The idea that social relationships affect health and longevity has enjoyed considerable popularity ever since the seminal paper by Berkman and colleagues showed a protective effect of social networks on survival.¹ Numerous studies have reported that being connected to people seems to confer various forms of health benefit. The paper by Giles and colleagues in this issue of the journal follows this line of research,² showing that social relationships remain an important health resource into very old age. The paper further illustrates that this health benefit may be restricted to social relationships that tend to be more discretionary, such as those with friends and relatives other than children. Similar findings have been reported for other important health outcomes in older age, both by this group of investigators³ and others.^4,5 Yet as appealing the idea is that social relationships, and friendships in particular, are good for our health, as elusive remains our understanding of why or how this is so.

The list of answers to this question is potentially long and complex. The usual suspects have been covered previously, and are nicely summarised in the paper by Giles et al.² There is little doubt that declining health affects the ability to maintain an extensive and active network of social contacts. This source of confounding is a vexing problem in observational research, given the imprecise variables of social function and health we typically are able to collect in large scale studies. But residual confounding attributable to poor measurement can hardly be a satisfactory explanation for a finding that has been validated in so many different settings. Two sets of theoretical perspectives have been offered to interpret the association between social relationships and health. The first perspective emphasises the health implications of the broader social-structural conditions in which we live. This perspective has been popularised by the notion of social capital, which is usually defined as the sum total of a person’s social connections, including the norms, values, and trust that build and sustain these connections.⁶ By and large, social capital is thought to promote health through the resources that flow between inter-connected members of social networks, social support often being thought of as one of the primary resources.⁷

The second perspective has put greater emphasis on the psychological mechanisms that form the putative link between social relationships and health. Insights from theoretical frameworks such as social cognitive theory⁸ and self determination theory⁹ have underscored the importance of social determinants in promoting feelings of competence and perceived control that regulate behaviour, including health improving behaviour. In other words, social contexts and social interactions shape patterns of behaviour that may be of direct consequence to physical and mental health. For example, in a study of older adults McAuley and colleagues have shown that social support had a significant positive effect on exercise self efficacy, which in turn predicted higher physical activity levels and greater maintenance of activity levels over time.¹⁰ It is of note that a recently proposed conceptual framework is an attempt to assimilate these two somewhat distinct perspectives into one model that incorporates the entire sequence of causal events, ranging from the broader structural “up-stream” conditions that shape individual behaviour to the intra-individual “down-stream” psychological and physiological mechanisms that link social relationships with concrete disease processes.¹¹

So why do friends seem to matter more than children? Giles et al suggest that discretionary ties, such as those with friends, provide a greater survival benefit than the less discretionary ties with children. As they indicate in their paper, socioemotional selectivity theory may offer an explanation by predicting that older adults become increasingly selective as they age in the type and number of social relationships in which they invest emotional attachment.¹² Thus, perhaps because of filial obligation and responsibility, older adults tend to turn to members of the nuclear family for emotional and instrumental support, especially when they experience or anticipate increasing dependence because of declining health.⁵ Another explanation may be that relationships with friends, more so than those with children, is marker of a person’s ability to develop and maintain the sort of connections with other people and groups that form the basis of social capital. The specific social capital resources that produce tangible health benefit can be thought of as the supports available in one’s social network or broader social environment that are conducive to healthy lifestyles—better access to healthy food, better dietary habits, more physical activity—and minimises exposure to stressful personal, work related, or neighbourhood related circumstances. Social capital may also include the resources available in networks of people that facilitate access to (health related) information and services, and even facilitate forms of political organisation that may be used to improve access to and quality of care.⁷ But friendships could be salubrious for other reasons. One could argue that friendship often is an end onto itself, valued for its own sake, offering intrinsic rewards that fulfil basic psychological needs of competence, autonomy, and relatedness.⁹ In other words, friendship, feeling connected to other human beings who are valued, trusted, and loved, may provide meaning and purpose that is essential to our human condition, and perhaps to longevity as well.

So far, the exact mechanisms by which friendship and related positively valued social relationships and conditions affect health remain poorly specified. Perhaps they act through a psychological equivalent of the notion of physiological reserve, a source of surplus psychic energy or capacity that generates a physiological benefit, which the organism applies to slow down chronic disease processes that typically lead to disability and death as we age. All we need to do is to identify the biological substrate of this surplus capacity, measure it in human beings, show that is it related to friendship networks and other positive social interactions, and test the degree to which it slows down disease, disability, and death. Although this may sound like a daunting task, a precedent exists in the form of the concept of stress, which portends to summarise and unify the negative experiences we encounter in our interactions with others and the world around us. There are also emerging ideas about a common physiological pathway such as embodied by the concept of allostatic load that links stress with the development of degenerative diseases.¹³ Perhaps we will find one day a similar unifying concept and common physiological pathway for the totality of our positive social experiences, a concept that hopefully will be defined on the basis of more than merely the absence of stress.