Culture and Comparative Social Science

“Culture” is a complex, contested, polysemous concept. Its meaning shifts from discipline to discipline, literature to literature. In some, it is aligned with discussions of identity, recognition, and belongingness; in others, descriptions of fundamental meaningfulness or cosmology. Like many polysemous concepts, “culture” supports a patchwork of partially overlapping analytical domains (Wilson 2006; Haueis forthcoming). In the comparative social sciences,Footnote 1 researchers take the “culture” of different “cultural groups” to explain differences in individual or aggregate behavior. I include in these sciences the work of experimental philosophers (e.g. Knobe et al. 2012), who have adopted the methods and concepts of comparative social science to investigate the universality of philosophical intuitions; for instance, about semantics (Machery et al. 2004), epistemology (Weinberg et al. 2001; Machery et al. 2017), and action (Knobe and Burra 2006).

Because these sciences use “culture” or “cultural groups” in the same way as other demographic categories (e.g. “gender”, “socioeconomic status”), I refer to their joint conceptualization in comparative work as demographic cultures or the demographic cultures concept. According to a prominent line of skepticism, however, because of the polysemous and ambiguous nature of the “culture” concept, the demographic cultures concept will similarly be unstable, empirically fraught, and theoretically pernicious. Footnote 2 Call this position demographic skepticism. My overall aim in this paper is to defend the cogency and use of the demographic cultures concept against such a skeptic.

My central targets are the metaphysical, ontological, and explanatory concerns raised by the skeptic. According to such a skeptic, these concerns arise because comparative social science is aligned with an outdated model of cultures and cultural groups—what I call the folk anthropological model. To the skeptic, this alignment renders the demographic cultures concept unfit for purpose: the concept generates ontologically mysterious and metaphysically problematic characterizations of cultural groups which cannot support principled comparative work.

I take the skeptic to be putting forward four problems for the comparative social scientist. Namely, that the demographic culture concept:

  • is committed to the strong empirical assumptions of the folk anthropological model (notably, holistic organization, homogenous communalism, and strong boundaries);

  • is committed to a particular account of essentialism about cultural groups, hidden structure essentialism;

  • lacks clear ontological operationalizations for “culture” and “cultural groups,” and;

  • does not identify causal mechanisms that explain the systematic similarities and differences between cultural groups and their constituent members.

These skeptical problems are real. Nonetheless their implications for comparative social science have been vastly overstated.

The problems do not target necessary elements of comparative social science. Because of this, one can turn the skeptical problems on their head—extracting from them recommendations for how comparative social scientific practice should be undertaken. When using the demographic cultures concept, comparative social scientists should:

  • avoid the strong empirical assumptions of the folk anthropological model, or clearly justify when and where these can be adopted;

  • adopt the metaphysical framework of relational essentialism;

  • explicitly articulate ontological operationalizations for “culture” and “cultural groups”, and;

  • link empirical claims about cultural groups to causal mechanisms that can explain the systematic similarities and differences between cultural group members.

There are two main upshots of this paper. The first is an analysis of demographic skepticism, which has not received sustained philosophical attention. This is needed because the skeptical position relies on substantive metaphysical ideas. Clarifying the skeptical position and bringing it into dialogue with the metaphysical literature is itself a useful contribution. The second is the articulation of four recommendations for the comparative social scientist. These are needed to avoid the real and unsavory empirical implications identified by the skeptic.

The above lists of problems and recommendations provide the rough outline of the paper. After a brief discussion introducing the folk anthropological model, I consider its strong empirical assumptions. Two sections then introduce metaphysical and philosophical machinery, which are applied to problems of metaphysical essentialism, ontological fickleness, and ontological uncertainty. I address skeptical arguments against using culture as an explanans before concluding.

Demographic Skepticism and the Folk Anthropological Model

Up to this point I have used the terms “culture”, “cultures”, and “cultural groups” indiscriminately. Going forward, some definitional clarity is needed. In this paper, I will be using the term “cultures” to identify a class that gathers “cultural groups”.Footnote 3 This is to distinguish the structure of the cultures class as the traditional target for anthropological theorizing, which (typically) used ethnographic material on particular “cultural groups” as evidence. “Culture”, by contrast, is what is possessed by members of any particular cultural group—and this tends to be talked about in generic or aggregate terms (e.g. “Quebecois culture.” More on this below). The “demographic cultures concept” is a joint conceptualization of both “culture” and “cultural groups” that imposes two constraints on the cultural groups it gathers, namely, that they are individuatable and that their respective members have qualitatively distinct culture. These two assumptions are what facilitate the application of comparative tools to cultural groups.

