We appreciate Moffett's call for an over-arching definition of society. However, we are not entirely convinced that his more human-focused definition can be neatly extrapolated to other primate/animal species. We will comment on how exactly the components of Moffett's definition of a society are reconcilable with multilevel societies, in which smaller social entities (“core units”) are embedded within multiple levels of higher-level entities (“intermediate levels,” “upper levels,” and “apex levels,” sensu Grueter et al., Reference Grueter, Qi, Zinner, Bergman, Li, Li and Swedell2020). This is especially important considering the relevance of multilevel societies to hominin/human social evolution (Chapais, Reference Chapais2013; Grueter, Chapais, & Zinner, Reference Grueter, Chapais and Zinner2012; Swedell & Plummer, Reference Swedell and Plummer2012, Reference Swedell and Plummer2019), the very societies upon which Moffett bases his arguments.

First, Moffett makes it clear that his definition of a society does not generally accommodate multilevel societies and he does not think species can “simultaneously possess two tiers that can be described as societies” except under restricted circumstances. However, this multiple membership phenomenon is precisely the point of multilevel societies, that is, we could in fact view them as “societies within societies” due to the perceived shared membership in these multiple levels (e.g., Schreier & Swedell, Reference Schreier and Swedell2009, Reference Schreier and Swedell2012).

Second, it is not entirely clear what main criterion should define a society in the case of multilevel systems. If it is individual recognition to be part of an “identity group” or “perceive one another as belonging together,” then, as Moffett states, the core unit of a multilevel society – that is, the socially bonded breeding unit – may indeed be equivalent to a society, but only in those multilevel societies wherein the degree of individual recognition abilities beyond the confines of the core unit is unclear. For example, Bergman (Reference Bergman2010) has shown that gelada males do not show vocalization-based recognition of individual males outside their core unit. However, only males were tested in this playback experiment, and it is possible that females recognize a wider range of individuals in this or other multilevel systems (Mac Carron & Dunbar, Reference Mac Carron and Dunbar2016). It is also possible that males do have the ability to recognize extra-unit individuals but are not motivated to so in the experimental setting, or are able to recognize one another by sight even if not by vocalizations. Moreover, in the multilevel system of hamadryas baboons, individual recognition extends to multiple layers of society, far beyond the core unit (e.g., Sigg, Stolba, Abegglen, & Dasser, Reference Sigg, Stolba, Abegglen and Dasser1982). Other examples where vocal recognition beyond a single social tier has been documented include African elephants (McComb, Moss, Sayialel, & Baker, Reference McComb, Moss, Sayialel and Baker2000), bottlenose dolphins (King et al., Reference King, Friedman, Allen, Gerber, Jensen, Wittwer and Krützen2018), and vervet monkeys (Cheney & Seyfarth, Reference Cheney and Seyfarth1982).

Third, if we use shared identity to define a society then we must also look beyond the core unit. In hamadryas baboons, for example, multiple core unit males negotiate a common direction of departure from their sleeping cliffs (Kummer, Reference Kummer1968). Such collective decision making and coordination of activities across core units are common in multilevel systems and contribute to their stability (Maeda, Sueur, Hirata, & Yamamoto, Reference Maeda, Sueur, Hirata and Yamamoto2021; Wu et al., Reference Wu, Zhao, Li, Su, Zhao, Ren and Pan2023). Moffett writes that in case of geladas “units share nothing with those greater collectives other than the habit of moving more or less across the same general ground,” However, they do seem to have a vested interest in keeping intruders at bay: Leader males of the core units sometimes join forces in warding off bachelors (Wrangham, Reference Wrangham1976). These cooperative interactions can also extend to higher levels of society, for example, in hamadryas baboons (Abegglen, Reference Abegglen1984; Kummer, Reference Kummer1968; Sigg et al., Reference Sigg, Stolba, Abegglen and Dasser1982). Another revealing observation is that core unit males in hamadryas baboons show “respect for possession” of females, an inhibition of competition that prevents them from stealing females from other males (Kummer, Götz, & Angst, Reference Kummer, Götz and Angst1974). Such respect would easily slide into more frequent aggressive competition over females if the core units represented the society.

Fourth, if, as Moffett states, a group “must extend beyond a simple, immediate family to be considered a society,” then we must define a society at a higher level of organization than the core unit. This element of the definition of a society is incongruent with the designation of the core unit as a society in multilevel systems because core units are reproductive units and can thus be construed as polygynous “families” sensu lato.

Fifth, another component of Moffett's definition of a society revolves around exclusive dominion over the same space at a given time and concern for minimizing encroachment by outsiders. If we apply this spatial criterion to multilevel systems, then we must define the society as the upper level, that is, the “band” in many species, as it is this level that largely maintains a shared activity space to the exclusion of other such units.

Given his comments about geladas, Moffett seems to favor considering the core unit to constitute a society in multilevel systems. While individual units do have some of the hallmarks of a mini-society, they are in fact just a constituent of a larger society. Therefore, if one must make a choice of just one level, then we feel it is currently more parsimonious to ascribe the upper level, that is, the band, as equivalent to a society in most multilevel societies. That said, in the case of geladas the society could also lie at an intermediate level between the core unit and the band, the “team,” a set of two to three core units (Mac Carron & Dunbar, Reference Mac Carron and Dunbar2016).

As an additional point, Moffett uses what we know about geladas to apparently argue that the higher tiers are of questionable social significance in all multilevel systems. We disagree. While the very highest level (apex level) is typically an aggregation resulting from shared attraction to localized resources and not a genuine social entity, the upper level, just below the apex level, is usually cohesive, longitudinally stable, and individualized (Grueter et al., Reference Grueter, Qi, Zinner, Bergman, Li, Li and Swedell2020, Reference Grueter, Qi, Zinner, Bergman, Li, Xiang and Fischer2021).

In sum, our commentary reveals how a proper understanding of the intricacies of multilevel societies is critical for informing and enriching discussions about what constitutes a society.