Variation in communicative complexity in relation to social structure and organization in non-human primates

Abstract

Communicative complexity relates to social complexity, as individuals in more complex social systems either use more signals or more complex signals than individuals living in less complex ones. Taking the individual group member's perspective, here we examine communicative complexity in relation to social complexity, which arises from two components of social systems: social structure and social organization. We review the concepts of social relationships and social complexity and evaluate their implications for communicative and cognitive complexity using examples from primate species. We focus on spider monkeys (Ateles geoffroyi), as their social organization is characterized by flexibility in grouping dynamics and they use a variety of communicative signals. We conclude that no simple relationship exists among social complexity, communicative complexity and cognitive complexity, with social complexity not necessarily implying cognitive complexity, and communicative and cognitive complexity being independently linked to social complexity. To better understand the commonly implied link between social complexity and cognitive complexity it is crucial to recognize the complementary role of communicative complexity. A more elaborated communicative toolkit provides the needed flexibility to deal with dynamic and multifaceted social relationships and high variation in fission–fusion dynamics.

This article is part of the theme issue ‘Cognition, communication and social bonds in primates’.

1. Introduction

The social complexity hypothesis for communicative complexity proposes that more complex social systems need more complex systems of communication [1]. In other words, variation in communicative complexity is linked to variation in social complexity, as individuals in more complex social systems either use more signals or more complex signals than individuals living in less complex ones. Although the association among social and communicative variables is persuasive, the hypothesis remains challenged by problems with operational definitions, conflicting results and weaknesses in testing paradigms [2]. In particular, it is unclear how primates use their communicative systems to manage the challenges arising from their social systems [3].

Our aim is to tackle the issue of communicative complexity from the perspective of each individual group member in relation to two components of social systems [4,5]: social structure, which refers to the patterns, content and quality of social relationships among group members, and social organization, which refers to group composition, size and cohesion. We first review concepts such as social relationships and social complexity, and then address their implications for communication and cognition. We use examples from various primate species, especially Geoffroy's spider monkeys (Ateles geoffroyi), a species with a high degree of flexibility in their social organization and their use of communicative signals.

2. Social relationships

Hinde [6,7] proposed a framework that links social interactions with social relationships and social structure. Social interactions are the observable components, characterized by the type of behaviour exchanged (e.g. grooming) and the manner in which it is exchanged (e.g. rapidly). In addition, any social interaction affects later interactions, and the interaction history constitutes the social relationship between two individuals. Accordingly, social relationships emerge from the interactions between individuals across time [8].

A social relationship is, therefore, captured by the types and manners of the interaction exchange between two individuals and their relative frequencies and patterning [6,7]. For example, two individuals that often groom and rarely fight have a different relationship from two individuals that rarely groom but often fight. How interactions are exchanged across time is also important. A relationship between two individuals that groom each other frequently one week but fought each other frequently the previous week has a different pattern from that between two individuals exchanging the same frequencies of grooming and attacks, but with each grooming bout being followed by a fight. In turn, the type of social relationship influences the social interactions two individuals may exchange. The social relationship constrains how two individuals interact as their previous interactions affect their subsequent interactions [9]. For example, dominance relationships affect how individuals interact in a competitive context.

The whole set of social relationships among group members constitutes a network of relationships, with each relationship being potentially influenced by any other relationship within such a network [6,9]. The network is the social structure of the group, which is described according to the characteristics of the constituent relationships. The social structure in turn affects the constituent relationships. For example, a group with a steep dominance hierarchy would have less-tolerant relationships among group members than a group with a more egalitarian hierarchy.

Viewing social relationships as investments for the individuals involved [10] led to the study of the mechanisms of relationship repair (e.g. post-conflict reconciliation: [11,12]) and their fitness benefits [13,14]. This perspective also focused attention on ‘strong’ social relationships, i.e. social bonds. However, social bonds are only one particular type of social relationship. As Hinde's framework underscores, individuals adjust their social interactions according to the identity of the partner, and social relationships are not homogeneous but can be highly differentiated across dyads of group members [6,7].

