The Anatomy of Friendship

make-public

For present purposes, I define friends as the people who share our lives in a way that is more than just the casual meeting of strangers; they are the people whom we make an effort to maintain contact with, and to whom we feel an emotional bond. That includes members of our extended family as well as more conventional friends (i. e., people not biologically related to us); is not includes our romantic partners. It it also ‘their uncommon for people to refer to their romantic partner or even their mother/sister as best friend’.

There is now extensive evidence that having friends protects you against both mental and physical illness, helps you recover quicker when you fall sick or have surgery, and makes you feel happier and more contented with life [170–181] (Figure I). Friends have a bigger effect on our susceptibility to disease and the risk we face of dying than anything else except giving up smoking [182]. In this respect, family relationships have the same effect as conventional friendships [183–185]. Similar findings have been reported for primates: female baboons who have more (grooming partners) have lower cortisol titres [186,187], have more offspring and live longer [188–190]. This is also true for wild horses and zebra [191–193]. While friends may promote health directly by giving assistance when needed, bonding activities that upregulate the endorphin system (physical contact, laughter, singing, dancing; Box 3) may also influence health directly through the effect that endorphins seem to have on the immune system [194]. ‘friends’

Data on self-ratings of emotional closefixed, those who ness suggest that, because the amount of social capital we have seems to be have larger networks on average have weaker relationships (they choose to spread their available social capital thinly across a large number of individuals), whereas those with smaller networks prefer to spread it thickly among fewer [28].

A second orthogonal dimension in social networks is the way we rank individuals in terms of emotional closeness. We do not invest either time or emotional capital (the two are in fact closely related; Figure 1) equally in our friends. Something like 40% of our total social effort (whether five most important to indexed as time or emotional closeness) is devoted to just us), with another 20% given to the 10 next most important [10]. In other words, 60% of our social effort is divided between just 15 people (those most likely to provide us with support [10]). five people (the

EGO 1.5 5 15 50 150 500 1500 Figure 2. The Circles of Friendship. An individual (ego) sits within a series of hierarchically inclusive circles, or layers, of friendship that have a very distinct scaling ratio (each layer is three times the size of the one inside it). The heavy line demarcates the 150 layer, the typical size of personal social networks. Beyond this are acquaintances (500 layer); the layer at 1500 appears to represent the limit on the number of faces we can put names to. Frequency of contact, rated emotional closeness, and willingness to act altruistically towards a given alter all decline across the successive circles [2,10].

In humans, the amount of time spent interacting with someone correlates with the perceived emotional closeness (Figure 1) [10,19], and this in turn correlates with the expectation of support [74].

How Time Limits Friendship Networks Time is a limited resource for all animals [77] including humans [78,79], and if the quality (and hence functionality) of a relationship depends on the time invested in it [10], each individual has to decide how to distribute his/her available social effort, or capital, across his/her network. The network layers of Figure 2 appear to be associated with very specific contact values (Figure 3) less often than the defining rate (once a week for the [10,19,20]. If someone 5-layer, once a month for the 15-layer, once a year for the 150-layer) for more than a few is contacted ... months, emotional closeness to that individual will inexorably decline to a level appropriate for the new contact rate [80–84]. That time is evident in romantic relationships: these are so focussed and costly in terms of time investment that, when we fall in love with someone (and hence bring a new person into the 5-layer in the centre of our social world), we typically lose one close family member and one close friend, reducing our to just four people for the duration of the period of infatuation [85]. limits the number of relationships we can have ‘5-layer’

– Not only do personal social networks vary considerably in their interconnectedness [87], but each of us seems to have our own personal signature for how we distribute our social effort fingerprint remains among our network members remarkably stable across time even when there is considerable change in network membership: when a particular friendship dissolves, it seems that we insert a new person into exactly the same emotional slot and see them just as often as the previous occupant [84]. in effect, a unique social fingerprint [84]. This

