key: cord-340101-n9zqc1gm
authors: Bzdok, Danilo; Dunbar, Robin I.M.
title: The Neurobiology of Social Distance
date: 2020-06-03
journal: Trends Cogn Sci
DOI: 10.1016/j.tics.2020.05.016
sha: 
doc_id: 340101
cord_uid: n9zqc1gm

Abstract Never before have we experienced social isolation on such a massive scale as we have in response to COVID-19. Yet we know that the social environment has a dramatic impact on our sense of life satisfaction and well-being. In times of distress, crisis, or disaster, human resilience depends on the richness and strength of social connections, as well as active engagement in groups and communities. Over recent years, evidence emerging from various disciplines has made it abundantly clear: loneliness may be the most potent threat to survival and longevity. Here, we highlight the benefits of social bonds, choreographies of bond creation and maintenance, as well as the neurocognitive basis of social isolation and its deep consequences for mental and physical health.

J o u r n a l P r e -p r o o f 3 are conventionally most concerned about all had much less impact on survival rates. Key factors included obesity, diet, alcohol consumption, how much exercise was taken, the drug treatments prescribed, and local air pollution. These authors conducted a follow-up analysis of 70 studies of longevity in older people, which followed ~3.5 million people over an average of ~7 years [16] : social isolation, living alone and feeling lonely increased the chances of dying by about 30%, even after accounting for age, sex and health status.

Many other studies have shown that social isolation (though not self-reported feelings of loneliness) was a significant predictor of the risk of death. For example, a longitudinal analysis of ~6,500 British men and women in their fifties [17] found that being socially isolated increases the risk that you will die in the next decade by about 25%. Quantitative analysis of nearly ~400,000 married couples in the American Medicare database revealed that, for men, the death of their spouse increased their own chances of dying in the immediate future by 18%. The death of the husband in turn increased the wife's risk of dying by 16% [18] .

Similar effects on morbidity rates have been found with respect to social support. A series of elegant prospective studies using data from the Framingham Heart Study [19, 20] found that the chances of becoming happy, depressed or obese were all strongly mirrored by similar changes in the closest friend. There was a smaller significant effect due to the behaviour of the friends' friend. Even a just detectable effect was present due to the friend of a friend's friend, but nothing beyond. This contagion phenomenon was especially strong if the friendship was reciprocal (i.e., both individuals listed each other as a friend). If the friendship was not mutual, the social contagion effect was negligible. The investigators also documented a strong effect of "geographical contagion". If you have a happy friend who lives within a mile radius, you are 25% more likely to become happy. And you are 34% more likely to be happy if your next-door neighbour is happy.

People who belong to more groups are less likely to experience bouts of depression. Such findings emerged from the UK Longitudinal Study of Ageing (ELSA) that repeatedly profiled around ~5,000 people from the age of 50 onwards. Previous research showed [21] that depressed people reduced their risk of depression at a later time point by almost a quarter if they joined a social group such as a sports club, church, political party, hobby group or charity. Indeed, joining three groups individuals were more immersed in their local community and trusted their neighbours more [22] [23] [24] . The causal directionality was difficult to pin down in these cases because of the cross-sectional nature of the data. Nevertheless, path analysis provided some indication that intensity of social exchange was the candidate driver.

The impetus to access social capital in the wider community [7] extends beyond humans.

There is now a wealth of evidence from long-term field studies of wild baboons that socially wellconnected females experience less harassment by other monkeys [7, 23] , have lower levels of cortisol stress hormones [25, 26] , faster wound healing [27] , produce more offspring and live longer [28] [29] [30] [31] . Such ramifications of social capital appear to hold up across a diversity of species, including chimpanzees [32] , macaques [33] [34] [35] , feral horses [36, 37] and dolphins [38] .

A key underlying reason for these effects, at least in humans, is likely that loneliness directly impairs the immune system, making you less resistant to diseases and infections. Research found [39] that freshmen students who reported feeling lonely had a reduced immune system response when they were given a flu vaccine compared to students who felt socially well engaged. Moreover, those students with only 4-12 close friends had significantly poorer responses than those with 13-20 friends. These two effects seemed to interact with each other: having many friends (a large social group of nineteen or twenty friends) seems to buffer against a weakened immune response. Yet, feeling lonely and having few friends results in a particularly poor immune defence. Other investigators [40] used data from the Framingham Heart Study to show that people with fewer contacts in their social network had elevated serum fibrinogen concentrations. In contrast, people enjoying many social contacts had low fibrinogen levels. Fibrinogen plays an important role in blood clotting when a blood vessel has been ruptured, as well as facilitating wound healing and tissue repair more generally: high concentrations thus signal poor health. Endorphins constitute a core component of the psychoendocrine mechanism underpinning friendship (see Box 1). Other research found [41] that social bonds stimulate the release of the body's natural killer cells, one of the white blood cells of the innate immune system whose core function is to destroy harmful bacteria and viruses.

People who are more socially integrated have better adjusted biomarkers for physiological function, as indexed by lower systolic blood pressure, lower body mass index, and lower levels of Creactive protein -the latter being another molecular response to inflammation. This insight was evident in each of four age groups (adolescents, young adults, middle age and old age) based on data from four large longitudinal American health databases [42] . The investigators found that, in J o u r n a l P r e -p r o o f adolescence, lack of social engagement had as big an effect on risk of inflammation as lack of physical activity. In old age, lack of friends had a bigger effect on risk of hypertension than the usually cited clinical causes like diabetes. Even more worrying, the effects of social relationships on these physiological measures of good health during adolescence and young adulthood can persist into old age. In a longitudinal study of 267 males, for example, research found [43] that the more socially integrated a child was at six years of age, the lower their blood pressure and body mass index (a measure of fatness) two decades later in their early thirties. This result held up when they controlled for race, body mass index in childhood, parental socioeconomic status, childhood health and extraversion.

