Social selectivity in aging wild chimpanzees

Abstract

Humans prioritize close, positive relationships during aging, and socioemotional selectivity theory proposes that this shift causally depends on capacities for thinking about personal future time horizons. To examine this theory, we test for key elements of human social aging in longitudinal data on wild chimpanzees. Aging male chimpanzees have more mutual friendships characterized by high, equitable investment, whereas younger males have more one-sided relationships. Older males are more likely to be alone, but also socialize more with important social partners. Finally, males show a relative shift from more agonistic interactions to more positive, affiliative interactions over the lifespan. Our findings indicate that social selectivity can emerge in the absence of complex future-oriented cognition, and provide an evolutionary context for patterns of social aging in humans.

Social bonds have adaptive consequences over the lifespan: strong social support enhances health, longevity, and biological fitness (1, 2). In humans, old age is characterized by increasing selectivity for positive, meaningful social interactions, manifesting as a cluster of behavioral and cognitive features we term the human social aging phenotype. First, older adults across societies have smaller yet more emotionally-fulfilling social networks than younger adults, due to an increasing focus on existing close relationships rather than new relationships (3–6). Second, older adults exhibit a positivity bias, showing greater attention to and memory for positive versus negative socioemotional information, and reduced engagement in tension and conflicts (7–9). The origin of this social aging pattern is therefore a central issue both for evolutionary perspectives on the life-course, and for promoting wellbeing in old age.

Socioemotional selectivity theory has emerged as the most influential explanation for the human social aging phenotype, arguing that the central process generating lifespan shifts in sociality is an explicit sense of future personal time and mortality (10, 11). The core idea is that when individuals perceive the future as expansive (as in youth) they prioritize building new relationships and interacting with many partners, whereas when time is perceived as short (as in old age) people focus on existing, important social ties. In support of this view, older adults perceive a more limited future than younger adults; people who anticipate curtailed time horizons—due to an illness diagnosis, natural disaster, or a geographic move—generally exhibit preferences like older adults; and experimental manipulation of future time perspective shifts socioemotional biases (9, 11–13). However, some evidence indicates that changes in socioemotional goals can be independent of future time perspective (14, 15). Thus, the role of shortened time perspectives in social selectivity during aging is currently unclear.

Here, we use a comparative approach to provide a new test of the origins of human social aging patterns. Socioemotional selectivity theory proposes that changes in social goals and behavior during aging are causally dependent on an awareness of shortened personal time horizons. However, there is no evidence that any other species are aware of their own future mortality or can imagine far-off future experiences in this rich way. Some nonhumans do engage in forms of future-oriented planning, but only in short-term food acquisition contexts such as saving a tool to access food hours or days later, and some of these instances may actually recruit lower-level mechanisms (16, 17). Even verbal young children show limitations in future-oriented cognition, and can struggle to imagine their future selves (18). Accordingly, if this kind of subjective future time perspective is causally necessary to generate the human social aging phenotype, then other animals should not show these characteristic shifts.

An alternative possibility is that the human social aging phenotype is mediated by proximate mechanisms that are more widely shared across species. Cost-benefit tradeoffs about whether to be social and with whom to socialize are critical for many animals. As aging imposes new constraints due to declines in physical condition, immunological health, and social status (19–22), older individuals might need to adjust their social choices. Accordingly, socioemotional selectivity in humans could represent an adaptive response where older adults focus on important social relationships that provide benefits, and avoid interactions that may have negative consequences. If so, other animals might also show social selectivity without necessarily possessing sophisticated future-oriented cognition like humans.

We test these alternatives by examining if the key characteristics of the human social aging phenotype are shared with wild chimpanzees (Pan troglodytes). Chimpanzees are an ideal comparand because they are one of humans’ two closest living relatives, have long lifespans of 50–60 years in the wild, and form flexible, long-term social bonds. They have a high degree of choice about who to interact with due to their large fission-fusion societies which comprise temporary and fluid sub-groups (‘parties’) that can range from one chimpanzee to almost the entire community. Our data come from Kibale National Park, Uganda, where we have documented social interactions in a community of wild chimpanzees on a near-daily basis for over 20 years, providing longitudinal observations that are unparalleled in human research.