Variation across social relationships is well captured by the concept of relationship quality, which likely comprises several different components. Initially, three independent components were proposed [15]: (i) value, i.e. the benefits provided by a relationship; (ii) compatibility, i.e. the tolerance between two individuals; and (iii) security, i.e. the predictability of interactions between two individuals. Additional components can be added to thoroughly illustrate relationship quality in different species [16].

3. Social complexity

Complexity emerges from the interplay of various elements that interact in different ways, often following simple rules [17,18]. The same applies to social complexity [1,5]. However, we focus on social complexity experienced by group members in relation to social structure and social organization, rather than social complexity that arises from their interactions [19].

As an aspect of social structure, the extent to which individuals experience differentiated social relationships contributes to social complexity, with the more differentiated the social relationships the more likely individuals experience higher social complexity [20]. The extent of complexity individuals encounter depends on how individuals consider each relevant characteristic influencing the differentiation [19]. For example, when individuals differentiate how they interact with others based on kinship, complexity is lower if group members are simply distinguished into two categories of kin and non-kin than if the differentiation takes into account kinship in multiple (ordered) categories or as a continuous variable. Complexity increases when individuals consider multiple characteristics of group members (e.g. familiarity, dominance, age). Similarly, assessing relationship quality becomes more complex when relationships include several components each having more than two categories. For example, Geoffroy's spider monkey takes into account not only compatibility with subgroup members during fission events, but also the degree of value and security with them [21].

The differentiation of social relationships and in turn social complexity may be affected by socio-ecological factors. Recently, a conceptual framework was proposed to illustrate how multiple socio-ecological factors may predict the extent to which social relationships are differentiated [22]. The framework includes four components capturing: (i) how within-group contest competition influences access resources; (ii) how individuals vary in providing services; (iii) how group-level cooperation is needed; and (iv) how social interactions are constrained. Although some of the components are similar to previous socio-ecological models [23,24], the goals are different. Socio-ecological models focus on how social and ecological factors lead to social relationships typical of each species (e.g. resident-nepotistic: [23]). By contrast, Moscovice et al.'s [22] framework examines the roles that social and ecological factors play in shaping the variation of social relationships within each group. The framework simultaneously examines the contribution of multiple socio-ecological factors to each component influencing the degree of relationship differentiation. Thus, the framework leads to testable predictions about within- and between-species variation in social complexity [22].

Another important source of social complexity linked to social structure derives from variation within social relationships, depending on possible changes over time [19]. Given the dynamic nature of their relationships [9], animals need to be ready to respond to a possible change in how a partner behaves. When the change is highly predictable, it does not result in a major challenge. However, unpredictable changes make the social environment more uncertain and therefore more challenging. For example, when dominance hierarchies are relatively unstable [25], group members are likely to perceive the social environments as more complex than when dominance hierarchies are stable [26].

Social complexity may also derive from elements of social organization. Most researchers have focused on group size (reviewed in [5]), whereas we examine the degree of fission–fusion dynamics, which refers to spatial and temporal cohesion in the formation of subgroups [27] and is a component of all social systems [28]. A low degree of fission–fusion dynamics is typically present in relatively cohesive groups, whereas in groups with a high degree of fission–fusion dynamics individuals are rarely, if ever, all together because they split and come together in subgroups that frequently change size and composition [28]. These dynamics impact the extent to which group members interact with each other, with a high degree of fission–fusion dynamics creating uncertainty about social relationships and social contexts in which group members interact [29]. When fission–fusion dynamics are high, individuals experience more variability in subgroup composition and spatial cohesion and, consequently, more variability in potential social partners, supporters and eavesdroppers. The degree of fission–fusion dynamics adds to the social complexity that individuals experience within the same relationship and across relationships with different group members because it brings another shorter-term source of variation [19]. Within a relationship, the variation depends on the need to adjust the interactions in relation to how long two individuals have been separated and who else is in the subgroup. Across relationships, relative relationship quality changes for individuals and their social partners depending on who is in the subgroup [28].