Social isolation may well have pervasive effects on brain connectivity. If rats are socially isolated when young (a condition that would give rise to feelings of loneliness in humans), neural function and plasticity are altered [44] [45] [46] [47] . In particular, episodes of social isolation can irretrievably alter the function of the prefrontal cortex (the part of the brain that is central to managing our social relationships [see below]), as well as its axon myelinisation (the laying down of the fatty sheaths around neurons that enable them to transmit signals faster and more efficiently) [44] . While short periods of loneliness in humans rarely have any long-term adverse outcomes, persistent loneliness escalates the risk of Alzheimer's disease and depression [48, 49] . Loneliness also leads to poor sleeping habits, with adverse psychological and physiological consequences [50] .

The fact that friends can have such dramatic effects on our health and well-being may lead us to suppose that the more friends we have, the better. However, the number of friends and family relationships we can manage at any given time is limited by cognitive constraints to ~150 [51, 52] .

There is, however, considerable individual variation, with social network sizes ranging between approximately 100-250. A number of fairly conventional factors are responsible for this variation: age (younger people typically have larger social networks than older people [53] ), sex (females usually have larger social networks than males [53, 54] ; though this does vary with age [55] ), personality (extraverts have larger social networks than introverts [56] ; women who score high on the neuroticism personality dimension have fewer acquaintances than those who score lower [57] ).

Friendships, however, require the investment of considerable time to create and maintain.

The emotional quality of a friendship depends directly on the time invested in a given social link [57] [58] [59] . One prospective study estimated that it takes around 200 hours of face-to-face contact over a three-month period to turn a stranger into a good friend [60] . Conversely, the emotional quality of a J o u r n a l P r e -p r o o f 6 relationship will decline rapidly ( Figure 1 ) if contact rates drop below those appropriate to the relationship quality [61] .

Time resources, however, are naturally limited: we devote only around 20% of our day to direct social interaction (excluding business-related interactions), equivalent to about 3.5 hours per day [62] . Given that our relationships are not all of equal value to us (friends serve a variety of different functions for us [63, 64] ), we allocate our valuable time across our social network in such a way as to maximise the different benefits that friends of different quality provide [65] . This dynamic results in a specific social fingerprint that is unique to each of us [66] .

Nonetheless, there are some broadly consistent patterns: a 40% share of our time is devoted to our five closest friends and family, and a further 20% to the ten next closest individuals. In other words, 60% of the 3.5 hours a day we spend in social interaction are devoted to just 15 people.

Social partners in the outermost layers of the social network each receive just 30 secs of our time a day on average. This gives rise to a very distinctive layering to our social networks, with layers that have a characteristic fractal pattern: the innermost layers of closest friends is very small (typically 5 people) but intense, the outermost ( ~150) very large but more casual [67, 68] . It is that inner circle of five closest friends and family that seems to matter most in terms of the buffering of both loneliness and disease.

Geographical distance also imposes strong constraints on the organization of friendship. The '30-min Rule' provides an empirical reminder that people are less willing to visit friends and family who live more than 30 mins away -no matter whether that involves travel on foot, by bicycle or by car [69] . Cutting across this effect is the influence of genetic relatedness: the kinship premium (i.e., the strong mutual benefits that kinship typically affords) incentivizes us to travel an extra mile to maintain contact with family than we are with friends [70] .

While the role of close contacts, like friends, is pivotal, other regular contacts can also contribute to one's social capital. Previous authors [71] famously claimed that weak -as opposed to strong, or close -ties provide important sources of external information. Analyses of information flow in social networks suggest that sources outside the 50 closest friendships offer few benefits [72] . Other benefits of interaction with more loose social ties can, of course, include heightened subjective well-being and sense of belonging to the local community [73] . However, as is often the case in such studies, it is crucial to precisely define the meaning of weak versus strong ties, since all weak ties belonged to the same community (a student class). Regular interaction with different people at the periphery of social networks can give rise to heightened perceived social and emotional fulfillment in ways that act as psychological buffers [24] , although this might depend on personality or social style [74] .

Social-affective processes in the presence of others take a different form than during the others' physical absence. Already in a nursery, if a baby starts crying, other nearby babies hear the distress signal and typically also start crying by mere emotional contagion. In addition to utterances and prosody, humans tend to align their communication towards each other by imitating vocabulary, grammar, mimics and gestures. For instance, humans tend to unconsciously synchronize their facial expressions even with people who are directing gaze at somebody else [75] . Such subliminal motor and emotional resonance is typically found to be intrinsically rewarding [76] . On the positive side, contagion processes can uplift an individual's happiness through people within the close neighborhood, but also miles apart [19] . On the negative side, loneliness also spreads rapidly through an individual's social interaction partners, thus affecting even friends of friends of friends [77, 78] .

Reading others' faces -impossible during a conventional phone call -may be an evolutionarily conserved means for exchanging pivotal information, which coevolved with the corresponding decoding machinery in brain and behavior responses (see next section). Faces offer a plethora of social information about an individuals' sex, age, ethnicity, emotional expression and potentially their intentions and mental state (all of which influence the strength of the bond between two individuals [59] ). Throughout development, learning and maturing critically hinge on joint attention of two individuals on the same object [79, 80] . Such mentalizing and eye gaze processes have been repeatedly linked to the higher associative and the striatal reward circuitry [79, [81] [82] [83] . Some authors even argue that the importance of such facets of interpersonal exchange may explain why humans developed wide and white sclera in the eyes -more easily visible than in most animals [84] . What may lead to greater vulnerability to predators for some species (by making the individual and her intentions more visible and exploitable) may have boosted learning and cooperation in human primates [85] . Such evolutionary adaptations facilitate how humans automatically represent the (visual) perspective of nearby others. Making statements about objects channels [88] . Compared to actual interpersonal encounters, a surprising number of psychological constants exist in how humans entertain and juggle with social relationships in digital environments.

For example, the upper bound of ~150 contacts (cf. above), as well as the structure of these networks, appears to hold across both the real world and a variety of virtual online contexts [53, 68, 89, 90] , suggesting that group size in today's society is still orchestrated by the same principles as when we were hunter-gatherers. Indeed, several neuroimaging studies [e.g., 51, 91] broadly confirm that our online social networks correlate with the volumes of the same core brain regions that resonate with the size of our offline networks [52, 92] .

