Insights From Evolutionary Anthropology on the (Pre)history of the Nuclear Family

Introduction

In The History of Human Marriage, Westermarck (1921) wrote,

Monogamy is the only form of marriage that is permitted among every people. Wherever we find polygyny, polyandry, or group-marriage, we find monogamy side by side with it. On the other hand, it is also in many cases the only form of marriage which is permitted by custom or law. This may be due to the mere force of habit ; or, possibly, to the notion that some men must not appropriate a plurality of wives when others in consequence can get none at all ; or to the feeling that polygyny is an offence against the female sex ; or to the condemnation of lust. (p. 104)

This excerpt, from the third volume of Westermarck’s monumental opus, illustrates some of the problems that have marred the study of kinship and marriage systems since its inception in the 19th century. For example, group-level norms are confounded with individual behavior, such that variation in behavioral strategies within societies is contrasted with variation across societies. In the following decades, these difficulties precipitated the field into “a subject in which specialists talked only to each other (and some talked only to God)” (Fox, 1967/1983, p. 10). Couched in arcane language and obscure conceptual categories, its basic assumptions came under severe scrutiny during the second half of the 20th century; by the end of the century, anthropology had essentially repudiated what it once viewed as its “basic discipline” (Fox, 1967, p. 10; see discussion in Holy, 1996, pp. 1-8).

As in other domains in the behavioral sciences, an evolutionary approach offers a unified conceptual and analytical framework to overcome these difficulties (see discussion in Mesoudi, 2011). After all, kinship and marriage systems represent “one form of assortative mating which has the interesting component of culturally defined categories” (Fox, 1967/1983, p. 3). Thus, the study of these features of social organization can take advantage of the suite of theoretical and methodological tools used in the analysis of social behavior, both human and nonhuman. In turn, to the extent that networks based on kinship and marriage created the social niche in which our species evolved, understanding their workings becomes crucial for understanding the evolution of human behavior. Ultimately, a comprehensive evolutionary account of the human phenomenon must be able to explain why individuals organize relatedness and reproduction in the ways they do (Fortunato, 2015).

Anthropologists have generally resisted application of the evolutionary paradigm to the analysis of kinship and marriage systems, the classic complaint being that “social” and “biological” categories overlap but are not identical (Fox, 1967/1983, p. 3; see e.g., Sahlins, 2013, 1976). At the same time, those advocating an evolutionary approach have perhaps failed to provide compelling explanations to account for these discrepancies. For example, one of the most powerful insights afforded by the evolutionary approach to the study of human behavior is the asymmetry that characterizes sexually reproducing species, deriving from the higher potential rate of reproduction of males relative to females (see review in Brown, Laland, & Borgerhoff Mulder, 2009). The cross-cultural prevalence of polygynous marriage is readily explained on this notion; why a small but substantial proportion of societies prescribe monogamous marriage is however not clear (Fortunato & Archetti, 2010). A norm enforcing monogamous marriage is especially puzzling where mating is effectively polygamous (e.g., through pre- and extramarital sexual relationships). The discrepancy between the “social” category of marriage and the “biological” category of mating would seem to disprove claims to the relevance of evolutionary explanations to these aspects of human social life (Fortunato, 2015).

My aim in this article is to elucidate the relevance of the evolutionary paradigm to the study of kinship and marriage systems. I begin with a discussion of conceptual and methodological issues that arise in approaching human social systems from an evolutionary perspective. I then narrow the focus on key tools used in contemporary cross-cultural research within evolutionary anthropology. Next, as a case study, I provide an overview of work aimed at reconstructing the (pre)history of the nuclear family in Indo-European-speaking societies. This exercise can help us evaluate theoretical claims to the centrality of the arrangement to European social organization and, by extension, to “Western” social organization. Such claims abound the in the social sciences, typically outside anthropology (Goody, 1996; Sear, 2016; Smith, 1993). The results also speak to renewed interest from across the social sciences in how family systems, and their historical trajectories, have contributed to shaping the contemporary world (e.g., Alesina & Giuliano, 2010; Duranton, Rodríguez-Pose, & Sandall, 2009).

