People often report arriving, somewhere around their forties, at a distinction between being alone and being lonely. It tends to get described as a stage: something you reach once the noise of early adulthood settles down.
The stage part is the bit we cannot find support for. In the lifespan data, average loneliness does not have a midlife moment.
This is a description of a common pattern, not a diagnosis of anyone.
What the lifespan data actually shows
Marcus Mund and colleagues published a meta-analysis in Personality and Social Psychology Review in 2020, pooling 75 longitudinal studies from Asia, Australia, Europe and North America covering 83,679 people. They examined two separate things: whether average loneliness rises or falls with age, and whether individuals hold their position relative to each other over time.
On the first, loneliness fell through childhood and then stayed essentially stable from adolescence into oldest old age.
On the second, the rank order proved about as stable as a personality trait, and it was this stability, not the level of loneliness itself, that followed an inverted U across the lifespan. Stability was lowest at the young end and peaked in midlife. The distinction is easy to garble, so to be plain: nothing in this paper says loneliness rises and then falls with age. It says the ordering of people holds together more tightly in the middle of life than at the start of it.
None of which forecasts any individual. Pooled averages are built by cancelling out exactly the movement a person cares about in their own case.
What the flat mean does undercut is the threshold story. Were something clicking for most people around forty, the average would move. Across 75 studies it does not.
Where the change actually appears: between generations
Midlife loneliness has shifted. It shows up when you hold age constant and compare the people born in different decades.
Frank Infurna of Arizona State University, with co-authors at Humboldt University Berlin and Brandeis, published an analysis in American Psychologist, online on 18 March 2024 and available in full through PubMed Central. It draws on three harmonised longitudinal surveys: the US Health and Retirement Study, the English Longitudinal Study of Ageing, and SHARE, which covers eleven European countries plus Israel. The full sample runs to 59,030 people, and a response counted only when the participant was aged between 45 and 65.
That age restriction is what makes the comparison work. Everyone is answering from the same stretch of life, so a difference between them is a difference of birth cohort or country.
Midlife loneliness in the United States was steadily elevated for baby boomers and Generation X compared with the Silent Generation, and ran well above the levels in peer European nations. Europe did not move as one bloc. England and Mediterranean Europe rose across the later-born cohorts. Continental and Nordic Europe were stable or slightly declining.
Loneliness here was measured with the three-item scale developed by Hughes and colleagues, which asks how often people feel they lack companionship, feel left out, and feel isolated. The items deliberately avoid the word loneliness. That is a sensible design choice, and it is also a reminder that the whole edifice rests on what people are willing to report about themselves, which is itself shaped by the culture doing the reporting.
The press release and the paper do not quite agree
The American Psychological Association’s press release on the study leads on a widening gap, with younger cohorts reporting more loneliness than older ones. That is accurate as a statement about generations within a country.
The paper’s own conclusion adds something the release does not. Having noted that American midlife loneliness sits much higher than in peer European nations, the authors write that because of historical increases, middle-aged adults in England and Mediterranean Europe today are “not that far behind the U.S.”
So one gap widened and another narrowed, and only one of those made the release.
The authors attribute the cross-national pattern to cultural norms, socioeconomic conditions and social safety nets: higher residential mobility in the United States, weaker family ties, job insecurity, income inequality, and thinner provision on family leave, unemployment protection and childcare. Every one of those is interpretation. The design compares countries and cohorts on a loneliness score; it does not test parental leave against loneliness outcomes, and it cannot rank the candidate causes. The authors say as much, calling for “more mechanism-oriented research to identify reasons.”
Their framing of the overall picture is worth borrowing even so. They argue it looks less like an epidemic, which they gloss as an outbreak that spreads rapidly, and more like something endemic, “regularly occurring within an area or community.” Endemic conditions are properties of a place. They tend to answer to structural change more readily than to individual effort.
What the solitude experiments measured
If the calendar does not deliver the alone-versus-lonely distinction, the experimental literature points at a different variable, and the paper itself gestures at it. In their closing limitations the authors note it would be intriguing to examine historical shifts in “related, yet distinct constructs such as solitude,” adding that parts of such aloneness reflect personal choice.
Thuy-vy Nguyen, Richard Ryan and Edward Deci ran four studies published in Personality and Social Psychology Bulletin in 2018 on what time alone actually does. Solitude produced a deactivation effect, damping both positive and negative high-arousal feeling. The fourth study followed 173 recruited undergraduates over two weeks of diaries, with 157 in the analysed sample. Autonomous motivation for solitude was associated with less loneliness and more calmness, but it did not moderate that headline deactivation effect on high-arousal affect.
A 2022 PLOS ONE study from Nguyen with Netta Weinstein and Ryan, which we covered on 1 August, found introversion predicted neither preference for solitude nor self-determined motivation for it, while dispositional autonomy predicted the motivation and was unrelated to the preference.
Both point at how much say a person had, not at temperament and not at age. Time alone that someone selected behaves differently in these measures from time alone that arrived by default, through a move, a redundancy, a bereavement or a diary that quietly emptied.
Which is a duller explanation for the forties than a developmental one, and a more testable one. That decade is often when people acquire some control over their own time, or find out they have not.
What none of this establishes
The Mund meta-analysis pools studies using different loneliness measures across four continents, and pooling costs precision. Its stability estimates are also centred on a one-year interval, so it does not license claims about who will be lonely decades later.
The Infurna analysis covers ages 45 to 65 and high-income countries only, as the authors note. Its Silent Generation data exists for the United States and England but not for the SHARE countries, and the authors state plainly that the basis for inferences about Generation X is “much smaller than for the earlier-born generations,” even though the Generation X results mirror the wider pattern. The measure captures the isolation facet of loneliness rather than its relational or collective facets, which the authors suggest may well have moved differently.
The solitude studies were run on undergraduates in fifteen-minute laboratory episodes and a fortnight of diaries. Nobody has yet run the equivalent on people in the life stage this article is about.
The paper’s own last word on the question is that we do not know why certain nations show elevated midlife loneliness and others do not. Everything above is a description of where the variation sits, and the variation sits between countries and birth years rather than along the arc of a single life.