These constancies suggest that lively virtual social interaction may similarly entrain faculties like memory and concept generation. Conversely, paucity of social interaction and loneliness may have deleterious effects on the cognitive and memory systems. It is conceivable that enhancement or decline of cognitive and neural reserve may be mediated by analogous pathways potentially involving dendritic arborization in the hippocampal and prefrontal regions [49] . The need for personalized interactions may already be reflected in the way that stock market traders sometimes add coded numbers to money transfers (e.g., 10,000,467 instead of 10,000,000 shares) as a potential replacement for the recognition of somebody's unique facial identity rather than remaining anonymous [93, 94] . This attractor for a full range of face-to-face cues during social interactions may explain why emojis have become so popular: they replace the important emotional signals in the absence of the ostensive facial cues that we use for the interpretation of utterances in the face-toface environment.

These considerations raise the important question how the brain implements toggling between real-world social interactions and virtual or imagined social interaction in the absence of physical contact [79] . The right temporoparietal junction was proposed as a key switching relay between two antagonistic classes of neurocognitive processes: those more anchored in one's current external sensory environment and more stimulus-independent ones relying on internally generated information [95] . This idea was later substantiated by a multi-modal neuroimaging study in 10,000 humans [96] : the right and left temporoparietal junction explained most variation in functional coupling changes between all major brain networks. Hence, these two association cortex regions may help mediate shifts of focus from the person in front of you to a person you are texting with on the phone, who is out of sight or touch.

Taken together, evidence of digital communication suggests that this new medium does not in fact change the general pattern of our social interactions or the numbers of people we contact [68, 89, 90, 97] . The sizes of the layers in our social networks are unchanged by using digital media or virtual communication. Also, the frequencies with which we contact certain people in each social J o u r n a l P r e -p r o o f layer are strikingly similar in the online and offline worlds. Some digital vehicles, however, lack the communicative richness of real face-to-face interactions: when asked to rate their satisfaction with interactions with their five closest friends each day, participants rated face-to-face and skype interactions as equally satisfying and both as significantly more satisfying than interactions with the same individual by phone, text messaging, SMS messaging, email or text-based social media such as Facebook [87] .

Human and non-human primates live in groups mainly to minimize external ecological threats, including predators, raiding by neighbors, and environmental risk. Advanced forms of cooperation are rare in non-primate species [98, 99] and probably emerged in non-human primates several million years ago. Today, the average humans spends up to 80% of waking hours in the presence of others [100, 101] . Investing cognitive resources in keeping track of friends, family and colleagues is highly demanding -more costly than contemplating the physical facts [102, 103] . Not only time limits (cf. above) but also neurocognitive limits [e.g., 104] effectively constrain how close one can be to how many individuals. But how is regular social stimulation reflected in neurobiology?

In monkeys [105, 106] and in humans [51, 52, 107, 108] , various indices of sociality and measures of social network size are robustly associated with specific regions of the neocortex. These same regions are responsible for processing social information such as predicting others' intentions [109, 110] . At least some of these brain-behavior associations may be cross-culturally consistent in humans, as evidenced by a structural neuroimaging study in the USA and China [111] . Whole-brain analyses have repeatedly highlighted a relationship between the ventromedial prefrontal cortex and measures of social network complexity and social competence [92, 105, 110, [112] [113] [114] [115] . The ventromedial prefrontal cortex and striatal nucleus accumbens have been found to play a key role in both social reward behaviors and the amount of social stimulation in humans [113] and other mammals [e.g., 44, 47] . Functional neuroimaging has shown that these neural correlates are also implicated in tracking others' popularity status in real-world social networks [116] . Similarly, positron emission tomography has shown that, in humans, the density of mu-receptors for betaendorphin, especially in the ventromedial prefrontal cortex, correlates with social attachment style, for which endorphins are more important than other neuropeptides [117] . Other evidence, such as in a functional neuroimaging study on maintenance and manipulation of social working memory [104] , has also related the dorsomedial prefrontal cortex to social network properties. There are similar correlations for social cognitive skills like mentalizing that are crucial to maintaining functional social relationships [118] [119] [120] .

Analyses of social richness and brain morphology in humans tend to identify a neural network involving the prefrontal cortex with several parts of the so-called default mode network as being crucial for managing social networks (e.g., Noonan et al., 2018) . This major brain network of the higher association cortex has probably recently expanded in primate evolution [121] . Its constituent regions are often thought to support several of the most sophisticated neurocognitive processes [122, 123] . In monkeys, there is evidence that experimental manipulation of social group size results in adaptations in the volume of frontal brain regions, the posterior superior temporal sulcus or temporo-parietal junction, as well as the amygdala and other parts of the limbic system [105, 106] . In humans, there is evidence for structural coupling between social network size measured by number of online friends and parts of the default mode network, including the hippocampus [51] . From a clinical perspective, functional connectivity alterations in the default mode network have been demonstrated as a consequence of feelings of loneliness in younger adults [124] . Moreover, the default mode network is especially subject to vulnerability in normal cognitive aging [125] , and is among the main brain circuits to be impacted by neuropathology in Alzheimer's disease [126, 127] .

Complementing higher associative parts of the human social brain [128] , amygdala volume is larger in individuals with more extensive social networks in humans [52, 107] . Amygdalar functional connectivity was also reported to increase with canonical brain networks implicated in face perception and approach-avoidance behaviour [107] . Indeed, previous authors reported [129] that a patient with complete bilateral amygdala lesions lacked a sense of appropriate personal space vis-àvis other people (Figure 3 ). This patient exhibited no discomfort when at close distances from another person, even to the point of touching the other's nose -despite the fact that their conceptual understanding of people's private physical space was intact. In contrast, healthy individuals typically show amygdala activation in response to close personal proximity. In a similar vein, the grey-matter volume of the amygdala correlated negatively with social phobia [130] . The amygdala may hence be required to trigger the strong emotional reactions normally associated with personal space violations, thus regulating interpersonal distance in humans.

