What Wellbeing Actually Is: PERMA, Meaning, and the Honest Science of a Good Life

What to know

• Wellbeing is not one feeling. Researchers separate how you rate your life overall from the good and bad emotions of the moment — and the two don’t always move together.
• The strongest finding here is that people with stronger relationships tend to live longer and healthier lives. But it’s a correlation, not proof — hold even the most famous study lightly.
• Money helps wellbeing more than the old ‘$75,000 plateau’ suggested, but on a log scale, so each extra step of happiness costs proportionally more.
• You can probably move your wellbeing somewhat — genes set a range, not a fixed point — but deliberate practices give modest, often short-lived gains, not magic.
• A life can feel good without feeling meaningful, and the reverse. Meaning and pleasure overlap but are not the same thing.

Ask ten people what makes a good life and you’ll get ten answers, most of them confident, few of them tested. “Be happy” sounds like advice until you try to say what happiness actually is. Is it the warm rush of a good evening? A life you’d choose again if you could see the whole thing? A sense that what you do matters? These are different things, and pretending they’re one word — happiness — is the first place most people get lost.

This article is a map, not a doctrine. It won’t tell you the right way to live. It will lay out what careful researchers have measured, how much they actually know, and where the honest answer is still “we’re not sure.” That last part matters more than it sounds. The internet is full of people who are certain about your happiness. Almost none of them will tell you when the evidence is thin. We will.

Happiness isn’t one feeling

Start with the most useful idea in the whole field: wellbeing has parts that come apart.

Original marker: observational · OECD, 2013; Diener, 1984 In wellbeing research and in official measurement guidelines such as the OECD’s, subjective wellbeing is commonly treated as having distinct components rather than a single quantity, separating life satisfaction — a cognitive, reflective evaluation of one’s life — from positive and negative affect, the momentary emotional states; these components are empirically distinguishable but only moderately correlated. In plain terms: you can rate your life as good and still have rough days, or feel fine moment to moment while sensing your life is going nowhere. Researchers also draw a second line. Original marker: observational · Ryan & Deci, 2001; OECD, 2013 Scholarship commonly distinguishes hedonic wellbeing, defined in terms of pleasure and the avoidance of pain, from eudaimonic wellbeing, defined in terms of meaning, purpose, and self-realization; this is a widely used distinction reflected in official measurement guidance, though the two are strongly correlated in practice and whether eudaimonia is a genuinely separate dimension remains debated.

Here’s the first teaching moment about the method itself. Even where the field broadly agrees on the parts, it doesn’t agree on how they fit. Original marker: contested · Busseri & Sadava, 2011 While the three core components — life satisfaction, positive affect, negative affect — are widely accepted, how they fit together structurally, whether as separate components, a hierarchy, a causal system, a composite, or configurations, remains unresolved; one peer-reviewed review identified several competing models and concluded it is premature to propose a definitive structure. So when someone hands you a tidy diagram of happiness, the honest response is curiosity, not belief.

One popular diagram is PERMA. Original marker: observational · Goodman et al., 2018; Butler & Kern, 2016 Introduced in Martin Seligman’s 2011 trade book Flourish, PERMA defines wellbeing through five elements — Positive emotion, Engagement, Relationships, Meaning, and Accomplishment — and has been widely adopted as a framework in positive psychology research and practice. It’s a genuinely good checklist for noticing what might be missing from a life. But “widely adopted” is not the same as “proven true,” and here the candor cuts close to home. Original marker: observational · Goodman et al., 2018 In a single cross-sectional self-report study of 517 US adults, the higher-order factor underlying PERMA correlated near-perfectly with a subjective-wellbeing factor — a latent correlation around r = 0.98 — leading the authors to conclude that overall PERMA and subjective wellbeing converge onto a single general wellbeing factor; the result is correlational and does not show the individual PERMA facets are redundant with one another. The measure built for it holds up better as a tool than as a theory. Original marker: observational · Butler & Kern, 2016 The PERMA-Profiler, tested across large samples, showed acceptable fit for a five-factor model with adequate reliability and convergent and divergent validity; the authors noted engagement was the weakest factor, and good model fit does not by itself establish that the five elements are conceptually independent constructs. Use PERMA as a flashlight, then — a way to ask “which of these am I living without?” — not as proof that life divides cleanly into five.