There is an important distinction between friendships and family relationships: family relationships seem to be much more resilient to lack of contact than friendships [91]. This may be because kinship carries with it an extra dimension of emotional closeness over and above that indexed by emotional closeness rating scales (the [1,2]) or because the kinship subnetwork [88]. Kinship is more networks are often associated with who make it their business to keep everyone updated on the state of the network [92,93]; this probably helps to maintain high levels of interconnectedness which may, in turn, allow relationships to be maintained by lower rates of dyadic contact. interconnected than the friendship subnetwork ‘kinship premium’ ‘kin-keepers’

During the later course of human evolution, when we needed to find new ways to increase the size of our social networks beyond the 50 limit seen in other primates, we had to trigger the endorphin system remotely so that we could bond with several individuals at the same time. In the likely order in which they appeared in human evolution [113], these have included laughter [114,115], singing [116,117], dancing [118,119], and emotional storytelling [120], all of which have been shown (either indirectly by changes in pain threshold or directly by positron emission tomography) to trigger the endorphin system and thereby generate an increased sense of bonding [117–121]. Because they do not involve direct physical contact, these behaviours bypass the intimacy restriction and allow us to create much larger networks.

The functionality of friendships (emotional support, unstinting help) depends implicitly on trust that, over the long haul, the relationship will be in approximate economic balance (i.e., debts will be repaid eventually). While close friendships (those in the innermost layers) may well involve unstinting altruism and, at least in the short term, less emphasis on scorekeeping, score-keeping and the monitoring of reputations are nonetheless likely to become increasingly important in the outer layers.

Developmental studies have identified theory of mind (ToM, the ability to understand another person’s mindstate and ‘suppose’) as a critical transition in children’s social intentions, and hence use words like development [204]. Although ToM (the most basic level of mentalising) may be a uniquely human trait, there is some just one step in a series of recursive evidence to suggest that great apes also possess it [205]. However, ToM is ‘wonders’ mentalising steps (I . . . ) known as the levels of intentionality [206,207]. These higher-order capacities play a crucial role in our ability to manage conversations [136] and fiction [209,210], as well as imposing limits on the number of friends we can our enjoyment of both humour [208] and have [128,129]. ‘supposes’ whether he ‘imagine’, ‘believe’, ‘believe’ ‘intend’, that you that she ‘think’

 important five orders of intentionalMentalising competences seem to have a natural upper limit at about including one’s own mindstate (Box 4). This seems to have ity, implications for conversational dynamics, for example. A number of observational studies of freely forming conversations in a variety of contexts and cultures indicate that there is a consistent upper limit of about four on the number of people that can be involved in a conversation [133–138]. Similar films constraints have even been noted in the sizes of conversations in plays, TV dramas, and [136,139–141]. If a fifth person joins the conversation, it is likely to break up into two or more conversations within a very short time [135,138]. This is an extremely sensitive effect: conversations concerned with factual matters or the mindstates of those actually involved in the conversation have an upper limit at four members, but when the mindstate of someone not physically present is being discussed the conversation has an upper limit of three members (and this is also true of Shakespeare’s plays) [136].

To maintain the coherence and cohesion of our social networks over the long haul required for the benefits they provide, we not only need to manage each of our dyadic relationships, but also need to be sensitive to the complex network of relationships within which these dyadic relationships are embedded: behaving badly towards a friend jeopardise my likely to relationship with that friend’s friends (a reputational effect), and may even invite retribution from them (a policing effect). While our time may be devoted mainly to a small number of people, our cognitive effort needs to be spread more widely because we have to be aware of the consequences that our actions will have for our entire social network. Aside from mentalising, this depends on sophisticated second-order cognitive skills like causal reasoning, analogical reasoning, one-trial learning, the comparison of alternative outcomes and, especially, the ability to inhibit prepotent actions. This set of skills is explicitly associated with the brain’s frontal pole (Brodmann area 10), a brain region that exists only in the anthropoid primates [142] and, at least in humans, forms part of the key centre for mentalising skills in the orbitofrontal and medial prefrontal cortex [128,129,131,143]. is

The two genders also differ in what maintains the emotional closeness of friendships over time. For women, this involves making the effort to spend more time talking together (either face-to-face, by phone or via the Internet), whereas talking has almost no effect on men's friendships; what maintains the emotional quality of men's friendships is increased involvement in activities (sports, drinking, etc.) [91]. Although doing things together does benefit women's investment in friendships, it has much less effect than it does for men.

