Dunbar explores how individual cognitive and structural mechanisms, such as grooming-based bonding and social cognition, enable primates to maintain group cohesion and overcome the constraints of living in large social groups. While Dunbar focuses on individual-level mechanisms, we propose that group cohesion and primate social networks should also be understood as emergent properties shaped by higher-level selection and self-organized processes driven by dynamic feedback loops and collective interactions.

Since the beginning of humanity, social relationships have been shaped by pressures affecting survival and interactions (Boyd & Richerson, Reference Boyd and Richerson2004; Harari, Reference Harari2014; Henrich, Reference Henrich2017). The COVID-19 pandemic exemplifies these dynamics, as social distancing mimics adaptive behaviours in the animal kingdom, where individuals modify social interactions to balance pathogen risk and resource access (Romano, MacIntosh, & Sueur, Reference Romano, MacIntosh and Sueur2020). This feedback loop between individual decisions and emergent social structures influences group cohesion through modularity and division of social roles. Darwinian natural selection extends beyond individual traits, incorporating cultural transmission and population dynamics, demonstrating that genetic, epigenetic, and cultural evolution collectively shape these adaptive responses (Ashe, Colot, & Oldroyd, Reference Ashe, Colot and Oldroyd2021; Birch & Heyes, Reference Birch and Heyes2021; Claidière et al., Reference Claidière, Smith, Kirby and Fagot2014; Henrich & McElreath, Reference Henrich and McElreath2003; Jablonka & Lamb, Reference Jablonka and Lamb1998; Nowak, Reference Nowak2006).

A study on human social interactions (Almaatouq et al., Reference Almaatouq, Noriega-Campero, Alotaibi, Krafft, Moussaid and Pentland2020) showed that dynamic networks with high adaptation rates improved group performance in changing environments. High adaptation led to more centralised networks, while less adaptable networks maintained prolonged social learning, reducing errors in stable environments. Network plasticity, as a key adaptive mechanism, refines individual actions. Similarly, adaptive networks in animal societies evolve through multilevel selection, favouring behavioural phenotypes that enhance network efficiency (Cantor et al., Reference Cantor, Maldonado-Chaparro, Beck, Brandl, Carter, He, Hillemann, Klarevas-Irby, Ogino, Papageorgiou, Prox and Farine2021; Fisher & McAdam, Reference Fisher and McAdam2017; Gross & Blasius, Reference Gross and Blasius2007; Sueur et al., Reference Sueur, Romano, Sosa and Puga-Gonzalez2019). Social network evolution including group cohesion and fission-fusion should be examined at multiple levels rather than solely at individual cognition (Fisher & McAdam, Reference Fisher and McAdam2017; Nowak, Reference Nowak2006; Sueur, Reference Sueur2023).

The relationship between network properties and cognition supports the social brain and cultural intelligence hypotheses (Dunbar, Reference Dunbar1998; Muthukrishna et al., Reference Muthukrishna, Doebeli, Chudek and Henrich2018; van Schaik, Isler, & Burkart, Reference van Schaik, Isler and Burkart2012). Network efficiency in primates correlates with neocortex size, suggesting an evolutionary link between social learning and cognitive abilities (Pasquaretta et al., Reference Pasquaretta, Levé, Claidière, van de Waal, Whiten, MacIntosh, Pelé, Bergstrom, Borgeaud, Brosnan, Crofoot, Fedigan, Fichtel, Hopper, Mareno, Petit, Schnoell, di Sorrentino, Thierry and Sueur2014). The ability of macaques to adjust social tolerance and group cohesion post-hurricane to optimise shade access for thermoregulation (Testard et al., Reference Testard, Shergold, Acevedo-Ithier, Hart, Bernau, Negron-Del Valle, Phillips, Watowich, Sanguinetti-Scheck, Montague, Snyder-Mackler, Higham, Platt and Brent2024) illustrates behavioural modulation based on physiological needs. Network topology also predicts brain structure in rhesus macaques (Testard et al., Reference Testard, Brent, Andersson, Chiou, Negron-Del Valle, DeCasien, Acevedo-Ithier, Stock, Antón, Gonzalez, Walker, Foxley, Compo, Bauman, Ruiz-Lambides, Martinez, Skene, Horvath, Unit and Sallet2022) and baboons (Meguerditchian et al., Reference Meguerditchian, Marie, Margiotoudi, Roth, Nazarian, Anton and Claidière2021), highlighting links between individual cognition and social structures (Sueur, Reference Sueur2023; Sueur et al., Reference Sueur, Romano, Sosa and Puga-Gonzalez2019).

