How Friendships Evolve From Childhood to Adulthood

Friendships evolve from childhood to adulthood because the psychological job they do keeps changing. Selman’s (1980) stage model traces children moving from egocentric play-partners to intimate, mutually chosen confidants by adolescence; Carstensen’s (1999) socioemotional selectivity theory explains why the pattern reverses again in later life, trading breadth for depth as time feels shorter.

Why Friendships Change Over Time

Friendship isn’t a single relationship type that simply thins out with age. It performs a different psychological function at each life stage, and the function determines the shape. A five-year-old’s “best friend” is often whoever is standing next to them at the sandbox; a fifteen-year-old’s best friend is a co-author of their emerging identity; a seventy-year-old’s best friend is often the one relationship left that reliably delivers warmth without complication.

Three forces drive the shift, and they recur at every stage covered in this article: cognitive development (a child’s ability to model another person’s inner life expands for over a decade), structural opportunity (school, college, and work generate contact automatically — or don’t), and time horizon (how much life you perceive as remaining changes what kind of relationship feels worth investing in). None of these forces act alone. A teenager’s obsession with peer approval is cognitive and structural; an older adult’s small, tight circle is a time-horizon effect and a response to decades of accumulated maintenance cost.

Childhood Friendship Development

Young children’s understanding of friendship develops in a predictable sequence, mapped in detail by Selman’s (1980) structural-developmental model — and the age bands deliberately overlap rather than snapping shut at a fixed birthday. In the earliest stage — roughly ages 3 to 6 — a “friend” is essentially whoever is physically present and willing to play; the relationship is defined by momentary proximity, not by any stable trait of the other child. Roughly ages 5 to 9, children move into “one-way assistance,” judging friends mainly by what they receive from them, without yet grasping that the other child has needs of their own.

The genuine shift happens around ages 7 to 12, in what Selman called “fair-weather cooperation”: children begin to consider a friend’s perspective alongside their own — though not yet hold both at once — and grasp that friendship involves give-and-take, not just receiving. Holding self and other’s perspective simultaneously is the next stage, “intimate-mutual sharing,” which spans roughly ages 8 to 15: trust, loyalty, and emotional disclosure become the defining features, and a friend is valued for who they are rather than what they provide in the moment.

This sequence matters practically because it explains why childhood friendships look interchangeable from the outside — any classmate can become “best friends” within a week — while still doing real developmental work. Early friendships are a low-stakes training ground for cooperation, fairness, and conflict resolution, skills that get tested for real once identity enters the picture in adolescence.

Adolescence: Identity, Intimacy, and Peer Influence

Adolescence is where friendship stops being a training ground and becomes central to identity itself. Erikson (1968) framed the teenage years around the crisis of identity vs. role confusion — the task of figuring out who you are, separate from your family. Peers become the primary mirror for that project, which is why teenage friendships carry an intensity that looks disproportionate to adults: approval from friends functions as evidence for or against the identity a teenager is actively constructing.

That intensity follows a specific curve, and it’s not the one researchers used to assume. Steinberg & Monahan (2007) found little change in resistance to peer influence between ages 10 and 14, then a roughly linear rise in that resistance from age 14 to 18, with little further change from 18 to 30. Independent judgment hardens fastest during the late-teen years, not gradually across the twenties. This is also the period when friendship’s emotional bandwidth expands fastest: disclosure, loyalty, and jealousy all intensify, because for the first time a friend can function as a genuine confidant rather than just a playmate.

For a closer look at how that intensity plays out in specific friend-group dynamics, see how friend groups drift apart — many of the mechanisms that fracture adult friend groups first appear, in miniature, during the high school years.

Young Adulthood: Network Expansion and Opportunity Structures

Friendship networks reach their lifetime peak in the mid-twenties, and the driver is structural, not emotional. Feld (1981) described the mechanism as “foci of activity”: shared organizational settings — a college campus, a first job, a shared apartment — mechanically generate friendship by putting the same people in repeated, low-stakes contact. College is arguably the most efficient friendship-generating machine most people will ever encounter, because it stacks several foci (dorm, major, clubs) on top of each other simultaneously.

Bhattacharya et al. (2016) put a number on it, using mobile-communication records from 3.2 million people: overall network size is largest in the mid-twenties and shrinks steadily from there for the rest of adulthood. That high point lines up almost exactly with the stretch of life when foci of activity are thickest and least scattered — before careers, geography, and partnerships start pulling people apart into separate worlds. The thirties drop-off isn’t friendship failing; it’s the foci disappearing one at a time, leaving the friendships they generated to run on deliberate effort instead.

The Time Economics of Adult Friendship

Adulthood often makes recurring contact harder to arrange. Hall (2019) found an association between shared time and closeness; the estimates do not turn a particular number of hours into a guaranteed friendship. A standing activity creates opportunities without setting a deadline for either person.

Less repeated contact can make adult friendship harder to develop. It is one influence among several, including health, confidence, access, caring responsibilities, and whether both people want more contact. Use the friendship time-budget calculator to plan opportunities, not to predict when closeness must arrive.

Midlife: Dyadic Withdrawal, Life Roles, and Friendship Types

Midlife friendship is shaped by a specific, well-documented pattern: dyadic withdrawal. Milardo, Johnson & Huston (1983) studied courting couples and found that as a new romantic relationship deepens — at any age, not only in midlife — couples systematically narrow contact with the more peripheral members of their social networks, concentrating time on each other especially during the early, intense phase. This isn’t dysfunction — it’s a predictable reallocation of finite social time, and midlife just adds more occasions to trigger it: a relationship after a divorce, a second marriage, a partner met later in life. But when a couple never widens back out — or when it’s compounded by a first child, which restructures daily schedules even more sharply than a new partner does — the friendships that depended on that free time can quietly disappear.