Reconstructing the (Pre)history of the Nuclear Family

Having discussed some key conceptual and methodological issues that arise in approaching kinship and marriage systems from an evolutionary perspective, I now illustrate the approach with a brief overview of work aimed at reconstructing the (pre)history of the nuclear family in societies speaking Indo-European languages (see Fortunato, 2009, Chapter 6, for details).

It is useful to start with definitions, as these vary across disciplines, and between technical and everyday usage. In anthropology, the nuclear family designates a family unit comprising parents and their dependent children (Fox, 1967/1983, p. 36; Keesing, 1975, p. 150; Parkin, 1997, p. 28); technically, therefore, this arrangement can be associated with either monogamous or polygamous marriage (Murdock, 1949, pp. 1-2; see discussion in Fox, 1967/1983, pp. 36-40). In other disciplines, as in everyday usage, the concept is typically used rather vaguely to designate the small isolated (i.e., monogamous and nonextended) family type characteristic of “Western” social organization (Smith, 1993). I enclose instances of the latter definition in quotes.

Is the “nuclear family,” so defined, a coherent arrangement, possibly conveying an advantage over alternative arrangements? For example, within anthropology it has long been suggested that both monogamy and neolocality serve to reduce potential conflicts of interests between spouses (e.g., Lowie, 1920; Murdock, 1949). Monogamous marriage eliminates the possibility of additional spouses and their children, whereas neolocal residence limits biases in access of husband and wife to their respective kin. Within the Tinbergen/Mayr framework (see “Levels of Explanation” section), this provides a possible explanation in terms of function (e.g., Alexander, 1987, pp. 70-73). More broadly, monogamy and neolocality have been variously linked to features of social organization viewed as markers of “complexity,” including industrialization, commercialization, modernization, and so on (Levinson & Malone, 1980, pp. 37-38; for example, Blumberg & Winch, 1972; de Leeuwe, 1971; Ember, 1967; Lee, 1979; McNett, 1973; Osmond, 1965, 1969; Sheils, 1971). The co-occurrence of the two traits in the “nuclear family,” and the preeminence of this arrangement in “Western” societies, would seem to uphold this view (Smith, 1993).

Enter what has come to be known in anthropology as “Galton’s problem”—the issue of the statistical nonindependence of human societies (see “Explaining Cross-Cultural Variation” section). The observation that two traits tend to co-occur provides necessary but not sufficient evidence that they are functionally related—that is, that they convey an advantage over alternative combinations of traits. Societies may share practices through descent from a common ancestor and/or through contact, hence the co-occurrence of traits in a cross-cultural sample may effectively be a historical artifact. Conversely, if changes in one trait tend to be associated with changes in the other trait, then this can be taken as evidence that the two traits coevolve, indicating that they may be functionally related. For example, it may be that societies with neolocal residence are more likely to acquire monogamous marriage from neighboring societies. Alternatively, it may be that where monogamous marriage emerges as an independent innovation, it is more likely to “stick” in the presence of neolocal residence. Evidence along these lines would suggest that the “nuclear family” is indeed a coherent arrangement, possibly conveying an advantage over alternative ones, and justifying its preeminent role in “Western” social organization.

Discriminating between these scenarios in the absence of detailed information about the history of the traits is problematic. Phylogenetic comparative methods offer one solution (see “Phylogenetic Comparative Approach” section). Using a phylogenetic tree to represent how the societies are related by way of descent from a common ancestor, they infer likely evolutionary scenarios that produced the observed distribution of traits across the societies. In testing for coevolution between two traits, the aim is to determine whether they have tended to change in correlated fashion over time. I used this approach to investigate the interplay between marriage and residence strategies in the history of Indo-European-speaking societies (Fortunato, 2009). Related results are reported in Fortunato (2011a, 2011b, 2011c), where I used a phylogenetic comparative approach to reconstruct the history of the traits.

