The claim that humans are social animals is repeated so often and in so many contexts that it has acquired the status of a biological fact rather than what it actually is: a partial description of a complex situation, deployed selectively to justify a particular model of social life.
Humans are social animals. They are also, under certain conditions, solitary animals. They are animals who form intense pair bonds. They are animals who cooperate in small groups. They are animals who compete within those groups. They are animals who need periods of withdrawal from social contact to process experience, regulate emotion, and restore the cognitive resources that sustained social engagement depletes. The full description is considerably more complicated than the slogan, and the slogan tends to be invoked at the moment when the complication would be inconvenient.
The inconvenient complication, in the context of the loneliness epidemic and its prescribed remedies, is this: the biological evidence for what humans actually need from social life points consistently toward depth rather than breadth, toward the quality of a small number of bonds rather than the quantity of social contacts, toward the specific neurobiological benefits of close, enduring, reciprocal relationships rather than toward general group membership. The evidence does not support the epidemic’s implicit model that more social contact is better. It supports a different model, one that the epidemic’s infrastructure—the community groups, the social prescriptions, the befriending services—is not well designed to produce.
Robin Dunbar’s work on primate social groups, which he extended to humans in the early 1990s, is the most frequently cited evidence for the claim that humans are adapted for group living. The Dunbar number—approximately 150 as the cognitive limit of stable social relationships a human can maintain—has entered popular culture as evidence that humans are naturally group animals. What is less frequently cited from Dunbar’s work is the layered structure he describes within the 150. The stable group of 150 is subdivided into layers of decreasing size and increasing intimacy: roughly 50 acquaintances, 15 people you can turn to for support, 5 people in the innermost circle of close friends and family.
The innermost circle is the biologically significant layer. The 5, not the 150, are the relationships that carry the neurobiological load of genuine social bonding. These are the relationships characterised by the sustained, reciprocal, physically present contact that activates the opioid and oxytocin systems associated with the stress-buffering effects of social connection. Dunbar’s research on the mechanics of this bonding—the role of synchronised activity, shared laughter, physical touch, the extended face-to-face conversation—shows that these effects are produced by specific kinds of interaction and not by the general fact of being in proximity to other people.
The distinction matters for the epidemic framing. What the opioid system requires for the stress-buffering effect is not contact in the general sense but the specific quality of contact available in the innermost circle. The community group attendance that the social prescription provides is an interaction at the level of the 50 or the 15. It does not produce the neurobiological effects of the 5. It is not biologically equivalent to the relationship it is being prescribed to supplement or replace.
The primate research is more direct about this than the human research tends to be, because primate social bonds can be studied longitudinally in ways that human relationships cannot easily be. Joan Silk’s long-term studies of baboon populations in the Amboseli basin in Kenya have tracked the relationship between social bond strength and survival outcomes over decades. The finding is consistent and striking: female baboons with strong, stable, enduring bonds with a small number of other females survive longer, reproduce more successfully, and show lower glucocorticoid levels—the hormonal indicator of chronic stress—than females without such bonds, regardless of their rank in the dominance hierarchy.
The rank finding is particularly important. Dominance rank confers material advantages—priority access to food, safety from predation, the social capital of high group status. These advantages predict survival. So does bond strength. But bond strength predicts survival independently of rank—a low-ranking female with strong bonds outlives a high-ranking female with weak ones. The bonds are not merely a proxy for the advantages that high rank confers. They are providing something that rank and group membership do not provide.
What they are providing is the specific stress-buffering effect of close, reciprocal, enduring relationship—the physiological regulation that occurs when a trusted other is physically present, that reduces the allostatic load of continuous social monitoring and competitive positioning, that allows the nervous system to rest in a way that the general social environment does not. The bonded baboon is not merely socially supported in the abstract sense. She is neurobiologically regulated in the specific sense that her stress hormone levels are measurably lower in the presence of her bonded partner than in the presence of the group.