Midlife also multiplies the number of roles competing for the same hours — spouse, parent, employee, sometimes caregiver to aging parents — each with its own claim on time and identity. Friendships that survive this period tend to be the ones that adapt their format: fewer spontaneous hangouts, more scheduled ones; less frequent contact, but contact that’s protected rather than left to chance. The friendships that don’t adapt their format are the ones most likely to fade — not because the affection is gone, but because the old format no longer fits the life stage. For the mechanics of exactly why this fading happens and how to reverse it, see why friendships fade as you get older.

In our experience, the friendships that survive this squeeze tend to split into two recognisable types: circumstantial friendships, which exist because a shared context — a job, a kid’s school run, a neighborhood — puts you in repeated contact, and chosen friendships, which persist because both people keep choosing the contact once that shared context ends. Most midlife circles are a mix of both, and the ratio shifts as roles pile up: circumstantial friendships tend to fade fastest once the context that generated them disappears, while a friendship that has already survived one change of context is more likely to survive the next one too.

Older Adulthood: Socioemotional Selectivity and the Positivity Effect

The pattern reverses again in later adulthood, and this reversal is the best-documented shift in the entire friendship lifespan. Carstensen (1992, 1999) proposed socioemotional selectivity theory: as people’s perceived remaining time horizon shrinks, they deliberately shift their social goals from information-gathering and novelty — useful when time feels open-ended — toward emotionally rewarding, well-known relationships. The result is a smaller, more curated circle, chosen on purpose rather than left over by attrition.

Mather & Carstensen (2005) documented a companion mechanism: the “positivity effect,” where older adults’ attention and memory skew toward positive over negative information relative to younger adults. Combined with socioemotional selectivity, this means older adults aren’t just keeping fewer friends — they’re experiencing the ones they keep more warmly, because both what they pay attention to and who they spend time with are being actively filtered for emotional payoff.

Carstensen and colleagues distinguish knowledge-seeking goals from emotional goals; perceived time horizons can change their relative priority. Bereavement, illness, accessibility, and unwanted isolation can also shrink a network. A smaller circle is not necessarily freely chosen.

The Friendship Recession in Modern Society

Every life-stage pattern above assumes some baseline level of friendship-forming opportunity — and that baseline has been shrinking across cohorts, independent of individual aging. The Survey Center on American Life’s (2021) national survey, led by Daniel Cox, found the share of American men reporting zero close friends rose roughly fivefold, from about 3% in 1990 to 15% in 2021, with the average number of close friends reported falling across both sexes over the same period.

The structural causes track directly onto the mechanisms already covered here. Sociologist Oldenburg (1989) coined “third places” for the cafés, clubs, bowling leagues, and other informal gathering spots outside home and work that once functioned as reliable foci of activity for adults, not just students; Putnam (2000) documented their decline as part of a broader erosion of civic and informal association across American life. Remote and hybrid work has removed another major foci-of-activity source for a large share of adults. Geographic mobility increases the frequency with which any given friendship loses its shared context. None of these forces are new individually; what’s new is how many are compounding simultaneously across an entire generation.

Health and Psychological Effects of Social Disconnection

The stakes here are physical, not just emotional. Holt-Lunstad, Smith & Layton’s (2010) meta-analysis of 148 studies covering more than 300,000 participants found that strong social relationships predict a higher survival odds (pooled odds ratio 1.50) — an effect size comparable to quitting smoking, and larger than the protective effect of exercise or maintaining a healthy weight. This wasn’t a fringe finding: it held across age, sex, and cause of death.

The U.S. Surgeon General’s 2023 advisory on the loneliness epidemic cited comparable evidence, describing chronic social disconnection as carrying a mortality risk on the order of smoking up to 15 cigarettes a day. The advisory’s broader point is the one worth taking seriously: friendship maintenance isn’t a soft, optional nicety squeezed in around “real” obligations . The evidence supports taking social connection seriously; it does not establish that friendship maintenance has the same causal effect as diet or exercise.

Practical Takeaways for Building and Maintaining Friendships Across Life Stages

The mechanisms above point to the same handful of practical moves at every life stage:

1. Manufacture the foci you no longer get automatically. After school and college stop generating contact for free, deliberately join something recurring — a class, a league, a standing group — that puts you in repeated, low-stakes contact with the same people. This is the single closest adult substitute for what school did by default.
2. Budget the hours, don’t hope for them. Hall’s estimates suggest that closeness often develops over substantial shared time, with large individual differences. The friendship time-budget calculator can help make room for contact without treating those estimates as deadlines.
3. Expect and plan around dyadic withdrawal. A new partner or a first child will pull time away from friendships for a while. That’s normal — the mistake is letting it become permanent instead of consciously restoring contact once the intense phase passes.
4. Curate on purpose, especially past midlife. Fewer, closer friendships aren’t a decline — they’re the pattern Carstensen’s (1999) research describes as healthy and deliberate. Don’t mistake a smaller circle for a failing one if the depth is there.
5. Treat maintenance as a health behavior. Given the mortality evidence above, a check-in can be a practical way to maintain contact. The friendship checkup tool is a quick reflection that surfaces which of your friendships are quietly cooling, so you know where to reach out first.