Collective social niche construction extends niche construction theory to social networks, where individual interactions shape and are shaped by social structures. Genetic, cultural, and ontogenetic processes drive network topology shifts, enabling populations to adapt dynamically. Cultural evolution, driven by social learning and information transmission, rapidly alters network structures, impacting the adaptive landscape of populations. Genetic selection on personality traits and social behaviours can lead to network topologies enhancing group fitness. Social network evolution is influenced not only by genetic selection but also by ontogenetic development and cultural transmission. Social structures can shift through learning and experience-based adjustments, such as age-related behavioural changes or modifications based on environmental feedback. These evolving networks influence selection pressures, information flow, and resource distribution, affecting individual fitness.

Collective social niche construction integrates social evolution, highlighting the feedback loop between individual behaviours and emergent social organisation. Individual exchanges between populations can facilitate the selection of network topologies, much like phenotypic evolution in sticklebacks (Farine, Montiglio, & Spiegel, Reference Farine, Montiglio and Spiegel2015), where social interactions drive covariance among individuals (Neumann & Bell, Reference Neumann and Bell2023). Multilevel selection favours phenotypic variation, leading to mutualistic benefits and specific network topologies (Costello et al., Reference Costello, Cook, Brodie and Formica2023; Farine et al., Reference Farine, Montiglio and Spiegel2015). In similar ecological conditions, populations favouring modular network structures show higher survival rates during disease outbreaks (Romano et al., Reference Romano, MacIntosh and Sueur2020). Modularity in networks can arise due to high interaction costs, while integrated networks may emerge when beneficial interactions are frequent (Marcoux & Lusseau, Reference Marcoux and Lusseau2013; Romano et al., Reference Romano, Shen, Pansanel, MacIntosh and Sueur2018, Reference Romano, Puga-Gonzalez, MacIntosh and Sueur2024; Sah, Mann, & Bansal, Reference Sah, Mann and Bansal2018). This adaptive structuring highlights the role of both genetic evolution and cultural evolution in shaping the network topology.

Beyond selection, adaptive networks in many species emerge through self-organisation. This principle, seen in complex collectives (Bonabeau et al., Reference Bonabeau, Theraulaz, Deneubourg, Aron and Camazine1997; I. D. Couzin & Krause, Reference Couzin and Krause2003; Puga-Gonzales et al., Reference Puga-Gonzales, Ostner, Schülke, Sosa, Thierry and Sueur2017), underscores the role of local interactions in shaping global network structures. Integrating insights from theoretical morphology, network science, and biological architectures enhances our understanding of adaptive networks. The ‘network morphospace’ framework (Avena-Koenigsberger et al., Reference Avena-Koenigsberger, Goñi, Solé and Sporns2015) categorises networks based on connectivity traits, aiding in the identification of generative rules and constraints. Network geometry (Boguñá et al., Reference Boguñá, Bonamassa, De Domenico, Havlin, Krioukov and Serrano2021), incorporating shortest paths and dynamic processes, reveals symmetries such as fractality and scale invariance, applicable to brain function and social topology. Biological systems exhibit multiscale architecture, where each level – from molecules to social groups – solves distinct problems through collective dynamics (Centola, Reference Centola2022; I. Couzin, Reference Couzin2007; McMillen & Levin, Reference McMillen and Levin2024). This nested structure underscores how collective intelligence drives adaptive functionality across biological and social scales including group cohesion and fission-fusion dynamics as Dunbar proposes.

By integrating niche construction and social network evolution, collective social niche construction provides a comprehensive framework for understanding adaptive connectivity including group cohesion and fission-fusion dynamics in animal societies. This approach emphasises the role of the network topology in enhancing fitness and survival, offering valuable insights into the evolution of sociality. Adaptive networks adjusting group cohesion and group size arise through a combination of genetic selection, cultural transmission, ontogenetic development, and self-organisation, illustrating the complex interplay between individual behaviours and emergent social structures. Understanding these dynamics deepens our knowledge of social evolution and offers new perspectives on the interdependence of cognition, sociality, and environmental adaptation.