4. The communicative toolkit

According to the social complexity hypothesis for communicative complexity, more signals and/or more complex signals are required in more complex than in simpler social systems [1–3]. Thus, the number of communicative signals is expected to depend on the degree of relationship differentiation, with a more elaborated communicative toolkit being present when relationship differentiation is higher [30].

Dominance/submission communicative signals deal with within-group contest competition, whereas affiliative signals are linked to within-group cohesion and cooperation. In most species both types of signals are present, but their relative number and use depend on the type of social relationship and social system [31]. For example, although the matrilineal system is at the basis of the social structure of all macaque species, there is between-species variation in the dominance style [32]. Rhesus macaques (Macaca mulatta) have a highly despotic dominance style, which constrains relationship differentiation, with variation occurring along the single axis of dominance/subordination. Under these circumstances, there is no need for an elaborated system of affiliative signals because a few unambiguous dominance signals are sufficient for social intercourse and coordination of group activities [33], suggesting a main role for only one of four components of Moscovice et al.'s [22] framework (i.e. how within-group contest competition influences access resources). Unlike rhesus macaques, pigtail (M. nemestrina) and stumptail macaques (M. arctoides), which exhibit less despotic dominance styles, use a variety of affiliative signals to maintain within-group cooperation and high levels of tolerance [34], suggesting the involvement of multiple components of Moscovice et al.'s [22] framework. Dominance style also affects communicative signals during post-conflict reconciliation across macaque species, in which more despotic species display lower conciliatory tendencies [35,36]. A comparison of nine macaque species revealed that more explicit signals (e.g. clasping) are used in groups where conciliatory tendencies are higher [37].

Variation in communicative complexity may also be linked to the degree of flexibility in the use of signals, with higher complexity when the same signal is used in multiple contexts or different signals are used in the same context [2,30,38]. For example, rhesus macaques use the facial display ‘lip-smack’ almost exclusively while approaching a group member and following an affiliative interaction, whereas pigtail and stumptail macaques use ‘lip-smack’ in a variety of contexts [34]. Chimpanzees (Pan troglodytes) kiss and embrace each other more frequently after conflicts than in other contexts, whereas they groom each other less often after conflicts than in other contexts [39,40]. Different signals may also be used in the same context depending on various aspects, including the timing, the proximity and the identity of the individuals involved. For example, stumptail macaques use two types of signals in post-conflict contexts [41]. Former opponents exchange brief touches in the immediate aftermath of a conflict when they are in close proximity, possibly preventing aggression from reoccurring. Grooming and other long-lasting behaviour occur later after the conflict, likely restoring valuable relationships between former opponents [41]. Another example of using different types of signals in post-conflict contexts depending on the circumstances is that of mandrills (Mandrillus sphinx), which use affiliative signals involving contact, such as grooming and muzzle contact, to reconcile with kin and subordinates and non-contact affiliative signals, such as ‘bared-teeth’ and ‘crest raise’, with individuals likely to renew aggression [42].

Characteristics of social organization may also influence variation in communicative complexity [28]. For example, fission–fusion dynamics may add layers of uncertainty about social relationships and the social contexts in which group members interact because of variation in spatial cohesion, subgroup size and composition [29]. Thus, fission–fusion dynamics impact the type of signals and the overall number of signals [43–47].

Aureli et al. [28] proposed a communicative framework in which the types of signals are related to relationship differentiation and fission–fusion dynamics (see the section ‘Implications of Fission–Fusion Dynamics for Communication’). For example, in groups characterized by a low degree of fission–fusion dynamics, individuals use signals mostly to deal with conflicts about resource competition and group coordination (see above). In groups characterized by a high degree of fission–fusion dynamics, individuals use signals to re-establish social intercourse and resolve uncertainties because of frequent spatial separations [29,30,48]. Following Aureli et al.'s [28] communicative framework, below we use our 20-year experience studying Geoffroy's spider monkeys to provide examples of the communicative toolkit in a species characterized by both a high degree of relationship differentiation and fission–fusion dynamics.