Such reports on the social brain often seemed to be in conflict about whether they highlight the prefrontal cortex or the amygdala of the limbic system. This apparent discrepancy was reconciled in a recent population neuroimaging study [131] : social traits such as daily exchange with family, friends, and work colleagues were associated with brain morphology in ~10,000 UK Biobank participants. Particularly prominent findings were reported in the limbic system, where volumes varied consistently with various indicators of social isolation. Less socially stimulated participants showed volume effects in various parts of the social brain including the ventromedial prefrontal J o u r n a l P r e -p r o o f cortex and the amygdala, in addition to the nucleus accumbens of the reward circuitry. Volume effects in these regions were reported for several markers of brittle social integration, such as living in a socially "emptier" household, knowing fewer individuals with whom to regularly share experiences and concerns, feeling unsatisfied with one's friendship circles, as well as having grown up without brothers or sisters and being unhappy with one's family situation [131] . This analysis also demonstrated wide-ranging sex differentiation in how traits of social isolation are linked to brain morphology. These findings underscore evidence from animals for a sex specific co-evolutionary relationship between the primate brain and social complexity [social brain hypothesis: 132, 133].

The perspective of brain network integration in loneliness was investigated in a seminal neuroimaging study of intrinsic functional connectivity in ~1,000 humans [124] . Careful analysis showed that feelings of loneliness especially affect the neural communication strength between the limbic system and the default mode network as well as the communication strength inside of the default mode network. As a particularly discriminatory pattern for loneliness, impoverished functional modularity was found for the default mode network and its interacting brain networks. In contrast, a positive sense of one's meaning in life was linked to strengthened functional differentiation of the canonical network ensemble. The collective evidence led the investigators [124] to argue that the default mode network and its coupling partners represents a neural signature reflecting one's own purpose in life versus social disconnection to others.

According to UNICEF estimates, ~140 million children worldwide live deprived of parents who could provide comfort and support. ~8 million of these children grow up in institutions without the socioemotional context of a regular family. In one of the earliest randomized clinical trials of its kind, orphans raised in institutions were systematically compared to orphans who were later welcomed into a foster home [134] . Abandoned children were randomly assigned either to remain under the care of the institution or to transition to the care of foster-parents. Their cognitive trajectories were monitored over several years. Those children who remained in the institution showed significantly lower development indices and lower IQs [of around 70: 134] than the adopted orphans. Being deprived of social bonds with caregivers also led to a pernicious reduction in grey-and white-matter tissue and lower fiber tract integrity as evidenced by brain MRI [134] . Institutional rearing was also shown to exacerbate the decay of the telomeres in cell nuclei [135, 136] . These protection caps normally prevent chromosome deterioration, which acts like a cellular sand clock of aging. Their shortening has major consequences for various biological pathways and health outcomes.

The younger the children were when adopted by a foster family, the better the cognitive performance later [137] . Impoverished cognitive domains include memory and executive function:

For orphans who transitioned to a foster home, some cognitive facets remained below-average throughout later life (e.g., short-term visual memory and attention allocation). Other cognitive dimensions (e.g., visual-spatial memory and spatial working memory) caught up with a normal trajectory at age 16 [134] . Such unique evidence underlines the fact that lack of socioemotional context in early life severely impedes brain development and maturation of the cognitive repertoire, which can be partially mitigated by developing social bonds to non-genetic parents (see Box 2) .

Early psychosocial deprivation also shows inter-generational effects, which are probably mediated through maternal and epigenetic effects [138] . Social isolation in childhood leads to molecular annotations of the genetic strand (such as methylation or phosphorylation of the histones that provide the structure for DNA strands) that are passed on to influence how children cope with stress and in turn how they raise their own children. For instance, in rats, socioemotional experience as a pup has an impact on how the rat's own pups later deal with stress and high anxiety levels [139] . Epigenetic regulation of gene transcription is involved in how maternal care promotes the rat pup's brain development and cognitive maturation. More licking and grooming by the mother increases protein expression of the Grm1 gene in the pup's hippocampus. This up-regulated gene transcription leads to greater availability of glutamate receptor proteins in hippocampal cells for inter-neuronal signaling [140] . In humans, a longitudinal neuroimaging study indeed showed that social support from the mother promotes volume growth trajectories in the hippocampus, and predicts socioemotional development and emotion regulation in early adolescence [141] .

In young rhesus monkeys, loss of social contact to the mother leads to behavioral aberrations that last right into adulthood. Such social isolation was shown to entail down-regulated dendritic growth in the prefrontal cortex and reduction in gene expression in the amygdala [142] .

Social adversity undergone by children with institutional upbringing led to disturbed functional connectivity between the prefrontal cortex and the amygdala [143] . Such perturbed brain maturation through social deprivation may be mediated by glucocorticoids, which are known to be inhibited by maternal care in primates [144] . Hence, maternal care is a critical enrichment of the social environment that promotes maturation, expression of growth hormones, and synaptogenesis in various brain circuits. In contrast, social neglect leads to disturbed social attachment, as well as increased aggression and hyperactivity, often potentially lifelong [145, 146] . How vulnerable an individual is to parental deprivation is subject to complex nature-nurture interactions that are strongly conditioned on personality and overall genetic endowment [147, 148] .

Rats separated early from their mothers were impaired in adult life in emotion regulation and arousal management [149] . Early socioemotional isolation of rat pups had impact on whether these rats later showed healthy responses to stress by mounting adequate cortisol levels [150] .

Hormones of the hypothalamic-pituitary-adrenocortical (HPA) axis are an important endocrine mechanism of stress neurobiology that plays a key role in social isolation. In baboon monkeys, infant survival is jeopardized for mothers who are more socially isolated and not well integrated in the local communities including ties to sisters, adult daughters, and other mothers [151] . Monkey mothers with a thinner social network are less likely to have infants which themselves have high fitness [28] .

Female baboon monkeys with a larger close social circle of grooming partners have healthy cortisol levels and typically deal better with stressful situations [25, 26, 152] . When one of these strong social bonds is disrupted, such as when a close member of the social group is killed by predators, cortisol titres rise in the blood. Such monkeys then tend to seek out new connections to "repair" the lost link in their social network [153] .