The you that lives it, and the you that remembers it

There’s a strange gap between experiencing a thing and recalling it, and it bends how we judge our own lives. Original marker: observational · Redelmeier & Kahneman, 1996; Kahneman et al., 1993 In experiments and clinical studies, people’s retrospective evaluation of a painful episode tends to track the most intense moment and the final moment, while the episode’s overall duration has comparatively little effect — a pattern called “duration neglect.” It produces choices that look irrational on paper. Original marker: observational · Kahneman et al., 1993 In a controlled cold-water experiment, most participants chose to repeat a longer trial that contained all the pain of a shorter trial plus extra time at a slightly less painful temperature, suggesting an easier ending can make an objectively worse experience preferred in memory. And it’s not just lab quirks. Original marker: observational · Redelmeier, Katz & Kahneman, 2003 In a randomized trial of 682 colonoscopy patients, briefly extending the procedure with a less painful final interval lowered patients’ remembered final-moment pain — 1.7 versus 2.5 on a 10-point scale — and made the overall experience rated as less unpleasant, 4.4 versus 4.9 on the same scale. The takeaway isn’t “manipulate your memories.” It’s humbler and more useful: the story you tell about an experience is not the same as the experience, and you partly choose which one to trust.

What actually predicts a good life

If you want one finding to take seriously, take this one. Original marker: observational · Holt-Lunstad et al., 2010 A 2010 meta-analysis of 148 prospective studies, with 308,849 participants, found that people with stronger social relationships had roughly a 50% greater likelihood of survival over follow-up; the authors described the magnitude as comparable to quitting smoking and larger than obesity and physical inactivity. Because the underlying studies are observational, this is a correlation, not proof that relationships cause longer survival. Read that last sentence twice — it’s the whole spirit of this site. The most quoted relationship study is the same way. Original marker: observational · Waldinger, Harvard Gazette, 2017 In the Harvard Study of Adult Development, a cohort begun in 1938, the directors report that participants’ satisfaction with their relationships around age 50 predicted their later physical health better than their midlife cholesterol levels — but this is a prediction within one non-representative, originally all-male US cohort, not experimental evidence that relationships cause better health. A careful person can act on this — invest in people — while still holding it loosely. Both can be true at once.

Money is where you get to watch science correct itself in public. For years the headline was a plateau. Original marker: observational · Kahneman & Deaton, 2010 In a 2010 analysis of more than 450,000 daily survey responses, Kahneman and Deaton reported that day-to-day emotional well-being rose with household income only up to roughly $75,000 a year and then leveled off, while overall life evaluation kept rising; this is a cross-sectional correlation, and later work — including a 2023 paper co-authored by Kahneman — indicates the sharp $75,000 plateau was partly an artifact of a coarse yes/no measure. Then better data arrived. Original marker: observational · Killingsworth, 2021 A 2021 study using over 1.7 million real-time smartphone reports from 33,391 employed US adults found that in-the-moment well-being kept rising roughly linearly with log income with no plateau above $75,000; but the sample was self-selected app users, the relationship is on log income — so each further gain requires proportionally larger dollar increases — and the design is correlational, not causal. Rather than just argue, the researchers teamed up. Original marker: observational · Killingsworth, Kahneman & Mellers, 2023 A 2023 adversarial collaboration reanalyzed the data and concluded happiness generally keeps rising with log income for most people, while the curve flattens mainly for an already-unhappy minority, roughly the least-happy 15 to 20 percent; this is a reanalysis of a single observational dataset, and published letters dispute whether cross-sectional data can support the causal framing. Notice what happened: a famous claim, better measurement, a public revision. That cycle is what trustworthy looks like.

Can you move the needle?

