Ants in anonymous societies (Moffett, Reference Moffett2012) do not need to recognise other members individually. Despite of lacking complex cognition, ants can form billion-sized societies comparable to those of humans. Anonymous societies also exist in mammals. Spotted hyenas live in societies with the same complexity and size as baboons (Holekamp, Reference Holekamp2007), and they can also recognise individuals anonymously (Moffett, Reference Moffett2019, p. 72). However, their behaviours are far less flexible than those of baboons (Holekamp, Reference Holekamp2007). These seem to challenge the social brain hypothesis (Dunbar, Reference Dunbar1993), one of the evolutionary hypotheses that link sociality with cognition (Bergman & Beehner, Reference Bergman and Beehner2015), because social complexity is no longer a good predictor of cognitive complexity.

Some scholars consider anonymous societies as exceptions or opposite to the social brain hypothesis. Dunbar and Shultz (Reference Dunbar and Shultz2023) argued that species living in large and anonymous societies are off-point in considering social–brain relationships. O'Donnell et al. (Reference O'Donnell, Bulova, DeLeon, Khodak, Miller and Sulger2015) proposed the distributed cognition hypothesis as an alternative to the social brain hypothesis, in which individual wasps reduce rather than increase their cognitive investment in response to social challenges, drawing on collective intelligence. Instead, Bergman and Beehner (Reference Bergman and Beehner2015) claimed that measuring social complexity should take into account “actual” social interactions, which is the opposite of the definition of society by Moffett. That is, it was referred to as the number of differentiated relationships, “the number of consistently different interactions that are seen. If members of a species treat all conspecifics exactly the same, the number of differentiated relationships is 0” (Bergman & Beehner, Reference Bergman and Beehner2015, p. 205). On this account, insect colonies can be considered as extended phenotypes of the reproductives (O'Donnell et al., Reference O'Donnell, Bulova, DeLeon, Khodak, Miller and Sulger2015). This suggests that the complexity in super colonies of billions (Moffett, Reference Moffett2012) may not be far comparable to that of other primates.

This solution proposed by Bergman and Beehner (Reference Bergman and Beehner2015) emphasises actual cognitive participation in social interactions, or real social skills involved (Dunbar & Shultz, Reference Dunbar and Shultz2023). We agree with Moffett that “societies are distinct from social networks” (introduction, para. 2). We hereby provide three examples that more clearly illustrate how this solution reveals the hidden social complexity masked under superficial social networks.

First, the social cognition of corvids may be comparable to that of primates, but most of them live in monogamous interaction networks. While their social complexity may be seemly simple, maintaining long-term pair bonds is thought to require complex relationship intelligence (Emery, Seed, Von Bayern, & Clayton, Reference Emery, Seed, Von Bayern and Clayton2007).

Second, we consider a modification of the social brain hypothesis, the social complexity hypothesis for communication (Freeberg, Dunbar, & Ord, Reference Freeberg, Dunbar and Ord2012), which states that complex societies lead to more complex communication systems. Most cetaceans are thought to support the hypothesis (Marino, Reference Marino, Di Sciara and Würsig2022), and humpback whales are a special case. Although male humpback whales live in individual-based societies, they exhibit complex hierarchical song structures. Dunlop (Reference Dunlop, Clark and Garland2022) revealed the hidden complexity of their communication networks. Despite of the apparent simplicity of their social networks, their intricate communication networks may help shape their complex vocal systems. Sexual pressure could be a primary reason for this complexity. Male humpback whales need to accurately send acoustic signals to females in an ever-changing environment in order to prevent competitors from eavesdropping.

Third, this hidden complexity can be more profound. The social brain hypothesis, originated from primate data (Dunbar, Reference Dunbar1993), does not adequately explain why great apes, which do not exhibit a high degree of social complexity, possess a significant level of encephalisation. The cultural intelligence hypothesis proposed by Whiten and van de Waal (Reference Whiten and van de Waal2017) suggests that the brain expansion of great apes can be better understood if their cultural complexity is taken into consideration. This hypothesis underscores the importance of social learning in shaping social complexity. By the same token, the ecological intelligence hypothesis, emphasising the importance of foraging complexity for cognitive complexity, should be treated as a complementary rather than competing hypothesis to the social brain hypothesis. This is because some seemingly foraging skills may possess important social functions, and vice versa (Rosati, Reference Rosati2017).

Previous studies have declared the difficulty of measuring social complexity (e.g., Hobson, Ferdinand, Kolchinsky, & Garland, Reference Hobson, Ferdinand, Kolchinsky and Garland2019; Kappeler, Reference Kappeler2019), and for the first time, we introduce a concept of jaggedness in measuring social complexity. Rose (Reference Rose2016, p. 80) proposed the jaggedness principle to explain the human tendency to simplify jagged phenomena. For example (Rose, Reference Rose2016, pp. 89–90), participants who achieve the same score on an IQ test can have vastly different scores on specific ability tests (e.g., jigsaw puzzles, number memory and vocabulary). This principle highlights huge variances in scores of different and weakly correlated sub-variables among individuals. Such a jaggedness principle in individual sciences also applies to social complexity studies.

To illustrate this principle more intuitively, we provide a modified version based on the social complexity in Figure 1. Simply put, two species assumed to have considerable cognitive capabilities cannot be described unidimensionally (e.g., group size) in terms of whether their societies are more or less complex when breaking down social complexity into multiple sub-variables, and vice versa. This suggests that social complexity research may yield opposite results due merely to differences in descriptive methods.

Figure 1.

Jagged profiles for social complexity suggest that any social feature alone is difficult to be as an indicator (adapted from Rose, Reference Rose2016, p. 81).

Some recent work (Raviv, Peckre, & Boeckx, Reference Raviv, Peckre and Boeckx2022) has recognised that the jaggedness in communicative complexity can cause conceptual misunderstanding, leading to inconsistent conclusions between human and non-human communication. This is due to the inconsistencies in the analysis units of communicative complexity (i.e., sub-variables), such as repertoire sizes versus compositionality. The distributed cognition hypothesis may be a confusion of cognitive complexity (see Traniello & Avarguès-Weber, Reference Traniello and Avarguès-Weber2023).

Finally, it is also important to discover local configurations that could reveal the hidden social complexity in line with the jaggedness principle. A similar approach has been applied to investigate the sociolinguistic paradigm shift (Eckert, Reference Eckert2012). The second wave of sociolinguistics demonstrated the importance of local configurations over macro-social features. For example, students' social affiliations appear to be better predictors of linguistic variation than their social classes. This causal and dynamic perspective helps recognise the jaggedness of social complexity and refine Moffett's definition of society. Nonetheless, we need more evidence to (dis)prove the social brain hypothesis.