With these definitions in hand I turn to articulate demographic skepticism in more detail.

The skeptical position is motivated by an alignment of the demographic cultures concept with the folk anthropological model. This model takes cultural groups to be geographical isolates: discrete populations with unique ways of life, whose distinctiveness is maintained through boundaries or barriers.Footnote 4 I call this a “folk” model because it is grounded in commonsense ideas about what anthropologists do—they go into the field and study a culture.

Whatever its intuitive appeal, the folk anthropological model has long been criticized for smuggling in troubling metaphysical and empirical assumptions.Footnote 5 Four stand out as particularly contentious:

Cultural group essentialism:

cultural groups have a set of essential properties that underpins their classification and explains the similarities and differences of group members;

Holistic organization:

the culture of group members is composed of tightly integrated, organized structures of interlocking elements including norms, practices, and values;

Homogenous communalism:

cultural group members share most (if not all) values, practices, and beliefs in common, and;

Strong boundaries:

cultural groups have geographic, linguistic, commercial, and/or ethnic boundaries that prevent the hybridization and diffusion of culture.

The folk anthropological model thus takes cultural groups to be essentialized wholes characterized by “coherence, timelessness, and discreteness.” (Abu-Lughod 1991, 146) As Clifford (1988) puts it, the folk anthropological model “constructs and reconstructs coherent cultural others.” (112) Narayan (2000) provides an excellent description of the model when she states:

This view understands cultures on the model of neatly wrapped packages, sealed off from each other, possessing sharply defined edges or contours, and having distinctive contents that differ from those of other “cultural packages.” (1084)

I think there is a good case to be made that the folk anthropological model informs and guides folk ideas about anthropological practice and cultural groups. Yet it is not clear to me that any social scientific or anthropological researcher has endorsed the folk anthropological model or uncritically adopted its assumptions.Footnote 6 And as I suggest below, it is a poor depiction of contemporary comparative social science. Nonetheless it continues to be used by skeptics as if it were a faithful depiction of contemporary research frameworks.Footnote 7

As noted above, the foundation of the skeptical position is the alignment of the folk anthropological model with comparative social science. Consider here remarks from Hirschfeld (2018), a demographic skeptic who will serve as one of the main interlocutors in this essay:

The growing relevance of cultural variation in psychology has not extended to a close examination of the notion of culture. Rather, there is wide acceptance that culture refers to a specific kind of social form that is entity-like, bounded, timeless, stable, symbolically fused, and highly shared in terms of the values, practices, and interests embodied in cultural performances and representations. (232)

By aligning comparative social science with the folk anthropological model, Hirschfeld implies that such work is committed to the same set of contentious assumptions: to cultural group essentialism (“timeless, stable” cultural groups), holistic organization (“symbolically fused”), homogenous communalism (shared “values, practices, and interests”), and strong boundaries (“entity-like, bounded”). The problem is—as the skeptic is only too happy to point out—that there is overwhelming evidence that culture and cultural groups are not stable, timeless, homogenous, holistically organized, or bounded. So by showing the commitments of the folk anthropological model to be untenable, the skeptic also takes themselves to be showing that comparative social science as a whole is untenable.

In what follows, I argue against the skeptic: questioning the alignment between the folk anthropological model and comparative social science, disaggregating the tight links between its contentious assumptions, and pointing to counterexamples in contemporary social scientific work. In so doing, I articulate a set of recommendations that social scientists should abide by in order to avoid the skeptic.

Skepticism Targeting Empirical Assumptions

Holistic organization, homogenous communalism, and strong boundaries are testable assumptions about cultural groups and the culture of their members. Unfortunately, when researchers have tested these claims, they find that they rarely hold.

One need not dwell on the failings of these assumptions in any great detail. Over the last fifty-some odd years, work in the historical and social sciences have amassed substantial evidence of their precariousness. Even deep in human history, cultural groups were sites of significant diffusion, change, migration, and hybridization. These are features that have only intensified over the centuries. If contemporary social science were indeed committed to these assumptions, they would be in conflict with this substantial body of evidence.Footnote 8

But these are not necessary assumptions. Social scientists do not need to assume cultural groups have strong boundaries, that group members are homogenous, or that their culture is holistic in order to support comparative work. Just like other demographic categories, one simply needs a principled means of generating appropriate reference classes (for instance, what “socioeconomic status” might mean across different populations). As a wealth of work in social science shows, this is an important, but far from impossible demand (Henrich 2020; Mohr et al. 2020; Vaisey 2021).