Spider monkeys exchange a variety of signals, including facial expressions, short-lasting contacts, long-lasting contacts and vocalizations [49–51]; for a detailed treatment of vocalizations see [52]. In many primate species, grooming is an affiliative signal [53], reflecting the value and compatibility of social relationships [15]. This pattern is present in spider monkeys, with grooming and proximity loading high on a component that maps onto compatibility [21,54]. Spider monkeys' communicative toolkit also includes short-lasting contacts such as embracing (i.e. the individual wraps one or both arms around another individual, while the two individuals are facing each other), pectoral sniffing (i.e. an individual places its face toward or touches its nose against the pectoral gland or armpit of another) and kissing (i.e. the individual puts its cheek and/or mouth close to another individual's face, without embracing), which are often collapsed into the behavioural category of ‘embrace’, given their relatively low frequency and similar function [55]. Interestingly, embrace loads on a different component derived from principal component analysis than grooming [54]. It loads high on a component with aggression [54], and the distributions of grooming and embrace across dyads are not correlated [55], suggesting that embrace serves a different function from affiliation.

Embrace may deal with relationship security, such as resolving uncertainties related to the re-establishment of social intercourse after separation (cf. [30]), which is supported by patterns of grooming and embraces following reunions. Aggression is much higher than at baseline when members from different subgroups join together after a period of separation [56]. Furthermore, those individuals exchange more embraces when reuniting in the aftermath of a fusion compared to baseline [50,56]. A similar result was obtained also after fusion-like events in a group housed at Chester Zoo [55]. Importantly for understanding the function of embrace, post-fusion aggression is dramatically reduced when embrace takes place [56].

The function of embrace in reducing uncertainty and risk is also supported by two other findings. The first supporting finding is about female–female interactions. In several primate species, females are attracted to the infants of others [57] and groom mothers to gain access to their infants [58,59]. Female spider monkeys, however, give embraces rather than grooming to access infants [60]. The second supporting finding is about male–male interactions. Grooming is reciprocated among males regardless of age; however, embraces are not reciprocated between males of different age, with younger males initiating most embraces with older males [61]. Given that aggression between males can be lethal, albeit rare [62,63], the likelihood of aggression from older to younger males may be reduced by the exchange of embraces. In addition, patterns of grooming and embraces between males are influenced differently by aggression patterns. Dyads characterized by higher aggression rates exchange less grooming and more embraces than dyads characterized by lower aggression rates [64]. Furthermore, embrace patterns change over time in concert with perceived risk as embrace increased in dyads characterized by lower aggression rates when males from the dyads characterized by higher aggression rates were no longer in the group [64].

Spider monkeys' communicative toolkit includes a longer-lasting, rarer and more complex form of interaction than embrace, which we currently label 'grappling' based on Eisenberg & Kuehn's [65] terminology and definition of a long-lasting sequence of behaviours, including some or all of the following: face greeting, face touching, embraces, pectoral sniffing, tail wrapping, genital contact and tee–tee vocalizations. Furthermore, during grappling it is common for individuals to come together and move apart repeatedly like a ‘push–pull dance’. Grappling may last anywhere from a few to over 30 min and occurs in contexts similar to those in which embrace occurs. We observed grappling events in male–male, female–female and female–male dyads. A typical pattern is that subadult males tend to initiate and maintain grappling with adult males [61], i.e. with individuals with whom subadult males have relationships that are deemed uncertain given the risk of lethal aggression [63]. During a period characterized by aggression among adult males, we observed several grappling events between adult males, including three cases with anal penetration [66].

Hence, in spider monkeys, embrace and grappling likely serve a communicative function other than affiliation, which is typically served by grooming. As spider monkeys deal with potential competition over resources, conflicts over the activity to pursue and travelling direction by fissioning into subgroups [67–69], they do not have dominance/submission signals in their communicative toolkit, which typically serve to reduce uncertainty and risk in competitive situations [70]. In spider monkeys, uncertainty and risk occur in different situations (see above), in which embrace and grappling are used to facilitate social intercourse by reducing uncertainty and the risk of conflict [8].

