in medieval China, centuries before interaction with Europeans (Campany 2003, 2012, 2018) and in medieval Islam (Abbasi 2020, 2021). The emergence of social kinds does not wait on language, and the development of language for social kinds is not only a Western project. If this is right, then the development of a concept for religion as a social genus is at least two thousand years old, though the social reality so labeled would be much older.


Schilbrack, K. (2022). The Concept of Religion. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2022). Metaphysics Research Lab, Stanford University.