The leads to the first recommendation around the use of the demographic cultures concept: avoid the strong empirical assumptions of the folk anthropological model. This is not to say that these strong empirical assumptions are everywhere and always wrong, or that they cannot support comparative social scientific work—they aren’t, and they can. But they are not the only principled means of generating the like-for-like juxtapositions needed for comparative social scientific work and should be avoided unless explicitly acknowledged and argued for.

Cultural Group Essentialism

The fourth commitment of the folk anthropological model is metaphysical. Here I examine the skeptic’s claims that comparative social science is committed to cultural group essentialism, and consider why the skeptic takes such a commitment to undermine successful empirical work.

It is important to begin by noting that many disciplines are allegedly committed to cultural group essentialism. Interpretive, critical, and empirical work across the humanities, social sciences, and natural sciences are often targeted by such accusations. In general, cultural group essentialism is taken to involve the attribution of an unchanging, “timeless” set of properties to cultural groups. And though skeptics vary in what they take to be the downstream effects of cultural group essentialism, all argue that assuming cultural groups are “timeless” leads to poor, inaccurate, or unjust empirical research. Footnote 9

But what is meant by “essentialism”? A good case can be made that when the demographic skeptic talks about “essentialized” or “timeless” cultural groups, they have in mind a familiar type of essentialism made famous by Kripke (1980) and Putnam (1975) about natural kinds. This position distinguishes between observable clusters of properties and an underlying essence or “hidden structure”. These hidden structures serve both to sort entities into kinds and to explain why members have the observable properties that they do. The canonical example in the literature is gold: gold has a number of observable properties (being yellow, shiny, malleable, conductive) explained by an underlying atomic structure. When skeptics suggest that social scientific work is committed to essentialism, they appear to be making the claim that cultural groups are kinds and that each has a hidden structure.

This helps to make sense of the claims about the “timelessness” of cultural groups. For hidden structures are typically assumed to be sets of essential properties, and essential properties cannot change without a corresponding change in the identity of the object.

There are many ways of understanding what it means to be an essential property and how to distinguish these from accidental ones. The most widely accepted account today understands this difference in terms of modality: essential properties are those that an object must have. Thus, what it is to be gold is to have a certain number of protons—since in any possible world, anything that does not have this number of protons is not gold. Non-essential properties, by contrast, may differ widely across these possible worlds. Atoms of gold may have greater or lesser numbers of electrons, be hammered into different shapes, and be in alloys of different colors. The key intuition here is that changing essential properties changes the identity of objects. Gaining a proton, say, would change both what an atom is (the kind it belongs to) as well as the observable properties it displays (its luster and color). To return to the classic example, if the atoms of some sample of gold were to each gain a proton, these shiny, solid, and yellow atoms of gold would go out of existence and some liquid, silver atoms of mercury would come into existence.

Applying this framework to cultural groups gives the following. Cultural groups are kinds and those kinds have a hidden structure. Any change to the hidden structure of cultural groups would mean a change in the identity of that group. So, for instance, if one holds that language is part of the hidden structure, then changing that language would mean that one group would go out of existence and a new one would spring into existence.

With these analyses in hand, we can distinguish two distinct problems for cultural group essentialism that the skeptic may be pointing toward. First, if cultural group essentialism means that cultural groups have hidden structures that determine what they are and when they exist, this suggests that some properties of cultural groups cannot change over time. This underpins the claim that cultural group essentialism is committed to a “timeless” nature of cultural groups. Yet the empirical record provides overwhelming evidence that cultural groups change frequently over time, and that no property is spared from such change. A dogged commitment to cultural group essentialism seems to entail such groups are ontologically fickle: they would frequently pop in and out of existence.

Second is the problem of ontological uncertainty. If there are hidden structures, contemporary social science does not agree as to what these are. This might be because researchers are either unclear about hidden structures or do not provide guidance as to what they might be. Both situations seem to undermine comparative social science. Such research requires clarity about when cultural groups exist, change, and go extinct.

These issues complicate the epistemology of comparative work. If hidden structures are obscure, or if entities are constantly changing their essential properties, it is hard to see how they could support empirical research or compelling explanations.

Individuation and Essentialism

Before addressing the metaphysical and ontological issues at the core of demographic skepticism, it is worth stepping back to interrogate their claims. Do comparative social scientists think cultures are “timeless”? Are they committed to cultural group essentialism?