More detailed analyses of friendship patterns suggest that friendships are based on a limited number of dimensions. We have identified seven key dimensions: language (or, better still, ... (i.e., where you grew up), educational history, hobbies/interests dialect), place of origin (including musical tastes), sense of humour, and worldview (moral views, religious views, political views) [160–162]. Broadly speaking, these seem to be interchangeable: a 4-star relationship can involve any combination of these seven dimensions. Taken together, they constitute the set of beliefs and rituals that remind us who we are, where we come from and why we form a single community with a common set of values and convictions. They thus play an important role in identifying the wider community as well as specific friendships. In respect of friendships, the more of these we share with someone, the more intimate our relationship with them will be, the closer

Communication channel may also be in this context. Participants in a diary study were asked to evaluate the quality of the interactions they five best friends each day: face-to-face and Skype outperformed the phone had had with their and text-based channels [texting, short message service (SMS), email and social networking sites] by a considerable margin [169]. Importantly, perhaps, interactions that involved laughter, whether real or digital (e.g., emoticons), were rated more highly than those that did not. What face-to-face and Skype interactions share is a sense of copresence (being in the same room flow of the together); in addition, they provide visual cues that allow us to monitor and adjust the interaction more effectively (thereby avoiding faux pas, for example) and radically increase the speed of interaction (facilitating repartee, and hence laughter). Single-channel (e.g., phone) or text-based media are simply too impoverished or too slow important

Friendships (including family) are the single most important factor affecting our health and wellbeing. However, friendships are costly to maintain, both cognitively and in terms of the time that needs to be invested in them. These limit the number of friends we can have to around 150, and obliges us to distribute our social time/capital unevenly among them as a function of the benefits they provide us with. The endorphin system seems to play a crucial role in the maintenance of friendships, and many of the behaviours we use in social contexts (laughter, singing, storytelling) seem to be especially good at triggering endorphins.

For present purposes, I define friends as the people who share our lives in a way that is more than just the casual meeting of strangers; they are the people whom we make an effort to maintain contact with, and to whom we feel an emotional bond. That includes members of our extended family as well as more conventional friends (i. e., people not biologically related to us); is not includes our romantic partners. It it also ‘their uncommon for people to refer to their romantic partner or even their mother/sister as best friend’.

There is now extensive evidence that having friends protects you against both mental and physical illness, helps you recover quicker when you fall sick or have surgery, and makes you feel happier and more contented with life [170–181] (Figure I). Friends have a bigger effect on our susceptibility to disease and the risk we face of dying than anything else except giving up smoking [182]. In this respect, family relationships have the same effect as conventional friendships [183–185]. Similar findings have been reported for primates: female baboons who have more (grooming partners) have lower cortisol titres [186,187], have more offspring and live longer [188–190]. This is also true for wild horses and zebra [191–193]. While friends may promote health directly by giving assistance when needed, bonding activities that upregulate the endorphin system (physical contact, laughter, singing, dancing; Box 3) may also influence health directly through the effect that endorphins seem to have on the immune system [194]. ‘friends’

Data on self-ratings of emotional closeness suggest that, because the amount of social capital we have seems to be fixed, those who have larger networks on average have weaker relationships (they choose to spread their available social capital thinly across a large number of individuals), whereas those with smaller networks prefer to spread it thickly among fewer [28].

A second orthogonal dimension in social networks is the way we rank individuals in terms of emotional closeness. We do not invest either time or emotional capital (the two are in fact closely related; Figure 1) equally in our friends. Something like 40% of our total social effort (whether five most important to indexed as time or emotional closeness) is devoted to just us), with another 20% given to the 10 next most important [10]. In other words, 60% of our social effort is divided between just 15 people (those most likely to provide us with support [10]). five people (the

EGO 1.5 5 15 50 150 500 1500 Figure 2. The Circles of Friendship. An individual (ego) sits within a series of hierarchically inclusive circles, or layers, of friendship that have a very distinct scaling ratio (each layer is three times the size of the one inside it). The heavy line demarcates the 150 layer, the typical size of personal social networks. Beyond this are acquaintances (500 layer); the layer at 1500 appears to represent the limit on the number of faces we can put names to. Frequency of contact, rated emotional closeness, and willingness to act altruistically towards a given alter all decline across the successive circles [2,10].