A lower-than-usual cortisol level in the morning is indicative of extended stress periods in adults [154] . The same diurnal cortisol dynamic is frequently observed in disturbed child-caregiver relationships [155] . In rhesus monkeys, a low hormone response has been observed after repeated separations from the mother. The same observation has been reported for children who were moved between several caregivers. An intact child-caregiver relationship probably provides a stress reserve to adrenoreceptor responses so that children get over stressful episodes quicker [156, 157] .

After undergoing adversity in early childhood, such as emotional or physical neglect, maltreatment, or maternal separation, enhancement of the child-caregiver relationship can mitigate the effect of previous hits to the HPA system. Early disturbance in important social relationships is linked to dysfunctional cortisol hemostasis in adult life [158] . In some neglected children, ensuing problems and behavioral disruptions can even be exacerbated in adult life [159] . Abnormal blood cortisol levels can potentially be prevented, mitigated or restored by family-based therapy and other interventions [160] . Nonetheless, dysregulated diurnal cortisol levels are further linked to various mental disorders including major depression, substance abuse, and post-dramatic stress disorder [154] , in addition to stress-induced impact on the immune, cardiovascular, and metabolic systems [161, 162] .

Further insight into the neurobiology of social isolation has also been derived from rigorous experiments with adult primates (see also Box 3). In one study, 20 monkeys were separated from others to live alone for 1.5 years [163] . Subsequently, monkeys were re-integrated into social groups of four monkeys housed together. Repeated positron emission tomography (PET) scanning revealed increased levels of D 2 receptors in the basal ganglia, which includes key nodes of the reward J o u r n a l P r e -p r o o f circuitry (see above), after being socially housed. This neurochemical adaptation in the monkeys' brain circuitry was apparent after as few as 3 months of social rehabilitation [163] . These authors also reported several differences in respect of social integration and social rank: monkeys of higher rank were groomed more by others. In contrast, subordinate monkeys spent more time by themselves. As a consequence at the behavioral level, the lower-rank monkeys were also significantly more willing to self-administer cocaine, which may also relate to heightened drug abuse in lonely humans [164] . Such molecular imaging evidence shows that changing from social deprivation to an environment with constant social stimulation causes neural remodeling in the dopaminergic neurotransmitter pathways in non-human primates, which may be clinically relevant for substance abuse disorders in humans.

We are social creatures. Social interplay and cooperation have fuelled the rapid ascent of human culture and civilization. Yet, social species struggle when forced to live in isolation. The expansion of loneliness has accelerated in the past decade. As one consequence, the United Kingdom has launched the 'Campaign to End Loneliness' -a network of over 600 national, regional and local organizations to create the right conditions for reducing loneliness in later life. Such efforts speak to the growing public recognition and political will to confront this evolving societal challenge. These prospects should encourage us to search for means to mitigate possible negative backlash. We offer some suggestions in Box 4. Additional insight into stress-responsive brain systems is imperative to tailor clinical decision making and therapeutic interventions to single individuals.

There is also a dire need for additional longitudinal research on the HPA axis and the cortisol response to psychological stressors.

We are grateful to Guillaume Dumas and Tobias Kalenscher for valuable comments on a previous version of the manuscript. DB was supported by the Healthy Brains Healthy Lives initiative (Canada First Research Excellence fund), and by the CIFAR Artificial Intelligence Chairs program (Canada Institute for Advanced Research).

Primates service their relationships through social grooming. Grooming triggers the endorphin system in the brain through a very specific neural system: the afferent CT fibres [165] .

These axon bundles have receptors at the base of most hair follicles, have the unusual properties of being unmyelinated (and hence very slow, especially compared to the pain receptors in the skin), with no return motor loop (unlike pain and other proprioceptive neurons), respond to a very specific stimulus (light slow stroking at ~2.5cm per sec) and directly trigger the endorphin reward system [166] . Although humans no longer have the full fur covering that encourages social grooming, we still have the receptors and instead use physical contact in the form of touching, stroking, caressing, and hugging as a means for strengthening social ties in our more intimate relationships [167, 168] .

Physical touch is intimate, and hence limited mainly to close family and friends ( Figure 2 ). To bond our wider range of relationships as well as our more intimate ones, humans exploit a number of behaviours that turn out to trigger the endorphin system. These joint activities include laughing [169, 170] , singing [171, 172] , dancing [173, 174] , feasting [22] and emotional storytelling [175] . An important feature of all these behaviours is that behavioural synchrony seems to ramp up the level of endorphin release [174, 176] .

In baby primates, close social interaction is not only beneficial, but critical for maturation and resilience. Experiments in baby monkeys showed that upbringing in social isolation during the first years causes a variety of social deficits. When separated early from their mothers, baby monkeys showed strong symptoms of social withdrawal: self-hurting behaviour like biting, stereotypical and repetitive motor behaviour, excessive avoidance behavior towards others as well as poor social and maternal skills as adults. When separated later from their mothers, baby monkeys tended to indiscriminately approach unknown monkeys without fear [cf. 177] .

Reports of human children in some crowded Russian and Rumanian orphanages painted a strikingly similar picture: socially and emotionally abandoned children showed either forward-J o u r n a l P r e -p r o o f backward rocking tics and social escape or overly strong attachment style, analogous to neglected baby monkeys [cf. 178] . These cases invigorated the then-contested claim that mother-child bonds are indispensable for normal development, and that foster-care parents can compensate many of these needs [134, 160, 179, 180] . Disruption of social interplay during critical development impacts negatively on cognitive, verbal, social and motor performance, and predisposes to mental health issues. In other words, early neglect remains measurable in brain and behaviour in later life.