Again, Hirschfeld (2018) serves as a useful exemplar of the skeptic as his arguments follow a common pattern found in the literature. This does not directly object to the empirical work of comparative researchers—something that should already raise eyebrows—but instead provides a narrative about the disciplinary history of anthropology. This narrrative purports to show that anthropology jettisoned cultural group essentialism along with any attempt at individuating and classifying cultural groups. Nonetheless, by freeing themselves of problematic metaphysics, anthropologists were then able to adopt more rigorous and principled methods for their interpretive and empirical projects.

This narrative is also meant to demonstrate a tension between cultural group essentialism and empirical work aimed at understanding such a varying, changing, and hybridizing entity.Footnote 10 And for reasons explored in more detail below, this skeptical argument might be correct about the failings of hidden structure essentialism. But as it stands, this argument does not show that comparative social scientific work is committed to such a metaphysical picture.

Making an even stronger argument, then, Hirschfeld mobilizes work on evolutionary psychology to argue that any individuation of groups will invariably lead to hidden structure essentialism. His argument runs as follows. Comparative researchers of all stripes make descriptive generalizations and explanatory claims about groups. These groups in turn will have some features that set them apart—they might look a certain way or do things in a different manner. One might even be able to explain these differences by pointing to deeper features like shared values or practices. However distinguished, these descriptive and explanatory generalizations feed into cognitive tendencies that essentialize groups (Hirschfeld 2018, 242; 1998; Morris 2014). In other words, because of our cognitive biases, hidden structure essentialism is the expected outcome of any attempt to individuate groups as groups of a particular kind, whether these are races, religions, cultures, or ethnicities. And while other demographic skeptics don’t pursue Hirschfeld’s psychological reasoning, most agree that any individuation of cultural groups will lead to hidden structure essentialism.Footnote 11.

I see two main reasons to be suspicious of this stronger argument. First, even if one believes that naïve theorizing about groups is guided by universalizing tendencies—for instance, that humans have a tendency to overestimate the similarities between perceived out-group members (Hirschfeld 1998)—this does not seem to be the right place to look for the metaphysical or ontological commitments of scientists. Such commitments are often embedded in the assumptions of formal models, empirical research strategies, and other methodological protocols. While it is true that scientists are human and thus subject to epistemological vices and biases, their tools and methods are often designed to explicitly control for—or counteract—such vices and biases. Moreover, scientists assemble themselves into broader scientific communities that provide opportunities for scrutiny and critique. So it seems suspect to look for metaphysical commitments in human psychology rather than in the actual practices of comparative social scientists.

Indeed, when one looks to comparative social science, one finds strategies explicitly designed to avoid cultural group essentialism. Consider here what Emirbayer and Goodwin (1994) have called the “anti-categorical imperative.” This is a recommendation that comparative researchers avoid frameworks and explanations that appeal solely to classificatory categories. And this is a recommendation that appears to have been widely taken up. Goldberg and Stein’s (2018) associative diffusion model of cultural variation, for instance, includes only simple mechanisms of learning and evaluation (semantic association, “constraint satisfaction”). Nonetheless, this model generates groups of individuals with similar clusters of traits (a “semantic network”). But the model itself makes no commitments to an underlying hidden structure of such groups. There is nothing fixed about the semantic networks nor the cultural elements being transmitted and evaluated.

Moreover, when researchers are actually interrogated about their commitments, they readily acknowledge the dynamic nature of culture and cultural change. This is true even for the cultural psychologists targeted by Hirschfeld (e.g. Kashima and Gelfand 2011). This should lead one to be suspicious of claims that comparative cultural work somehow fails to recognize cultural change and variation. While it is possible that researchers might be confused about their own metaphysical commitments, it would be strange indeed for them to identify cultural change as central to their work, recognize the problems of “timeless” cultural groups, adopt explicit methodological strategies to avoid such problems, and yet remain committed to metaphysics that entail such timelessness.

A second reason to doubt that comparative researchers are committed to cultural group essentialism concerns the evidence used by demographic skeptics. What skeptics identify as problematic are general explanatory claims that appeal to social categories: for example, that “theory of mind develops differently in strongly collectivist cultures” or that “the United States has a distinctive set of individualistic values”.Footnote 12 Such claims are supposed to be evidence that cultural groups are being essentialized, and that the properties or features identified are part of the hidden structure. Yet even when such claims make specific attributions to cultural groups—for instance, that culture can explain different philosophical intuitions on Gettier cases (Machery et al. 2017)—it is not clear these claims embed metaphysical commitments about cultural groups.