In humans, the amount of time spent interacting with someone correlates with the perceived emotional closeness (Figure 1) [10,19], and this in turn correlates with the expectation of support [74].

How Time Limits Friendship Networks Time is a limited resource for all animals [77] including humans [78,79], and if the quality (and hence functionality) of a relationship depends on the time invested in it [10], each individual has to decide how to distribute his/her available social effort, or capital, across his/her network. The network layers of Figure 2 appear to be associated with very specific contact values (Figure 3) less often than the defining rate (once a week for the [10,19,20]. If someone 5-layer, once a month for the 15-layer, once a year for the 150-layer) for more than a few is contacted ... months, emotional closeness to that individual will inexorably decline to a level appropriate for the new contact rate [80–84]. That time is evident in romantic relationships: these are so focussed and costly in terms of time investment that, when we fall in love with someone (and hence bring a new person into the 5-layer in the centre of our social world), we typically lose one close family member and one close friend, reducing our to just four people for the duration of the period of infatuation [85]. limits the number of relationships we can have ‘5-layer’

– Not only do personal social networks vary considerably in their interconnectedness [87], but each of us seems to have our own personal signature for how we distribute our social effort fingerprint remains among our network members remarkably stable across time even when there is considerable change in network membership: when a particular friendship dissolves, it seems that we insert a new person into exactly the same emotional slot and see them just as often as the previous occupant [84]. in effect, a unique social fingerprint [84]. This

There is an important distinction between friendships and family relationships: family relationships seem to be much more resilient to lack of contact than friendships [91]. This may be because kinship carries with it an extra dimension of emotional closeness over and above that indexed by emotional closeness rating scales (the [1,2]) or because the kinship subnetwork [88]. Kinship is more networks are often associated with who make it their business to keep everyone updated on the state of the network [92,93]; this probably helps to maintain high levels of interconnectedness which may, in turn, allow relationships to be maintained by lower rates of dyadic contact. interconnected than the friendship subnetwork ‘kinship premium’ ‘kin-keepers’

when, during the later course of human evolution, we needed to find new increase the size our social networks beyond the 50 limit in other primates, we had to ways of triggering the endorphin system remotely so that we could several individuals at the same time. In the likely order in which they appeared in human evolution [113], these have included laughter [114,115], singing [116,117], dancing [118,119], and emotional storytelling [120], all of which have been shown (either indirectly by changes in pain threshold or directly by positron emission tomography) to trigger the endorphin system, and thereby generate an increased sense of bonding [117–121]. Because they do not involve direct physical contact, these behaviours bypass the intimacy restriction and allow us to create much larger networks. ‘groom’

The functionality of friendships (emotional support, unstinting help) depends implicitly on trust that, over the long haul, the relationship will be in approximate economic balance (i.e., debts will be repaid eventually). While close friendships (those in the innermost layers) may well involve unstinting altruism and, at least in the short term, less emphasis on scorekeeping, score-keeping and the monitoring of reputations are nonetheless likely to become increasingly important in the outer layers.

Developmental studies have identified theory of mind (ToM, the ability to understand another person’s mindstate and ‘suppose’) as a critical transition in children’s social intentions, and hence use words like development [204]. Although ToM (the most basic level of mentalising) may be a uniquely human trait, there is some just one step in a series of recursive evidence to suggest that great apes also possess it [205]. However, ToM is ‘wonders’ mentalising steps (I . . . ) known as the levels of intentionality [206,207]. These higher-order capacities play a crucial role in our ability to manage conversations [136] and fiction [209,210], as well as imposing limits on the number of friends we can our enjoyment of both humour [208] and have [128,129]. ‘supposes’ whether he ‘imagine’, ‘believe’, ‘believe’ ‘intend’, that you that she ‘think’