The socioemotional dialogue between caregiver and baby is mediated in several important ways. Mothers speak to their offspring in "baby talk", which potentially evolved only recently in humans [181] . Accompanied by direct face-to-face exchanges, these communication bouts with characteristic vocabulary and prosody promote infant development milestones. The interpersonal stimulation grabs the baby's attention, she gains weight faster, modulates her emotional state, and enhances various health outcomes. Mother-infant communication is also delivered through direct skin-to-skin contact [166] . Postnatal touching bolsters mother-infant bonding, alleviates anxiety, and provides intrinsic pleasure through endorphin release [182] [183] [184] . Throughout life and quite independent of geography, primate societies are orchestrated by the creation, curation, and cultivation of social bonds though purposeful social closeness.

Among the many consequences of loneliness on body and mind, the scarcity of social contact encourages drug compensation behaviour, such as alcoholism, possibly via non-social rewards triggering dopaminergic neurotransmitter pathways [163] . At the genetic level, loneliness was shown to entail under-expression of anti-inflammatory genes involved in glucocorticoid response and over-expression of genes related to pro-inflammatory immune responses [185] .

Fortunately for future clinical intervention, loneliness may be a modifiable determinant in healthy aging [11] .

As people grow older, the social network typically becomes smaller -naturally diminishing the cognitive stimulation through frequent and intense social interaction on a daily basis, thus potentially reducing the neural reserve. Over the last century, the average human lifespan in developed nations has increased by nearly three decades. On the other hand, older people were also reported to show a decline in the capacity to take other people's point of view, as demonstrated in three separate mentalizing tasks [163] . These authors showed that social cognition J o u r n a l P r e -p r o o f deficits were related to decreased neural activity responses in the medial prefrontal default mode network [163] . This capacity is likely to be particularly important when introspecting other people's minds who are not physically present -where social cues like facial expression, mimics, and gestures are missing.

Both limited social stimulation and weakening social reflection capacities relate to the sense of loneliness in complicated and important ways [13] . Once lonely, bias for negative information processing of cues from others hinders social rehabilitation in a downward cycle [4, 186] . Many recent studies have corroborated the corpus of empirical evidence that the feelings of loneliness escalate the risk of certain neurological diseases and especially Alzheimer's disease in later life [49] .

Social isolation at massive scale risks creating cohorts of individuals who are socially dysfunctional. It may therefore be important to identify ways of mitigating the worst of the effects so as to alleviate the consequences. The following possible countermeasures may be worth exploring:

 One promising intervention would involve creating opportunities where mutual social support relationships (friendships) could develop naturally. You cannot, however, force people to become friends: both parties need to be willing to devote resources to each other in a context where available time budget for social engagement is limited [187, 188] and there are competing friendship interests [66] . However, by providing more opportunities for people to meet in congenial environments, new friendships may blossom.

 Social neuroscientists [189] undertook a longitudinal intervention study on 332 matched adults who underwent regular training sessions. Several months of cognitive training improved empathy for others' affective state or perspective-taking of others' mental state, which resulted in structural remodeling in brain regions belonging to the social brain network, including the frontoinsular network and the default mode network. Daily affective training resulted in thickening of the right anterior and mid-insula, with correspondingly enhanced compassion ratings. Different training regimes correlated with different brain regions. Further research is urgently needed to explore therapeutic interventions using training of social capacities in socially deprived humans.

J o u r n a l P r e -p r o o f  One important lesson is that joining clubs can have important benefits in reducing both a sense of loneliness and psychological or psychiatric conditions [21] . One obvious solution is to encourage vulnerable individuals to join social groups and communities that suit their interests and abilities. Establishing a wide range of such clubs is likely to be much cheaper than paying for carehomes and prisons.

 Singing is known to have a dramatic, immediate effect on creating a sense of social engagement and elevating psychological well-being [the "ice-breaker effect" : 171] .

Vulnerable individuals could be encouraged to join choirs and community singing groups.

Encouragement and funding may need to be invested in establishing a network of choirs.

 Use of video-embedded digital communication is likely to gain in importance. This is especially true where family and friendship groups can meet in the same virtual space. The visual component of the interpersonal encounter appears to play a key role in creating a more satisfying experience of digital social media [87] . Emotional closeness at the start of the study is set at 0 for both groups. Redrawn from [61] .

regions. In 1,368 people from several countries, this study investigated the permissibility of social touch [167] . The authors showed that human social touch is particularly dependent on the nature of the relationship. The topography of accepted social touching depends on many factors, including a) emotional relationship, b) type of interpersonal bond including kinship, c) sex, and d) power dynamics. Close acquaintances and family members are touched for more different reasons. Culture influence, measured in five countries, was small. Female, rather than opposite-sex, touch was evaluated as more pleasant, and it was consequently allowed on larger bodily areas. Reproduced from [167] .  What further refinements of online digital media might improve people's function in creating and maintain friendships, especially for the housebound? It is insufficiently known which types of modern medium best mimic which neurocognitive facets of real social interaction.

 Which neurobiological mechanisms explain how the default mode network and its connections to subordinate brain systems support higher social capacities, and their decline in social deprivation? This associative brain network needs to be more completely understood; especially regarding the congruencies and idiosyncrasies between healthy aging trajectories, the experience of social isolation, and vulnerability to neurodegenerative pathologies. In terms of progress towards causal understanding, putting a premium on longitudinal studies holds out unprecedented promise.

 Across the entire lifespan, to what extent does reduced social stimulation or too few social contacts lead to loss in general capacities of the cognitive repertoire? How much do people struggling with cognitive load have issues maintaining many active social relationships? Or both? Progress in this chicken-and-egg problem will shed light on the aetiopathology of the loneliness, and usher towards new intervention strategies.