Rather than being claims about an essence of a cultural group, one can read such claims as generics. Generics are a kind of linguistic construction that can remain true in the face of exceptions; for instance, “bread has gluten” is true, even though there are many gluten-free loaves. Bare plural generics (“bread has gluten”, “tigers have stripes”, “cultures are highly integrated”) generate a number of philosophical problems but are unproblematically used in ordinary and scientific language and reasoning. For instance, researchers use bare plural generics in research and discussion about species (“tigers have stripes”, “platypuses hunt using electrolocation”) even while recognizing that such generics are made about changing and heterogeneous populations. The same is true of cultures. It seems equally reasonable to understand claims—including many causal-explanatory claims—about demographic cultures as being bare plural generics.

Taken together these suggest that attributions of cultural group essentialism to comparative social science are suspect.

Skepticism Targeting Hidden Structure Essentialism

Let me return to consider the metaphysical concerns of the skeptic. As mentioned above, the skeptic argues that hidden structure essentialism generates a dilemma when combined with what we know about culture. Hidden structures cannot change without changing what the cultural group fundamentally is. If changes to hidden structures are frequent then cultural groups are ontologically fickle, and if so, hidden essences are an unstable foundation on which to rest empirical claims. Recognizing that cultures in general change frequentlysuggests either that one give up on hidden structure essentialism or commit oneself to a fickle and empirically suspect picture of cultural groups.

Joining the skeptic, I will argue that the comparative social scientist should reject hidden structure essentialism. Yet this does not mean giving up on demographic cultures. There are other metaphysical positions that can support comparative social scientific work.

First, however, it is important to recognize the extent to which hidden structure essentialism is compatible with change. This fact is not well-recognized by the skeptic. As mentioned above, hidden structure essentialism allows for non-essential properties to vary. Depending on how researchers identify and draw boundaries around the hidden structure, the existence of cultural groups might be compatible with substantial change.

Yet hidden structure essentialism would only be able to ground research in comparative social science if there were a set of properties that all and only particular cultural groups held, remained fixed over time, and explained other observable properties. And here I agree with the skeptic that there are no good candidates for such hidden structures. From all we know about cultural groups and culture, they are inherently fluid and changing entities, constantly hybridizing, shifting, melding, and combining. All their elements are subject to change. The fluid character of cultural groups thus wholly undercuts hidden structure essentialism—it is not a plausible metaphysical framework for comparative social scientific work.

Nonetheless, rejecting hidden structure essentialism does not mean rejecting demographic cultures nor comparative projects and methods that use them. A different metaphysical picture, relational essentialism, can underwrite the classificatory needs of comparative researchers.

Relational essentialism identifies the essence of kinds with relational properties. Consider, as an example, Okasha’s (2002) relational essentialist account of species. On this account, species are what they are in virtue of genealogical relationships holding between organisms. For many sexually reproducing fauna, we can characterize this by saying that species are particular branches on the tree of life. Tracing branches back to the trunk, in effect, is retracing the evolutionary trajectory of that species back through time. On this picture, what it means to be a particular branch—a particular species—is to be a lineage of organisms. What it means to be a platypus, for instance, is to be a member of a populational lineage that connects up with a separate branch (the echidnas) about 19 million years ago, and so on.

Relational essentialism is compatible with substantial change. Platypuses could change considerably—they might grow in size, change the color of their pelt—yet remain platypuses just in case they emerge from the right genealogical relationships. Relational essentialism can also accommodate radical similarity across different species. For instance, platypuses have “duck bills” and “beaver tails”. But whatever their similarities, ducks and beavers are not platypuses. Those have different sets of essential relational properties and thus are different kinds of critters.

I think relational essentialism is a plausible metaphysical picture for work on demographic cultures. And while a more full-throated endorsement would require further exploration and defense, it is worth noting that the picture already underwrites accounts of social kinds like race and gender (Bach 2012; Godman 2020), as well as Alan Patten’s (2014) social lineage account of cultural groups.

A brief exploration of Patten’s account demonstrates how a relational essentialist metaphysics can do work classifying cultural groups while also accommodating substantial variation and change in group culture. As Patten argues, the social lineage account understands cultural groups to be sociohistorical isolates.Footnote 13 These persist just when group members control the institutions of education and enfranchisement of new members (2014, 47–54). Control over these institutions means that already enfranchised group members inculcate new recruits with prevailing ideas, values, practices, and norms. Control, socialization, and enfranchisement together generate sociohistorical lineages that individuate cultural groups and explain features of cultural group members.