Like spider monkeys' embraces, ritualized greetings between male baboons (genus Papio) depend on aspects of social structure and social organization that affect social complexity. They involve a combination of highly stylized repetitive elements, from quick walking-by, head bobbing and quick touches to potentially harmful behaviours such as genital fondling [71,72]. There is a wide variation in the frequency and the degree of elaboration of male–male greetings across baboon species, being rare and rather simple in species experiencing high male–male contest competition and frequent and with risky components, such as genital fondling, in species characterized by low male–male contest competition and some degree of cooperation [73]. Ritualized greetings are most elaborate in Guinea baboons (Papio papio) that live in a multilevel social organization in which basic social units fission and fuse, forming larger units labelled parties and gangs [74]. Ritualized greetings occur almost exclusively between males from the same party and likely serve to facilitate social intercourse within their multilevel social organization [73,74]. Thus, variation in ritualized greetings across baboon species seems to reflect the degree of relationship differentiation in terms of competition, tolerance and cooperation between males and the degree of fission–fusion dynamics.

5. Cognitive implications

In the previous section, we showed that one way in which communicative complexity is related to social complexity is that a more elaborated communicative toolkit is expected when there is a higher degree of relationship differentiation. In this section, we evaluate whether this aspect of social complexity has implications in terms of cognitive complexity. The degree of social complexity individuals face depends on how individuals account for factors that influence relationship differentiation, with higher complexity expected when individuals need to take into account multiple characteristics of other group members and multiple components of relationship quality, especially when based on continuous or multiple-set ordinal assessment (see §3 above). In addition, individuals need to take into account the changes in each of their relationships over time. To do so, individuals must track their past interactions (cf. [6]). In essence, they need to integrate information about the pattern of previous social interactions with each of their partners, convert information from different types of interactions into common currency, and continuously update this integrated information [8,70]. This is a bookkeeping process that seems cognitively burdensome.

We have proposed that such a bookkeeping process can be achieved without a major burden on cognitive abilities because individuals can rely on their emotions to inform their decisions [75,76]. Emotions act as an intervening variable [77,78] in which the emotion experienced constrains how the individual responds to a given situation [78]. As an individual's emotional state is influenced by the patterning of previous interactions with each group member, partner-specific emotional states can be developed [75,76,78,79]. The neuroendocrine mechanisms underlying partner-specific emotional states remain uncertain; however, oxytocin and endorphins, which are released in association with the exchange of grooming, are potential candidates ([80,81]; see also [82] for a review). Such partner-specific emotional states can provide a crucial integration of the various interactions between two partners at any given moment, varying as interaction patterns change. Thus, partner-dependent emotional states function like bookkeeping, which is crucial for assessing variation in social relationships [8] and may be essential in navigating social complexity that arises from the degree of relationship differentiation [19].

A high degree of fission–fusion dynamics provides another facet of social complexity that may affect the communicative toolkit (see §4 above). Here, we evaluate whether such an aspect of social complexity has cognitive implications. The degree of fission–fusion dynamics hinges on variation in spatial cohesion, subgroup composition and size, which in turn influences how often group members can interact with one another [28]. When groups are characterized by a high degree of fission–fusion dynamics, group members potentially face more social complexity because of an additional, shorter-term source of variation in their relationships with different partners [19], which creates further uncertainty [29].

A high degree of fission–fusion dynamics may imply that individuals face difficulties in behaving consistently toward group members because of their prolonged absence due to being in different subgroups. After a subgroup fusion, individuals need to resume social intercourse. In addition to communicative signals specific to this context (e.g. post-fusion embraces in spider monkeys [56]), resumption of social intercourse after separation may require enhanced memory abilities to enable individuals to remember past interactions for longer so that they can retain their influence on the attitude toward a partner, which may imply enhanced episodic or semantic memory [83,84] and longer-lasting emotional consequences (cf. [28]). Furthermore, prolonged separation potentially poses challenges for cooperative interactions [85,86], with which animals may cope by shortening the timeframe of social interchange to reduce errors [87].