The phenotype of loneliness

Evolutionary mechanisms for loneliness

Perceived social isolation and cognition

The growing problem of loneliness

Loneliness and attention to social threat in young adults: Findings from an eye tracker study

The Cultural Context of Loneliness: Risk Factors in Active Duty Soldiers

It is all about who you know: Social capital and health in lowincome communities

Counting on Kin: Social Networks, Social Support, and Child Health Status

Impact of social connections on risk of heart disease, cancer, and all-cause mortality among elderly Americans: Findings from the Second Longitudinal Study of Aging (LSOA II)

Associations of social networks with cancer mortality: A meta-analysis

Social and emotional support and its implication for health

The effect of social relationships on survival in elderly residents of a Southern European community: a cohort study

Social Networks and Health

Social isolation, social activity and loneliness as survival indicators in old age; a nationwide survey with a 7-year follow-up

Social Relationships and Mortality Risk: A Meta-analytic Review

Loneliness and Social Isolation as Risk Factors for Mortality: A Meta-Analytic Review

Social isolation, loneliness, and all-cause mortality in older men and women

The Effect of Widowhood on Mortality by the Causes of Death of Both Spouses

Dynamic spread of happiness in a large social network: longitudinal analysis over 20 years in the Framingham Heart Study

Connected: The surprising power of our social networks and how they shape our lives

Social group memberships protect against future depression, alleviate depression symptoms and prevent depression relapse

Breaking Bread: the Functions of Social Eating

Social structure as a strategy to mitigate the costs of group living: a comparison of gelada and guereza monkeys

Religiosity and religious attendance as factors in wellbeing and social engagement. Religion

Focused grooming networks and stress alleviation in wild female baboons

Social stressors and coping mechanisms in wild female baboons (Papio hamadryas ursinus)

Social affiliation matters: both same-sex and opposite-sex relationships predict survival in wild female baboons

Social Bonds of Female Baboons Enhance Infant Survival

Strong and Consistent Social Bonds Enhance the Longevity of Female Baboons

The benefits of social capital: close social bonds among female baboons enhance offspring survival

Network connections, dyadic bonds and fitness in wild female baboons

Social support reduces stress hormone levels in wild chimpanzees across stressful events and everyday affiliations

Family network size and survival across the lifespan of female macaques

Decoupling social status and status certainty effects on health in macaques: a network approach

Responses to social and environmental stress are attenuated by strong male bonds in wild macaques

Sociality increases juvenile survival after a catastrophic event in the feral horse (Equus caballus)

Social bonds between unrelated females increase reproductive success in feral horses

Social and genetic interactions drive fitness variation in a free-living dolphin population

Loneliness, Social Network Size, and Immune Response to Influenza Vaccination in College Freshmen

Social connectedness is associated with fibrinogen level in a human social network

Opiate Antagonist Prevents μ-and δ-Opiate Receptor Dimerization to Facilitate Ability of Agonist to Control Ethanol-altered Natural Killer Cell Functions and Mammary Tumor Growth

Social relationships and physiological determinants of longevity across the human life span

Friends With Health Benefits: The Long-Term Benefits of Early Peer Social Integration for Blood Pressure and Obesity in Midlife

Postweaning social isolation enhances morphological changes in the neonatal ventral hippocampal lesion rat model of psychosis

Abnormalities of presynaptic protein CDCrel-1 in striatum of rats reared in social isolation: relevance to neural connectivity in schizophrenia

Region-specific impairments in parvalbumin interneurons in social isolation-reared mice

A critical period for social experience-dependent oligodendrocyte maturation and myelination

Feelings of loneliness, but not social isolation, predict dementia onset: results from the Amsterdam Study of the Elderly (AMSTEL)

Loneliness and Risk of Alzheimer Disease

Loneliness Is Associated with Sleep Fragmentation in a Communal Society

Online social network size is reflected in human brain structure

Social brain volume is associated with in-degree social network size among older adults

Emotional arousal when watching drama increases pain threshold and social bonding

Social networks, support cliques, and kinship

Sex differences in social focus across the life cycle in humans

Extraverts Have Larger Social Network Layers: But Do Not Feel Emotionally Closer to Individuals at Any Layer

Individual differences and personal social network size and structure

Social Relationships and the Emergence of Social Networks

The Anatomy of Friendship

How many hours does it take to make a friend?

Managing Relationship Decay: Network, Gender, and Contextual Effects

The social brain hypothesis

Closeness, loneliness, support: Core ties and significant ties in personal communities

Different Strokes from Different Folks: Community Ties and Social Support

Hamilton's rule predicts anticipated social support in humans

Persistence of social signatures in human communication

Social network size in humans

Calling Dunbar's numbers

Did distance matter before the Internet?

Going That Extra Mile: Individuals Travel Further to Maintain Face-to-Face Contact with Highly Related Kin than with Less Related Kin

The Strength of Weak Ties

Structure and function in human and primate social networks: Implications for diffusion, network stability and health

Social Interactions and Well-Being: The Surprising Power of Weak Ties

Limited communication capacity unveils strategies for human interaction

What's in a smile? Neural correlates of facial embodiment during social interaction

Fairness and cooperation are rewarding: evidence from social cognitive neuroscience

Alone in the crowd: the structure and spread of loneliness in a large social network

Attraction and close relationships

A second-person approach to other minds

Joint attention, shared goals, and social bonding

The Computation of Social Behavior

Reward value of attractiveness and gaze

ALE meta-analysis on facial judgments of trustworthiness and attractiveness

The eyes have it: the neuroethology, function and evolution of social gaze

Unique morphology of the human eye

Seeing it their way: Evidence for rapid and involuntary computation of what other people see

Effects of Duration and Laughter on Subjective Happiness Within Different Modes of Communication: Happiness and mode of communication

Technology-Mediated Communication in Familial Relationships: Moderated-Mediation Models of Isolation and Loneliness

Human Social Networks

The structure of online social networks mirrors those in the offline world

The social network-network: size is predicted by brain structure and function in the amygdala and paralimbic regions

Ventromedial prefrontal volume predicts understanding of others and social network size

The Modular Neuroarchitecture of Social Judgments on Faces

The social brain: allowing humans to boldly go where no other species has been

Characterization of the temporo-parietal junction by combining datadriven parcellation, complementary connectivity analyses, and functional decoding

Subspecialization within default mode nodes

Use of Social Network Sites and Instant Messaging Does Not Lead to Increased Offline Social Network Size, or to Emotionally Closer Relationships with Offline Network Members

Coevolution of neocortical size, group size and language in humans

Games people play-toward an enactive view of cooperation in social neuroscience

A Survey Method for Characterizing Daily Life Experience: The Day Reconstruction Method