Cultural groups, on Patten’s account, are just like species: the culture of group members can change yet the cultural group can nonetheless remain identical so long as the particular relationships brought about through control, socialization, and enfranchisement remain intact. As he writes:

At any given moment, its content consists in various beliefs, meanings, and practices, but what makes these the beliefs, meanings, and practices of a shared culture is that the people who hold them share a common social lineage. (Patten 2014, 51)

These beliefs, meanings and practices can change—and change radically—so long as they are endorsed by enfranchised members and used to inculcate new recruits.

Let me sum up. Skeptics move too quickly in aligning demographic cultures with hidden structure essentialism, breezing past what “timelessness” actually involves, and moving to a wholescale rejection of comparative approaches to culture. I am doubtful that many comparative social scientists are committed to hidden structure essentialism. Nonetheless, clarifying the skeptic’s arguments allows for a clear prescriptive recommendation. Comparative social scientists should adopt relational essentialism: it provides a plausible metaphysical framework for individuating cultural groups that avoids the problems of ontological fickleness and timelessness.

Skepticism Targeting Ontological Operationalization

What about the problem of ontological uncertainty? Often, skeptics complain that the demographic cultures concept relies on unclear ontological and empirical operationalizations, or worse, is never operationalized at all. In either case, there appears to be a lack of ontological specificity and theoretical clarity that can guide researchers. As Hirschfeld puts it, such work deploys an “intuitive, unscrutinized notion as the central concept in a causal argument.” (Hirschfeld 2018, 234).

What might the problem be with such an unscrutinized, under operationalized concept? Here it is useful to look at a formulation of the problem by the cognitive anthropologist Roy D’Andrade (1992):

There is no clear relation between culture and action. Of course, one can say “people do what they do because their culture makes them do it.” The problem with this formulation is that it does not explain anything. (23)

D’Andrade’s formulation identifies two problems: a lack of (i) ontological specificity about the nature of culture and (ii) a clear causal-explanatory link between culture and human behavior. As he puts it, “Unless there is some specification of how culture “makes” people do what they do, no explanation has been given.” (ibid.) While the two problems are linked, it is worth addressing each separately. In this section, I suggest that providing a clear ontological operationalization is not an onerous task—and that there are in fact several such operationalizations at work in the comparative social scientific literature today. In the next, I explore the claim that such comparative work fails to provide clear causal-explanatory linkages.

A lack of ontological specificity is often seen as endemic to comparative work on culture. Hirschfeld, for instance, surveys a landmark paper in cultural psychology (Markus and Kitayama 1991) and claims that the paper “used the concept of culture 105 times without discussion of what it might in fact refer to.” (Hirschfeld 2018, 237) He further notes “their paper is not unusual in this regard.” (ibid.) This underwrites expansive and general claims about comparative work: that “culture is never identified nor given any material description (beyond the action that is supposedly evidence of it)” (235).Footnote 14

Providing ontological operationalizations that are both encompassing and concrete is indeed difficult as these features trade off with each other. This because culture has a complex structure that is embedded in a range of material vehicles, behavioral expressions, vocalizations, institutions, and social roles. Yet this does not mean researchers lack ontological operationalizations suitable for empirical work.

D’Andrade, for instance, endorses an account where culture is identified with systematically organized “informational pools” embedded in material artefacts, behaviors, and environments. The links between culture and action are “cultural schema”: mental representations, instantiated in the brain, which allow agents to interpret their environment in terms of means-ends relationships (D’Andrade 1992, 31).

While it is true there is no prevailing consensus on the ontology of culture, there are clear clusters of research that adopt the same or similar operationalization. The informational operationalization favored by D’Andrade (1981; 1992) can be found elsewhere: in cultural evolution (Mesoudi 2011; Henrich 2015), sociology (Goldberg and Stein 2018), and archaeology (O’Brien and Lyman 2002). Beyond this, philosophers have identified a range of candidate ontologies for culture (Risjord 2012), with practice-based ontologies in particular finding adherents in the sociological (Bourdieu 1977), anthropological (Ingold 2011), and social ontology literature (Rouse 2015).

Of course, an ontology of “culture” is not the same as an ontology of “cultural groups” (in the technical sense in which these terms are being used here). There is room for significant divergence of opinion in how the two are linked. Those who identify culture with “information” tend to identify cultural groups with particular “informational pools”. This is especially prevalent in areas of sociology and the evolutionary human sciences. In evolutionary anthropology, economics, and political science, by contrast, cultural groups are frequently identified with ethnolinguistic populations—that is, with the practices, behaviors, and norms of populations with shared biological ancestry. And as noted above, in philosophy, Patten’s (2014) social lineage account identifies cultural groups with particular sociohistorical lineages resulting from institutional control, socialization, and enfranchisement.

