We agree with Dunbar that for primates: i) Living in large groups creates stresses arising from managing a larger set of differentiated social relationships; ii) Managing weak ties in groups in the upper grade cannot be achieved through direct or indirect grooming relationships and therefore requires more sophisticated cognitive mechanisms; iii) Communication plays a key role in managing these differentiated social relationships by helping primates read others’ intentions (Section 5). A key question posed by Dunbar (Section 3.2.3) is how specialised forms of social cognition are actually used to manage social relationships. We argue that one key function of this specialised cognition is to allow for flexible and intentional communication which increases the predictability of conspecifics’ behaviour, thereby reducing the physiological stresses associated with group living (Seyfarth & Cheney, Reference Seyfarth and Cheney2018; Roberts et al., Reference Roberts, Dunbar and Roberts2022). For signallers, communication is used to influence the receivers’ behaviour and therefore make it more predictable, whilst for receivers, communication allows them to better predict the future behaviour of signallers, especially behaviour directed towards themselves (Seyfarth & Cheney, Reference Seyfarth and Cheney2018). As an unpredictable social environment is a key factor in causing stress (Sapolsky, Reference Sapolsky2021; Seyfarth & Cheney, Reference Seyfarth and Cheney2018, Roberts, Dunbar, & Roberts, Reference Roberts, Dunbar and Roberts2022), any reduction in the uncertainty of conspecifics’ behaviour associated with communication will be beneficial for both signallers and receivers.

Primates use a wide variety of communication to help meet the demands of living in complex social groups, including vocalisations, gestures and facial expressions (Seyfarth & Cheney, Reference Seyfarth and Cheney2018). For both vocalisations (Fischer, Reference Fischer2017) and gestures (Roberts et al., Reference Roberts, Vick, Roberts, Buchanan-Smith and Zuberbühler2012a), there is flexibility in both the production and usage according to the social context (Roberts, Vick, & Buchanan-Smith, Reference Roberts, Vick and Buchanan-Smith2012b). For example, across a number of primate species, a range of low-amplitude vocalisations (grunts) are used to signal affiliative intent when approaching another conspecific. This increases the predictability of behaviour between interacting dyads when they are in close proximity, which leads to vulnerability to potential aggression. In female chacma baboons (Silk, Seyfarth, & Cheney, Reference Silk, Seyfarth and Cheney2016) and olive baboons (Silk et al., Reference Silk, Roberts, Städele and Strum2018), production of grunts on approach increased the likelihood of affiliative interaction and were used less frequently with strong bonds (mothers and daughters) and more frequently with weaker bonds (sisters and non-kin), where the risk of aggression is higher. This demonstrates the value of communication in increasing the predictability of behaviour of weak bonds for both signallers and receivers of communication.

In addition to vocal communication, gestural communication - voluntary movements of the arms, head, body postures and locomotory gaits – plays a particularly important role in managing weak ties and increasing the predictability of conspecifics’ behaviour, as there is greater flexibility and intentionality in the use of gestures compared to vocalisations among the great apes (Tomasello & Call, Reference Tomasello and Call2019; Roberts & Roberts, Reference Roberts and Roberts2020). Intentional gestural communication – communication that is targeted at a specific audience to produce a specific response – is a key feature of gestural communication (Roberts & Roberts, Reference Roberts and Roberts2020; Tomasello & Call, Reference Tomasello and Call2019) that appears to play a central role in managing the demands of group living (Roberts & Roberts, Reference Roberts and Roberts2020). By targeting communication to a specific individual through body posture, gaze and use of the appropriate modality, and persisting and elaborating in that communication until the desired response is achieved, for the signaller intentional gestural communication provides a way of increasing the probability of the desired response from the receiver (Tomasello & Call, Reference Tomasello and Call2019). For receivers of communication, intentional communication makes it clearer that the signal is directed specifically at them and is personally relevant to them, enabling them to detect and respond appropriately to the signal in a complex social environment with competing demands for their attention (Roberts & Roberts, Reference Roberts and Roberts2020).

In larger social groups, using communication to increase the predictability of conspecifics behaviour becomes more important as there are more weak ties to manage, leading to a greater risk of misreading intentions and aggression (Section 4). Flexible use of communication with these weak ties can effectively signal both aggressive and affiliative intentions, facilitating group cohesion and avoiding physical aggression (Silk et al., Reference Silk, Seyfarth and Cheney2016; Silk et al., Reference Silk, Roberts, Städele and Strum2018; Damjanovic, Roberts, & Roberts, Reference Damjanovic, Roberts and Roberts2022). In these larger groups, more complex communication repertoires - systems that contain a larger number of functionally distinct elements (Freeberg, Dunbar, & Ord, Reference Freeberg, Dunbar and Ord2012) – enables a more nuanced understanding of conspecifics intentions. For example, combining signals across different modalities (Heesen et al., Reference Heesen, Austry, Upton and Clay2022), or using sequences of vocalisations (Leroux et al., Reference Leroux, Schel, Wilke, Chandia, Zuberbühler, Slocombe and Townsend2023) or gestures (Roberts & Roberts, Reference Roberts and Roberts2019), can increase the efficiency of the communication signal. Further, social stress can itself impair social bonding by disrupting attention allocation and conscious awareness, leading to stimulus-driven processing where individuals react primarily to the immediate stimulus features of the communication signal, rather than integrating wider contextual information (Roberts et al., Reference Roberts, Dunbar and Roberts2022). Intentional communication directed at specific receivers (e.g. through eye gaze) mitigates these effects by enhancing attention allocation and conscious awareness, enabling the receiver to integrate broader contextual information (e.g. prior relationship, behavioural context, audience) to better interpret the signaller’s intentions (Roberts et al., Reference Roberts, Dunbar and Roberts2022).

This communication is underpinned by a suite of cognitive skills identified in Section 3.2.3, including inhibition and executive functioning. For signallers, these skills enable flexibility in signal production according to both the receiver and wider audience (e.g. selective production of copulation calls; Townsend, Deschner, & Zuberbühler, Reference Townsend, Deschner and Zuberbühler2008), and adjusting the signal according to the visual attention of receivers (Tomasello & Call, Reference Tomasello and Call2019). For receivers, detecting whether communication is directly relevant to them, inferring the goal of signaller and deciding on an appropriate response also requires executive function skills (Roberts & Roberts, Reference Roberts and Roberts2020). To conclude, a key selection pressure in the evolution of communicative complexity is the stresses arising from living in a large social group with many weak ties to manage (Seyfarth & Cheney, Reference Seyfarth and Cheney2018; Roberts & Roberts, Reference Roberts and Roberts2020). Intentional signals involving audience checking, response waiting, attention getting, turn-taking, communicative innovation and culture help overcome the stimulus-driven processing that can occur due to stress, enhancing attention allocation and enabling greater predictability of conspecifics’ behaviour (Seyfarth & Cheney, Reference Seyfarth and Cheney2018; Roberts & Roberts, Reference Roberts and Roberts2020).