There are two steps to a phylogenetic comparative analysis. The first is to obtain a phylogenetic tree, or a sample of trees, and the matching comparative data. The second step is the statistical analysis. I used a sample of trees generated by Pagel, Atkinson, and Meade (2007) through the application of a phylogenetic tree-building method to vocabulary data for Indo-European languages; the data were collated by Dyen, Kruskal, and Black (1992). In previous analyses of these data, Gray and Atkinson (2003) used the same approach to test between the two main competing hypotheses for the origin of the language family (see also Atkinson & Gray, 2006a, 2006b; Atkinson, Nicholls, Welch, & Gray, 2005; Bouckaert et al., 2012), providing support for the scenario proposed by Renfrew (1987). This posits the expansion of Indo-European languages from Anatolia with the spread of agriculture beginning around 7000 to 6000 bce. The alternative scenario, proposed by Gimbutas (1973a, 1973b), places the homeland of the language family in the Pontic steppes north of the Black Sea, with expansion driven by nomadic horse-riding pastoralists starting after 4000 bce (see discussion in Diamond & Bellwood, 2003).

With the tree sample in hand, I matched speech varieties in the vocabulary database with societies in the Ethnographic Atlas, to obtain comparative data on marriage and residence strategies. This resulted in a sample of 27 societies, with marriage scored as monogamous versus polygamous, and residence as neolocal versus non-neolocal (Figure 1). As expected, monogamy and neolocality tended to co-occur in the sample, with monogamy present in 93.8% of neolocal societies (15 of 16 societies) versus 27.3% of non-neolocal ones (3 of 11 societies).

OPEN IN VIEWER

Figure 1.

Diagram summarizing Pagel, Atkinson, and Meade’s (2007) sample of 750 trees, pruned to 27 Indo-European speech varieties plus the outgroup Hittite, with comparative data on marriage and residence strategies.

This pattern is suggestive of a functional association between the two traits, but mapping the comparative data onto the phylogenetic trees reveals clustering in specific “branches,” corresponding to linguistic subgroups (Figure 1). With few exceptions, monogamy and neolocality are found in the European branches, and their co-occurrence is exclusive to them. Overall, the distribution of marriage and residence strategies at the “tips” of the trees can be accounted for through a limited number of instances of correlated change in the traits. These do not provide sufficient evidence that the traits have tended to change in correlated fashion. In other words, the co-occurrence of monogamy and neolocality in this sample can be accounted for by the patterning of descent of the societies from a common ancestor, which likely practiced monogamous marriage (Fortunato, 2011a) and neolocal residence (alongside virilocal residence as the prevailing mode; Fortunato, 2011b). In testing for coevolution, two models were fitted to the data: a model of independent evolution, in which the traits change independently of each other, and a model of dependent evolution, in which they change in correlated fashion. Statistically, the dependent model did not provide a better fit to the data than the independent model; hence, there was no support for a scenario of correlated change. Although the investigation of a larger sample may provide stronger support for the hypothesis of coevolution, this seems unlikely given the tight clustering of marriage and residence strategies within linguistic subgroups (Fortunato, 2009).

Toward a Biologically Based Social Anthropology

Previous theorizing about human family systems has often assumed, or at least implied, a functional association between monogamous marriage and neolocal residence, in what is commonly referred to as the “nuclear family.” This assumption is based on the observation that the two practices tend to occur together, and it underlies theories that make this family type central to “Western” social organization. However, the co-occurrence of traits across societies may reflect the history of the traits, rather than a functional relationship between them (Fortunato, 2008).

I have provided an overview of phylogenetic comparative analyses, applied to cross-cultural data, aimed at reconstructing the history of marriage and residence strategies in Indo-European-speaking societies, with a focus on the interplay between monogamy and neolocality (Fortunato, 2009, 2011a, 2011b, 2011c). Results show that any evidence of association between the two traits cannot be untangled from the patterning produced by the phylogenetic relationships among societies in the sample. These results challenge the notion, prevalent in history and sociology of the family and in demography, of the centrality of the “isolated nuclear family” to the social organization of Europe, and of western Europe in particular (Goody, 1996; Smith, 1993; see e.g., Hajnal, 1965, 1982). They imply that the widespread co-occurrence in the region of two key elements of this family type—monogamy and neolocality—may be an artifact of history, rather than the result of “the individualistic or nuclear tendencies of the European family system” (Smith, 1993, p. 328). Consistently, phylogenetic ancestral state reconstructions reported in Fortunato (2011b) show that neolocal residence occurred alongside virilocal residence throughout the history of Indo-European-speaking societies; by definition, virilocality results in extended family organization.