We analyzed social interactions from 78,000 hours of observations from 1995–2016, comprising 21 male chimpanzees ranging from 15 years (when males are physically mature and enter the adult hierarchy) to 58 years, with an average of 10.6 years of data per individual and 141.6 observation days per year. We examined males because they exhibit stronger bonds and more frequent social interactions than relatively asocial females (23, 24). We used mixed models to test the importance of age in the longitudinal data. We always controlled for an individual’s dominance rank, which has a pervasive effect on chimpanzee social interactions and declines in old age (19). Few males had adult maternal brothers so we did not account for kinship, but we controlled for other predictors such as year or female presence when appropriate for the social metric (see supplementary materials).

We first examined whether older chimpanzees focused their social interactions on important partners, a key signature of the human social aging phenotype. To characterize relationships, we used a spatial proximity metric indexing the time that pairs of individuals spend near each other. Close proximity is an important marker of affiliation in primates, as it is a pre-requisite for other cooperative interactions like grooming and signals social comfort (1). We indexed social preferences by examining how often two individuals were within 5m when in the same party. We then categorized male-male dyads as ‘mutual friends’ (both showed a preference to sit near the other, above their individual average rates of association), ‘one-sided friends’ (one individual showed this preference, but their partner did not), or ‘non-friends’ (neither preferred to associate with the other). Thus, mutual friendships are reciprocated, whereas one-sided friendships are not. We found that the number of mutual friends increased with age [χ²=6.89, df=1, p<0.01; Fig 1a], whereas one-sided friendships declined [χ²=9.76, df=1, p<0.005; Fig 1b]. For example, 15-year-olds had an average of 2.1 one-sided friends and 0.9 mutual friends, whereas 40-year-olds had 0.6 one-sided friends and 3.0 mutual friends. Finally, age and dominance had independent effects on relationships (see supplementary materials): both higher rank and older age predicted more mutual friendships, but fewer one-sided friendships.

Figure 1: Friendships in aging chimpanzees.

(a–b) Older males had more mutual friendships, whereas younger males had more one-sided friendships. (c–d) Chimpanzees invested more and had more equitable grooming patterns with mutual friends. (e–f) Older males (35+ years) were mutual friends with peers, and attractive partners for younger males’ (15–20 years) and prime-aged males’ (20–35 years) one-sided friendships. Ribbons and error bars indicate 95% CI estimates.

To test whether mutual friendships were high-value bonds like those prioritized by older humans, we then examined grooming, a principal form of primate social investment (1). We first assessed whether mutual friends were more likely to groom, and found that chimpanzees of all ages engaged in more total grooming with mutual friends compared to one-sided and non-friends [χ²=94.38, df=2, p<0.0001; p<0.0001 for pairwise comparisons; Fig. 1c], with similar results for grooming given and received (see supplementary materials). Second, mutual friends engaged in longer grooming bouts [χ²=25.03, df=2, p<0.0001; p<0.05]. Third, mutual friends had more equitable patterns of grooming, both within bouts [χ²=38.23, df=2, p<0.0001; Fig. 1d] and across the year [χ²=58.13, df=2, p<0.0001]. Using this dyadic data, we further found that mutual friendships were more common amongst dyads of older males than dyads of prime-aged or younger males, but one-sided friendships were driven by younger males seeking out prime-aged and older partners (see Fig. 1e–f and supplementary materials). Thus, chimpanzees invested more, and more equitably, in mutual than one-sided relationships, and older adults had more mutual friendships than younger adults.

Older human adults have smaller but more selective social networks, so we next examined how chimpanzee gregariousness changed with age. We assessed the likelihood that an individual was observed alone, and found that although males were rarely alone (less than 1% of observation time), solitariness did increase with age [χ²=4.51, df=1, p<0.05]. Yet when chimpanzees did socialize, age predicted an increased likelihood that they were in a party included at least one other adult male, as opposed to only females or juveniles [χ²=15.73, df=1, p<0.0001; Fig. 2a]. Moreover, older males were observed in larger male parties [χ²=19.50, df=1, p<0.0001] and were more likely to be in close proximity of another adult male [χ²=18.02, df=1, p<0.0001; Fig. 2b]. For example, 15-year-olds were found in parties averaging 5.8 other males, and were physically near another male on 36.6% of observations. In contrast, 40-year-olds were in parties with 6.6 other males, and in proximity to another male on 53.7% of observations. Thus, older males showed some declines in their overall tendency to be gregarious, but were more likely to be in parties with and sit near important social partners when they did choose to socialize.

Figure 2: Gregariousness and positivity bias in aging chimpanzees.

Older males were more likely to be (a) in parties with other males, and (b) sitting in proximity to those males. (c) While both directed aggression (with specific targets) and non-directed aggression (displays) declined with age, grooming remained fairly constant. Ribbons indicate 95% CI estimates.

