It may not be the job of a definition to explain the phenomenon it names. Societies have been named in various ways since before the dawn of sociology as a discipline, with emphasis placed either on shared cultural values or structural components (e.g., Conerly, Holmes, & Tamang, Reference Conerly, Holmes and Tamang2021). Yet, our comprehension of social phenomena has proceeded. What we find problematic about Moffett's proposed definition of societies is the irrelevant role ascribed to social interactions, which in our view set the stage for social identity to arise. We present several arguments supporting this position.

The interdependence of the units comprising any collective is a basic condition of its existence and permanence. Even the boundaries of systems as simple as cell aggregations are established not by an overarching, a priori collective identity but by the possibilities they have for exchanging information about their states and the environment (Levin, Reference Levin2019). Similarly, constraints on how many social relationships can be maintained by an individual (Dunbar, Reference Dunbar1993) and on how a group can maintain its coherence over time are determined by the type of interactions that occur within it (Henzi et al., Reference Henzi, de Sousa Pereira, Hawker-Bond, Stiller, Dunbar and Barrett2007).

Social interactions are the crucial link between societies and the mind of individuals, which we agree with Moffett is an important mechanism underlying group life. But it is precisely research into the social minds of animals that has yielded the best support for the idea that it is relationships, not simply group membership, that animals pay attention to (Aureli & Schino, Reference Aureli and Schino2019). The “social categories” in the mind of baboons, for example, are about matrilineal affiliation or relative dominance rank, independently of the particular individuals involved in these relationships (as has been shown by field playback experiments, e.g., Bergman, Beehner, Cheney, & Seyfarth, Reference Bergman, Beehner, Cheney and Seyfarth2003).

To us, social identity is a result, and not a precursor, of social structure (although we acknowledge that the two concepts are interrelated). Emerging from the overall pattern of social interactions and relationships, the structure of a group is often invariant over generations, because of the fact that there are top-down influences on the types of interactions individuals can engage in as they develop. Dominance hierarchies, assortment patterns between the sexes, kinship effects, and (in humans) cultural norms and values transcend the particular individuals that compose the group at any given time, constraining social interactions and providing a degree of temporal stability to the resulting social structure (Flack, Reference Flack2017; Hinde, Reference Hinde1976). Social identity is one of those individual-level features that is influenced by the social structure in a top-down fashion and reinforces, in most cases, the social structure. For example, group-level features, such as dialects (Cantor et al., Reference Cantor, Shoemaker, Cabral, Flores, Varga and Whitehead2015; Henry, Barbu, Lemasson, & Hausberger, Reference Henry, Barbu, Lemasson and Hausberger2015), behavioral traditions (Laland & Evans, Reference Laland, Evans, Call, Burghardt, Pepperberg, Snowdon and Zentall2017), or even more elusive ones like the “social microbiome” (Archie & Tung, Reference Archie and Tung2015; Sarkar et al., Reference Sarkar, Harty, Johnson, Moeller, Archie, Schell and Burnet2020), are all mediated by social interactions and effectively serve as identity markers. Thus, the social structure based on the social interactions between group members allows dynamic and shared group identities to emerge, rather than some pre-existing and somehow abstract identity being a “soil” for the growth of societies.

We consider networks to be models of societies, not a separate organizational level. Since social interactions are fundamental to societies, networks are preeminent tools for understanding societies. Although one-dimensional networks are too limited to capture what a society is, better representations can be achieved by multiplex network models, which can be constructed in a variety of ways, with weighted links, suprastructures, and modules, and most importantly, with various types of links (representing various types of social interactions) into the same structure (Smith Aguilar, Aureli, Busia, Schaffner, & Ramos-Fernandez, Reference Smith Aguilar, Aureli, Busia, Schaffner and Ramos-Fernandez2019). Furthermore, addressing the question of how groups form and maintain continuity in time and space, theoretical explorations of the dynamic features of networks show that their structure can be invariant over long temporal scales in spite of the addition or deletion of nodes (e.g., Moore, Goshai, & Newman, Reference Moore, Goshai and Newman2006; Murase, Jo, Török, Kertész, & Kaski, Reference Murase, Jo, Török, Kertész and Kaski2015).

The way in which Moffett considers social identity seems not to allow for those cases in which an individual might have more than one sense of identity, on different levels of social organization. People, for example, have a sense of belonging to the neighborhood, in addition to the state, and to the country they live in. Dolphins could have strategies for interacting not only with their close allies but also with the second- or third-order units in which their alliances participate (Connor, Reference Connor2007). If we force the existence of one important level (society) based on a feature that we believe is only related to that level (identity sense), then we lose sight of the relevance of the other levels and the potential mechanisms that allow them to exist. These multiple levels of organization are not purely casual groupings which mean nothing in social terms for the animals. There is clear evidence of consistent subgroup memberships in various species of mammals (Cantor et al., Reference Cantor, Shoemaker, Cabral, Flores, Varga and Whitehead2015; Ramos-Fernandez et al., Reference Ramos-Fernandez, King, Beehner, Bergman, Crofoot, Di Fiore and Boyer2018; Wittemyer, Douglas-Hamilton, & Getz, Reference Wittemyer, Douglas-Hamilton and Getz2005) and birds (Papageorgiou & Farine, Reference Papageorgiou and Farine2021). We should thus consider that an individual may actually have multiple identity senses, one for each level of organization to which in reality it belongs.

Social identity is certainly an important component of culture, and we humans obviously devote plenty of resources to signaling and maintaining our identity, which allows us to collaborate and coordinate our actions in complex cultural settings through the signaling of norms, values, and group traditions (Smaldino, Reference Smaldino, Love and Wimsatt2019). But even here, whether the very nature of societies is built around their social identity begs the question of what is the role of social interactions in these collaborations and coordinations. We contend that membership is not the crucial aspect of ordinary human existence, it is the social interactions at the heart of societies, based on collaboration and coordination, which underlie our various senses of belonging.