As noted above, the occurrence of the “nuclear family” is often linked to features of social organization viewed as indicators of “societal complexity” and “modernization,” based on the prevalence of this family type among the “complex,” “modern” societies of Europe. Yet the historical evidence (discussed in Smith, 1993; see e.g., Laslett, 1977) and the phylogenetic reconstructions in Fortunato (2011a, 2011b) point to an “early” origin of this family type across European societies, and across Indo-European-speaking societies more generally. Unless social and/or ecological determinants can be identified that account for both its distribution and its history, then the occurrence of this arrangement across Europe is best viewed as a result of the nexus of descent linking societies in the region. Detailed historical demographic data (e.g., Szołtysek & Gruber, 2016) will likely play a role in identifying possible social and ecological determinants.

An “early” origin of this family type also suggests revisiting explanations that ascribe the “nuclear tendencies” of the European family to the influence of Christian ideology on kinship and marriage systems (Smith, 1993; for example, Goody, 1983). The available evidence is, at present, anecdotal; detailed historical and ethnographic data will be required to formally evaluate the alternative pathways that may have produced the posited associations between Christianization and specific features of social organization (see discussion in Fortunato, 2009).

Finally, together with previous analyses (Fortunato, Holden, & Mace, 2006; Fortunato & Mace 2009), the reconstructions of Indo-European social organization by Fortunato (2011a, 2011b) point to a society practicing monogamy, virilocality with neolocality, and dowry. On the ethnographic evidence, this type of social structure is compatible with interpretations of the linguistic evidence that attribute to Indo-European society an economy based on domesticated plants and animals, with use of the plow, a settled lifestyle, and possibly some form of social ranking based on social status differentiation (Mallory & Adams, 2006, p. 284). The reconstructions seem harder to reconcile with interpretations of the linguistic evidence that tend to emphasize elements pointing to a pastoral economy and a nomadic lifestyle; these interpretations provide the basis for the traditional view that Indo-European languages propagated across Eurasia with warlike nomadic pastoralists (e.g., Anthony, 2007; Gimbutas, 1991; Mallory, 1989). In fact, the traditional view has strongly influenced reconstructions of other aspects of early Indo-European social life; in turn, these reconstructions have been used to substantiate the traditional view of Indo-European origin (Renfrew, 1987, pp. 260–262). The analyses in Fortunato (2009, 2011a, 2011b) avoid this circularity because the inferences are derived entirely from the cross-cultural data, in conjunction with the model of population history captured by the linguistic evidence; consequently, the inferential process is relatively assumption-free with regard to early Indo-European social organization. One implicit assumption is that practices observed in the “ethnographic present” can be used to characterize early Indo-European social structure. The analytical categories used in these analyses are based on the simplest possible conceptualization of marriage and residence strategies; as such, they apply to the range of social systems found in the ethnographic record. They are therefore likely to apply also to the range of social systems involved in the history of dispersal of Indo-European-speaking societies from a common ancestor.

Overall, the findings reviewed here emphasize the importance of using systematic and, where possible, explicitly historical approaches to cross-cultural comparison. The formulation of analytical categories that are theoretically justifiable and empirically useful is one of many advances that derive from extension of the philosophical, theoretical, and methodological foundations of the evolutionary paradigm to the study of kinship and marriage systems (see “Introduction” section). More generally, placing variation in human social systems within this paradigm offers a unique opportunity for progressing the study of human behavior toward what Hughes (1988) envisioned as “a biologically based social anthropology” (p. 21)—a discipline focused on understanding the social system of the human species, while at the same time situating human behavior within a broad cross-specific perspective.

A “mature science of society” will comprise general principles applicable to any species, and specific principles applicable to individual taxonomic groups (Hughes, 1988, p. 141); together, they will need to account for the ultimate causes of behavior and for its proximate determinants, including any interactions between genetic and nongenetic pathways for the transmission of information (Rogers, 1988; see “Levels of Explanation” section). Given the emphasis on cultural transmission in the study of human behavior (e.g., Boyd & Richerson, 1985; Cavalli-Sforza & Feldman, 1981), this will likely involve “a mutual transformation” of biology and anthropology (Wilson, 1979). In turn, this will represent an indispensable step toward integration of evolutionary approaches to human behavior (see Laland & Brown, 2011), culminating in the development of a biologically based social anthropology.