Our final set of analyses tested whether chimpanzees exhibit an increasing positivity bias, the other key component of the human social aging phenotype. We examined the overall rates that individual male chimpanzees spent giving grooming to all others as an index of positive social interactions, versus rates they spent giving aggression as an index of negative interactions (comprising both directed aggression where the individual targets another specific individual by hitting, biting, or chasing them; and non-directed aggressive displays without particular targets). We found that whereas grooming remained fairly constant across the lifespan, aggression decreased with age [χ²=69.09, df=1, p<0.0001; Fig. 2c]. Grooming and aggression received from others showed similar patterns (see supplemental materials). Chimpanzees therefore show a behavioral shift from relatively more negative interactions to more positive interactions during aging, analogous to the human positivity bias.

Overall, our data provide the first evidence for social selectivity during aging in nonhumans. We found that older chimpanzees, like humans, prioritized high-quality relationships: whereas younger adults had more asymmetrical friendships characterized by reduced investment, older adults exhibited more mutual friendships characterized by a high, equitable investment. Second, older chimpanzees were more likely to be observed alone, but tended to socialize more with important partners by joining parties with other males, joining larger male parties, and sitting in close proximity to other males. Finally, chimpanzees exhibited an increasing positivity bias, showing consistent grooming but reductions in aggression across the lifespan. Importantly, these patterns were independent of dominance rank: despite their falling status, older males were more likely to be mutual friends with each other, and were the targets of ‘one-sided’ advances from younger males. Together, our data indicate that chimpanzees demonstrate key behavioral signatures of the human social aging phenotype, showing that increasing social selectivity can occur in the absence of a rich future time perspective.

These data support the view that senescence drives fundamental shifts in the costs and benefits of social interactions across species. Yet although many species face new constraints during aging, chimpanzees and humans show a response to these constraints that is not universal. In particular, several other primate species exhibit social withdrawal during aging—reducing social interactions overall without focusing on important partners—as well as a negativity bias characterized by declines in affiliation but steady rates of aggression (25–28). Why do aging chimpanzees and humans instead show social selectivity and a positivity bias? We propose that optimal social responses to aging depend on a species’ social organization and life history. The relationships of many species are primarily based on kinship, and senescence makes it difficult to form new relationships as close relatives die. Consequently, social withdrawal may be a common pattern. Yet social relationships are flexible, can occur outside of kinship, and last many years in long-lived humans and chimpanzees (23, 24). Thus, strongly-established relationships may be more reliable for older chimpanzees than for other primates. This may be especially important for species with relatively low reproductive skew like chimpanzees, as older individuals can still obtain fitness benefits via cooperative alliances despite changes in health and social status (19–22).

A second question concerns the specific proximate mechanisms underpinning social selectivity in chimpanzees. Socioemotional selectivity theory proposes that the human social aging phenotype causally depends on an explicit sense of the self in time, but given that other animals have constrained future-oriented cognition (16, 17), other proximate mechanisms must play a role in nonhumans. In some primates, age-related shifts in social behavior have been proposed to stem from declining capacities to cope with stress (28), yet older chimpanzees do not exhibit major increases in sensitivity to either energetic or social stressors despite higher overall glucocorticoids (22). An alternative possibility is that chimpanzee social aging patterns are driven by shifts in emotional reactivity, given that increasing capacities for emotional regulation is a feature of human aging (8, 14). Thus, a key question is whether older chimpanzees also exhibit shifts in affective processes, such as a less reactive temperament or lower rates of reactive aggression.

Our findings demonstrate how data from long-lived, socially-flexible animals are crucial to disentangle the proximate and ultimate causes of human social aging patterns. In addition to testing how different species respond to the constraints of aging, this can inform our understanding of evolution of social roles. In other long-lived mammals like cetaceans and elephants, older individuals serve as stores of ecological knowledge that benefit groupmates (29). This indicates that a prolonged life-course can shape adaptive strategies for information-processing, and suggests that the social aging phenotype characteristic of chimpanzees and humans may be due in part to the social knowledge that long-lived, socially-flexible species can acquire. Indeed, older adult humans exhibit greater crystallized intelligence and skillfulness at reasoning about social conflicts (30). Similarly, older chimpanzees might acquire social knowledge over their lifespan that makes them attractive social partners despite falling dominance status. In sum, while humans exhibit extraordinary cognitive features that allow for complex reasoning about the future, commonalities in social aging between humans and other animals are key to understanding the evolution and function of these mechanisms.