The human neurobiological evidence points in the same direction. Naomi Eisenberger’s research at UCLA on the neural correlates of social pain demonstrates that the brain processes social exclusion through the same neural circuitry that processes physical pain—the anterior cingulate cortex and the anterior insula, the regions associated with the aversive quality of physical injury. This is why social exclusion hurts in a way that is not metaphorical. The neural response is not analogous to pain. It is pain, processed through the same alarm system.
What matters for the present argument is the specificity of what triggers the pain response. Eisenberger’s work, and the broader literature on the neuroscience of social bonding, shows that the pain response is triggered less by the general fact of low contact than by the specific experience of not being seen, not being known, not being received by the close relationships that carry the neurobiological significance of genuine bonding. A person in a crowded room who does not feel genuinely received by the people around them activates the same neural alarm as a person in physical pain. The crowd does not buffer the response. In some conditions it intensifies it, by making the absence of genuine reception more salient through contrast.
This is the neurobiological explanation for the paradox that Riesman identified sociologically—being lonely in a crowd—and that the epidemic framing struggles to address because its model of social need does not distinguish between the neurobiological requirement and the social contact that superficially resembles its satisfaction.
The high-arousal, group-based social environment that the contemporary other-directed culture produces is not neurobiologically neutral for the people within it. The continuous social monitoring, the performance management, the antagonistic cooperation, the ongoing calibration of presentation to the crowd’s response—these are activating rather than regulating. They maintain the nervous system in a state of low-level alert, the allostatic cost of continuous competitive positioning within a group whose approval is the primary source of orientation. The stress-buffering effects of genuine close relationship require the opposite condition: the trusted other in whose presence the monitoring can stop, the radar can rest, the performance can be suspended.
The evolutionary logic of this is straightforward. The hominid groups in which humans evolved were small—estimates based on archaeological evidence and comparisons with existing hunter-gatherer societies suggest band sizes of between 20 and 50 individuals, with the key cooperative unit being smaller still: the pair bond, the sibling relationship, the small cluster of close kin and trusted non-kin who constituted the immediate reliance group. These relationships were not selected for because group membership was good for you in the general sense. They were selected for because they provided specific survival advantages—coordinated defence, cooperative childcare, food sharing, the reciprocal grooming and physical contact that activated the opioid system and lowered the stress response.
The evolutionary pressure was toward depth rather than breadth. Breadth—the management of relationships across the wider band and beyond it—was also adaptive, but for different purposes: information exchange, resource access, the negotiation of inter-group relations. The quality of the innermost bonds was what determined health and survival. The width of the social network was what determined access to resources and information. These are different functions, served by different kinds of relationship, producing different kinds of biological effect.
The constructed epidemic model has collapsed this distinction. It has treated the absence of broad social contact as the deficit requiring correction, without examining whether the deficit that is actually producing the health consequences is the absence of the deep bonds that carry the neurobiological significance. The community group referral addresses the broad contact layer. The befriending service addresses, at best, the support layer. Neither is designed to produce the innermost bonds that the biological evidence identifies as the stress-buffering relationships that predict health and longevity.
The person who is genuinely lonely in the biologically meaningful sense—whose innermost circle is empty or insufficient—cannot have that deficit addressed by the addition of social contacts at the outer layers of the network. The contacts are not the wrong kind in a qualitative sense alone. They are the wrong kind in a neurobiological sense: they do not activate the bonding systems, they do not produce the opioid response, they do not lower the glucocorticoid levels, they do not provide the rest from monitoring that the trusted other provides.
The slogan that humans are social animals is true in the sense that the social environment matters enormously for human health and wellbeing. It is misleading in the sense that it implies the solution to the deficit of social life is more social life in the general sense.
What the evidence points toward is not more social life but deeper social life—the specific quality of bond that the evolutionary environment selected for, that the neurobiological systems are calibrated to require, and that the contemporary other-directed, high-arousal, performance-organised social environment is singularly ill-suited to produce.
The baboon with strong bonds lives longer than the baboon with high rank.
The epidemic is counting rank.
The bonds are elsewhere.
The Vorpel Foundation