 important five orders of intentionalMentalising competences seem to have a natural upper limit at about including one’s own mindstate (Box 4). This seems to have ity, implications for conversational dynamics, for example. A number of observational studies of freely forming conversations in a variety of contexts and cultures indicate that there is a consistent upper limit of about four on the number of people that can be involved in a conversation [133–138]. Similar films constraints have even been noted in the sizes of conversations in plays, TV dramas, and [136,139–141]. If a fifth person joins the conversation, it is likely to break up into two or more conversations within a very short time [135,138]. This is an extremely sensitive effect: conversations concerned with factual matters or the mindstates of those actually involved in the conversation have an upper limit at four members, but when the mindstate of someone not physically present is being discussed the conversation has an upper limit of three members (and this is also true of Shakespeare’s plays) [136].

To maintain the coherence and cohesion of our social networks over the long haul required for the benefits they provide, we not only need to manage each of our dyadic relationships, but also need to be sensitive to the complex network of relationships within which these dyadic relationships are embedded: behaving badly towards a friend jeopardise my likely to relationship with that friend’s friends (a reputational effect), and may even invite retribution from them (a policing effect). While our time may be devoted mainly to a small number of people, our cognitive effort needs to be spread more widely because we have to be aware of the consequences that our actions will have for our entire social network. Aside from mentalising, this depends on sophisticated second-order cognitive skills like causal reasoning, analogical reasoning, one-trial learning, the comparison of alternative outcomes and, especially, the ability to inhibit prepotent actions. This set of skills is explicitly associated with the brain’s frontal pole (Brodmann area 10), a brain region that exists only in the anthropoid primates [142] and, at least in humans, forms part of the key centre for mentalising skills in the orbitofrontal and medial prefrontal cortex [128,129,131,143]. is

The two genders also differ in what maintains the emotional closeness of friendships over time. For women, this involves making the effort to spend more time talking together (either face-to-face, by phone or via the Internet), whereas talking has almost no effect on men’s friendships; what maintains the emotional quality of men’s friendships is increased (sports, drinking, etc.) [91]. Although doing things together does benefit women’s investment in friendships, it has much less effect than it does for men.

More detailed analyses of friendship patterns suggest that friendships are based on a limited number of dimensions. We have identified seven key dimensions: language (or, better still, ... (i.e., where you grew up), educational history, hobbies/interests dialect), place of origin (including musical tastes), sense of humour, and worldview (moral views, religious views, political views) [160–162]. Broadly speaking, these seem to be interchangeable: a 4-star relationship can involve any combination of these seven dimensions. Taken together, they constitute the set of beliefs and rituals that remind us who we are, where we come from and why we form a single community with a common set of values and convictions. They thus play an important role in identifying the wider community as well as specific friendships. In respect of friendships, the more of these we share with someone, the more intimate our relationship with them will be, the closer

Communication channel may also be in this context. Participants in a diary study were asked to evaluate the quality of the interactions they five best friends each day: face-to-face and Skype outperformed the phone had had with their and text-based channels [texting, short message service (SMS), email and social networking sites] by a considerable margin [169]. Importantly, perhaps, interactions that involved laughter, whether real or digital (e.g., emoticons), were rated more highly than those that did not. What face-to-face and Skype interactions share is a sense of copresence (being in the same room flow of the together); in addition, they provide visual cues that allow us to monitor and adjust the interaction more effectively (thereby avoiding faux pas, for example) and radically increase the speed of interaction (facilitating repartee, and hence laughter). Single-channel (e.g., phone) or text-based media are simply too impoverished or too slow important

Friendships (including family) are the single most important factor affecting our health and wellbeing. However, friendships are costly to maintain, both cognitively and in terms of the time that needs to be invested in them. These limit the number of friends we can have to around 150, and obliges us to distribute our social time/capital unevenly among them as a function of the benefits they provide us with. The endorphin system seems to play a crucial role in the maintenance of friendships, and many of the behaviours we use in social contexts (laughter, singing, storytelling) seem to be especially good at triggering endorphins.