This leads to a third recommendation for the comparative social scientist: clearly operationalize both “culture” and “cultural groups.”

This effort need not be aimed at producing a single, monistic accout of “culture” or “cultural groups.” There is no reason why comparative social science cannot be pluralistic about the demographic cultures concept in the same way that other disciplines—like the life sciences—are pluralistic about their concepts. Philosophers of science have long noted how complex entities support different yet empirically fruitful ontological operationalizations (Dupré 1993). Demographic cultures do not stand out as distinct in this regard.

I do not think the demand for clear operationalizations to be onerous. In fact, contrary to the skeptic’s claims, I think much work already employs sufficiently clear operationalizations that are well-suited to particular research efforts. While the skeptic is surely right that some research lacks clear and explicit operationalizations—and others, any operationalization at all—I do not think this is representative of all (or even, most) comparative social science.

Skepticism Targeting Culture as an Explanans

There is one final component of the skeptical position. This links the ontology of “culture” and “cultural groups” to explanatory concerns. Again, it is helpful to return to Hirschfeld. As he writes:

The overwhelmingly dominant view in psychology is that the conjunction of features — spatially bounded, territorialized, convergent and shared systems of belief and practice derived from key symbols, key values, etc. — typifies a culture and as a result of which cultures are seen to possess a causal force. The Japanese rear their children as they do because of culture values for communitas and veneration to senior generations. (Hirschfeld 2018, 245).

It is the idea that “cultures” are a force that is so problematic.

Anthropologists and cultural psychologists should be interested in accounting for how universal phenomena are culturally inflected, treated in cultural discourses, and come to have the cultural profiles they enjoy. But they shouldn’t say that cultural inflection, discourse, and profile are caused by some cultural force. (Hirschfeld 2018, 246)

Alberto Acerbi (2020, 217) comes to a similar conclusion: “we should not think of culture as an entity and, especially, as an entity with causal powers.” Footnote 15

But why? What is so problematic about treating culture in this way? Here I think looking at both Hirschfeld's and Acerbi’s positive projects helps to reveal why they think the idea of a “cultural force” is so problematic and where they think the demographic culture concept goes wrong.

Both Hirschfeld and Acerbi express a commitment to an epidemiological framework inspired by the work of Dan Sperber (1996). This framework, broadly, explains the spread of mental representations from individual to individual by appealing to environmental and psychological causes. Such mental representations can be more or less cultural, but they are not, just by dint of being socially transmitted, a cultural thing.

These are ontological commitments. There are no cultural groups on this account. All that exist are elements with more or less of a scalar quality: mental representations that are more or less pervasive. Cultural groups, if they are anything, are merely assemblages of agents, perhaps in specific contexts, that create conditions for traits to spread.Footnote 16 Consider, Hirschfeld’s (2018) discussion of the “Little Red Riding Hood” folk tale:

What makes Little Red Riding Hood cultural is the convergence of constraints on transformations inherent in communication, the evocation of a “catchy” narrative structure, and a suite of memory and relevance functions. The cultural environment defined by Little Red Riding Hood is not itself a community; rather it defines a set of representational dispositions identified with a population in which a genre of children’s folk tales is widely distributed and has stabilized. (248).

Let us call this picture cultural pervasiveness, since whether something counts as being cultural hinges on the pervasiveness of mental representations in space and time. For demographic skeptics—especially those influenced by Sperber—this pervasiveness is in turn explained by concrete mechanisms.Footnote 17 These explain the differential distribution of traits in a population over time.

We can now better articulate Hirschfeld and Acerbi’s pronouncements against using culture, understood as a “cultural force”, as an explanans. What they are in fact demanding is that similarities and differences among individuals be explained by causal mechanisms. Failing to do so generates comparative social scientific explanations that are unilluminating and obscurantist.

I think this demand is reasonable. This suggests a fourth and final recomomendation for comparative social scientists: researchers should ground explanations of similarities and differences in causal accounts.

But like the previous recommendation about ontological operationalization, I think there is a wealth of comparative work that already follows such a suggestion. Bad comparative work might merely appeal to culture as a force, but there is plenty of good work that puts forward compelling causal-mechanistic explanations.