Gossip, reputation, and social adaptation

Higher order intentionality tasks are cognitively more demanding

Mental time travel into the past and the future in healthy aged adults: An fMRI study

Social Working Memory Predicts Social Network Size in Humans

Social network size affects neural circuits in macaques

Baboons (Papio anubis) living in larger social groups have bigger brains

Amygdala volume and social network size in humans

Orbital prefrontal cortex volume predicts social network size: an imaging study of individual differences in humans

Ventromedial prefrontal volume predicts understanding of others and social network size

Orbital prefrontal cortex volume correlates with social cognitive competence

Gray matter volume of the anterior insular cortex and social networking

Segregation of the human medial prefrontal cortex in social cognition

The brain structural disposition to social interaction

The structural and functional brain networks that support human social networks

Orbital prefrontal cortex volume correlates with social cognitive competence

Neural mechanisms tracking popularity in real-world social networks

Adult attachment style is associated with cerebral μ-opioid receptor availability in humans: Opioids and Attachment

Building blocks of social cognition: Mirror, mentalize, share?

Social cognition and the brain: A meta-analysis

Are there theory of mind regions in the brain? A review of the neuroimaging literature

Surface-based and probabilistic atlases of primate cerebral cortex

Dark Control: The Default Mode Network as a Reinforcement Learning Agent. Hum Brain Mapp

The brain's default network: anatomy, function, and relevance to disease

Loneliness and meaning in life are reflected in the intrinsic network architecture of the brain

Social-Cognitive Deficits in Normal Aging

Cortical Hubs Revealed by Intrinsic Functional Connectivity: Mapping, Assessment of Stability, and Relation to Alzheimer's Disease

Structural covariance of the default network in healthy and pathological aging

Computing the Social Brain Connectome Across Systems and States

Personal space regulation by the human amygdala

Reduced amygdalar and hippocampal size in adults with generalized social phobia

000 Social Brains: Sex Differentiation in Human Brain Anatomy. Science Advances

Neocortex evolution in primates: the 'social brain' is for females

Primate brain evolution: genetic and functional considerations

Cognitive Recovery in Socially Deprived Young Children: The Bucharest Early Intervention Project

Telomere length and early severe social deprivation: linking early adversity and cellular aging

Accelerated telomere shortening: Tracking the lasting impact of early institutional care at the cellular level

Long-term effects of institutional rearing, foster care, and brain activity on memory and executive functioning

Nongenomic Transmission Across Generations of Maternal Behavior and Stress Responses in the Rat

Behavior of adult rats is modified by the experiences their mothers had as infants

Variations in postnatal maternal care and the epigenetic regulation of metabotropic glutamate receptor 1 expression and hippocampal function in the rat

Preschool is a sensitive period for the influence of maternal support on the trajectory of hippocampal development

Amygdala Gene Expression Correlates of Social Behavior in Monkeys Experiencing Maternal Separation

Early developmental emergence of human amygdala-prefrontal connectivity after maternal deprivation

Psychobiological mechanisms underlying the social buffering of the hypothalamic-pituitary-adrenocortical axis: A review of animal models and human studies across development

Maternal Care, Gene Expression, and the Transmission of Individual Differences in Stress Reactivity Across Generations

Early Adversity and Critical Periods: Neurodevelopmental Consequences of Violating the Expectable Environment

Beyond diathesis stress: Differential susceptibility to environmental influences

Biological sensitivity to context: II. Empirical explorations of an evolutionary-developmental theory

Critical periods for the effects of infantile experience on adult learning

Maternal and Environmental Influences on the Adrenocortical Response to Stress in Weanling Rats

Social relationships among adult female baboons (papio cynocephalus) I. Variation in the strength of social bonds

Sociophysiology of relationships in squirrel monkeys. I. Formation of female dyads

Behavioural and hormonal responses to predation in female chacma baboons ( <i>Papio hamadryas ursinus</i> )

Stress and disorders of the stress system

The potential role of hypocortisolism in the pathophysiology of stressrelated bodily disorders

Maternal behavior predicts infant cortisol recovery from a mild everyday stressor

Dampening of Adrenocortical Responses during Infancy: Normative Changes and Individual Differences

Disentangling psychobiological mechanisms underlying internalizing and externalizing behaviors in youth: Longitudinal and concurrent associations with cortisol

Effects of a therapeutic intervention for foster preschoolers on diurnal cortisol activity

Importance of Studying the Contributions of Early Adverse Experience to Neurobiological Findings in Depression

The Influence of Social Hierarchy on Primate Health

Social dominance in monkeys: dopamine D2 receptors and cocaine selfadministration

Loneliness and alcohol abuse: A review of evidences of an interplay

The neurophysiology of unmyelinated tactile afferents

Social touch modulates endogenous μ-opioid system activity in humans

Topography of social touching depends on emotional bonds between humans

Cross-cultural similarity in relationship-specific social touching

Sharing the joke: the size of natural laughter groups

Social Laughter Triggers Endogenous Opioid Release in Humans

The ice-breaker effect: singing mediates fast social bonding

Singing and social bonding: changes in connectivity and pain threshold as a function of group size

Naltrexone Blocks Endorphins Released when Dancing in Synchrony

Synchrony and exertion during dance independently raise pain threshold and encourage social bonding

Emotional arousal when watching drama increases pain threshold and social bonding

Rowers' high: behavioural synchrony is correlated with elevated pain thresholds

Social Recovery by Isolation-Reared Monkeys

Research Network on Early Experience and Brain Development

Early Adverse Care

Romania's abandoned children

Function of infant-directed speech

Stroking modulates noxious-evoked brain activity in human infants

The Communicative Functions of Touch in Humans, Nonhuman Primates, and Rats: A Review and Synthesis of the

Infants Autonomic Cardio-Respiratory Responses to Nurturing Stroking Touch Delivered by the Mother or the Father

Social regulation of gene expression in human leukocytes

Examining the visual processing patterns of lonely adults

Time as a limited resource: Communication strategy in mobile phone networks

Cognitive resource allocation determines the organization of personal networks

Structural plasticity of the social brain: Differential change after socioaffective and cognitive mental training