Self-control is another cognitive ability to be enhanced when the degree of fission–fusion dynamics is high [88]. Following a subgroup fusion, individuals should demonstrate restraint in interacting as usual with other group members because the social milieu has changed, and then assess the new situation and adjust their behaviour accordingly [28]. Experimental evidence supports this view as more inhibition and more flexible responses are shown by individuals of species with a higher degree of fission–fusion dynamics [89–91].

An enhancement of cognitive abilities can also be expected to deal with the influence of other individuals (i.e. third parties) on the social interactions between two partners [92–95]. When the degree of fission–fusion dynamics is high, an individual may make decisions about whether to fission depending on subgroup membership. Here are two examples of hypothetical strategies for such decisions. First, an individual might join a subgroup to attack a particular individual, but would avoid joining if the target's ally is also present. Whereas we do not have direct evidence for such decisions, there is evidence for individuals fissioning with group members with whom they have high-quality relationships [21]. Second, we can expect individuals to engineer subgroup membership to their benefit [28]. For example, they could choose subgroups with the strongest group members in order to be protected by them. Alternatively, they can select subgroups with the most vulnerable individuals to serve as their protector and as a consequence receive in exchange more grooming or mating opportunities (cf. [48,96]). Such strategies may require enhanced planning-related cognitive abilities [28]. So far, no test of such an enhancement has been done using comparisons across species with different degrees of fission–fusion dynamics. Similarly, no between-species comparison has been done examining the cognitive abilities potentially required to cope with highly variable social environments, although metrics to quantify the uncertainty level due to such social environments have been developed [29].

6. Conclusion

We reviewed evidence for communicative complexity in relation to aspects of social structure and social organization regarding social complexity and examined its potential cognitive implications. Based on our review, we can conclude that there is no simple relation between social complexity, communicative complexity and cognitive complexity.

We showed that social complexity related to relationship differentiation has implications for communicative complexity, with a more elaborated communicative toolkit being expected when the degree of relationship differentiation is higher. Nevertheless, such social complexity does not necessarily imply cognitive complexity. Taking into account all factors affecting relationship differentiation, such as the multiple characteristics of other group members and the multiple components of relationship quality, and their changes over time, certainly appears demanding. We argued, however, that the bookkeeping of all those factors can be achieved by emotional mediation, without an enhancement of specific cognitive abilities.

We indicated that social complexity related to fission–fusion dynamics, which influences how often group members can interact with one another, has likely cognitive implications, with an enhancement of memory-related abilities, self-control and potentially planning when the degree of fission–fusion dynamics is higher. Under this scenario, cognitive complexity would increase with higher social complexity along with an increase in communicative complexity in terms of additional signals to enable individuals to resolve uncertainties and re-establish social intercourse after periods of separation.

Social complexity involves aspects that we did not cover in our review, such as the assessment of the social relationships between other group members due to the potential influence of third parties on the social interactions between two individuals. Such an assessment has important implications for cognitive complexity (reviewed in [79]). However, we do not envision that communicative complexity should be influenced by such an assessment.

Summarizing, two aspects of social complexity, relationship differentiation and fission–fusion dynamics, seem related to communicative complexity, but only the latter might result in increased cognitive complexity. We thus propose an approach to better understand the link between social complexity and cognitive complexity that is based on recognizing different aspects of social complexity and an additional, crucial role of communicative complexity. A more elaborated communicative toolkit provides the needed flexibility and effectiveness to deal with dynamic and multifaceted social relationships and high variation in fission–fusion dynamics.

Data accessibility

This paper has no additional data.

Authors' contributions

F.A.: conceptualization, writing—original draft; C.M.S.: conceptualization, writing—review and editing; G.S.: conceptualization, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

Writing of this paper was facilitated by a Short-term Mobility grant from CNR to G.S. Our long-term research on spider monkeys has been funded by British Academy, Chester Zoo, CONACYT, Leakey Foundation, National Geographic Society and Wenner-Gren Foundation.