Consider as an exemplar the ambitious claims of Joseph Henrich (2020) on the deep origins of WEIRD psychology.Footnote 18 Over the last few years, Henrich and his colleagues have argued that WEIRD individuals are statistical outliers on a range of metrics: they are more individualistic and meritocratic, inordinately trusting of strangers, motivated by fairness, and “analytic” in reasoning (Henrich et al. 2010; Muthukrishna et al. 2020). In his recent work, Henrich has argued that the suite of values and motivations associated with WEIRD psychology can be measured by a metric he and his colleagues call the Kinship Intensity Index (“KII”). This is an aggregate measure that combines a range of indicator variables: cousin marriage, the structure of families, residence patterns, marriage structures, and marriage restrictions (particularly around endogamy). In a nutshell, the KII tracks the intensity and influence of kin-based structures (around kinship and marriage) on everyday life: who one sits next to (not that cousin), who counts as a father or uncle (not from that side of the family), who can be married (not from that clan). Kin-based structures also influence cognition—and this is where the KII meets WEIRD psychology.

On Henrich’s account, cultures that rank high on the KII will be those with greater “conformity to peers, deference to traditional authorities, sensitivity to shame, and an orientation toward the collective (e.g., the clan) over oneself.” (Henrich 2020, 198) In populations that rank lower on the KII—where there is rampant exogamy, neolocality, and importantly, a lack of cousin marriage—agents live in “individual-centered worlds” with “greater independence, less deference to authority, more guilt, and more concern with personal achievement.” (Ibid.) The broad causal claim Henrich makes is that WEIRD populations are outliers compared to most other cultures because they have “fewer and weaker” relational bonds, as measured by the KII.

This account is ambitious and is bolted onto an even more ambitious historical narrative. Henrich claims the lower KII scores of Western Europeans were driven by a package of ecclesiastical and legal decisions by the Catholic church. These outlawed cousin marriage to an unusual degree and boosted a package of ideas supporting the individual accumulation of wealth. The historical details of that case are contentious, but not obviously wrong. Whatever its merits, Henrich’s account provides a clear and unambiguous case of demographic cultures featuring in a powerful cross-cultural comparative account, with agential similarity in values grounded in concrete mechanisms of kinship practices.

The demographic skeptic is right to demand that similarities between agents be grounded in causal mechanisms—but this is a demand that many comparative researchers of cultures will be able to answer.


Henrich’s work raises an important consideration. He takes demographic cultures mostly for granted, showing that multiple measures of cultural groups—sometimes operationalized as national cultures and sometimes as ethnolinguistic populations—offer supporting evidence for his clams around WEIRD psychology. But nowhere does Henrich defend a metaphysics of cultural groups. And as the ambivalence about operationalizations suggests, he is open to multiple ontologies to boot. Does this suggest that empirical work explaining differences between cultural group members can be agnostic about broader metaphysical and ontological commitments?

In large part, I think this is so. Comparative social scientific work explains similarities and differences within and between different groups. It relies on the two minimal ontological commitments identified above: that groups are individuatable and qualitatively distinct. Yet these minimal assumptions are all that is needed to support the variety of tools and methods at work in comparative social science.

Okasha (2002) makes analogous points about researchers in the life sciences. Much work in evolutionary biology looks at sets of facts—ecological relationships, geographical distributions, gene flow, genome structure—that make little contact with the genealogical facts used to classify species. Relational essences might be essential for classificatory efforts in the life sciences, but the precise nature of these classifications might do little work in the day-to-day research of many scientists. The same might be true for demographic cultures. The explanations of similarity across groups may appeal to sets of facts—cognitive dispositions, geographical location, network size, background information—that make little contact with the relational facts underwriting classification.

The broader point is this. Demographic skepticism emerges out of an alignment between comparative social science and the folk anthropological model. The latter model posits tight connections between metaphysical, ontological, and explanatory commitments. But these are not necessary for comparative work. The minimal commitments needed to pursue comparative explanations cuts considerable ontological slack; enough to support considerable freedom in the operationalizations and empirical frameworks used by researchers.

The four recommendations I’ve made suggest how such varied frameworks should be developed and defended: one should adopt relational essentialism because it provides a means of individuating cultures that is not committed to implausible “timeless” essences; one should avoid unusual and strong empirical assumptions in one’s work (such as those of the folk anthropological model) unless a case can be made for them; and one should provide clear ontological operationalizations of key terms that can feed into causal-explanatory frameworks. And as I’ve pointed out throughout, a great deal of work in the comparative social sciences already abides by such recommendations.