Maybe you’re stuck with the happiness you were born with. The data says: partly, but far less rigidly than the slogans claim. Original marker: observational · Bartels, 2015 Twin studies estimate that a moderate share of the differences in subjective well-being between people is heritable, with meta-analytic estimates converging on roughly 36 to 40%; older single studies citing figures near 50% sit at the high end and are not the meta-analytic consensus. The famous “you always bounce back” story is shakier than its fame suggests. Original marker: contested · Brickman, Coates & Janoff-Bulman, 1978 The 1978 Brickman study found that lottery winners did not differ significantly in happiness from controls while paralyzed accident victims were significantly less happy; because it was a small one-time cross-sectional comparison with no follow-up, it documents a snapshot, not the complete long-term adaptation it is often cited to prove. Original marker: observational · Lucas, 2007; Diener, Lucas & Scollon, 2006 Large nationally representative panel studies show people only partly adapt to some major life events — long-term disability, for example, was associated with moderate-to-large lasting drops in life satisfaction with incomplete recovery — indicating the strict fixed-set-point model is overstated, even though these are group averages with wide individual variation.

And the tidy pie chart you may have seen? Watch it deflate, honestly, with its own authors helping. Original marker: observational · Lyubomirsky, Sheldon & Schkade, 2005 The popular “happiness pie” — roughly 50% genes, 10% circumstances, 40% intentional activity — originates with Lyubomirsky, Sheldon, and Schkade in 2005, who presented the figures as approximate estimates suggested by past research, not as exact or causally established measurements. Original marker: contested · Brown & Rohrer, 2020 A 2020 critique by Brown and Rohrer found little empirical evidence for the specific 50/10/40 decomposition, arguing it conflates population-level correlations with individual determinants, likely understates circumstances, and omits measurement error. Original marker: observational · Sheldon & Lyubomirsky, 2021 In a 2021 response, the original authors said they mostly agreed and that the pie chart “appears to have outlived its usefulness,” while maintaining that people can raise their well-being through intentional activity — an effect they put at around 15% of variance, based on meta-analytic and quasi-experimental rather than large-RCT evidence.

So deliberate practices do something — just don’t expect fireworks. Original marker: observational · Bolier et al., 2013; White et al., 2019 Meta-analyses of randomized trials find that positive psychology interventions — gratitude, savoring, acts of kindness, using personal strengths — tend to produce small, mostly short-term gains in well-being, though the effect shrinks under stricter methods and the studies show signs of publication bias. Original marker: contested · White, Uttl & Holder, 2019 When small-study and publication bias are corrected, the effect on depressive symptoms appears to shrink close to zero and is often no longer statistically significant; this rests mainly on a single reanalysis, and concerns non-clinical interventions rather than established treatments such as therapy or medication. The most honest map is the most specific one. Original marker: observational · Folk & Dunn, 2024 A systematic review limited to preregistered experiments found some widely recommended strategies, such as random acts of kindness, lacked reliable supporting evidence, while a few others, such as being more sociable, more dependably raised happiness; because only around four dozen preregistered studies qualified, several strategies remain barely tested and these conclusions are tentative.

Meaning is its own thing

Finally, a feeling many people know but can’t name: a life can be pleasant and feel empty, or hard and feel deeply worth it. The science backs the instinct. Original marker: observational · Baumeister et al., 2013; Steger et al., 2006 Research treats a sense of meaning or purpose and momentary pleasure as overlapping but separable: survey studies find they share many predictors yet diverge on others, and meaning in life can be measured with instruments distinct from positive affect. Meaning may even register in the body. Original marker: observational · Hill & Turiano, 2014; Alimujiang et al., 2019 In large prospective cohort studies of US adults, a higher self-reported sense of purpose is associated with lower subsequent all-cause mortality, even after adjusting for affect; because these are observational, they show association rather than causation and cannot rule out reverse causation or confounding.

Where this leaves you

Here is the map, then. Wellbeing isn’t one thing; it’s how you feel moment to moment, how you judge your life overall, and whether it feels like it’s for something. Relationships are the steadiest bet, money helps with sharply diminishing returns, your genes set a range rather than a verdict, and effort buys real but modest movement. None of this is a command. It’s a set of questions worth asking yourself — and a set of claims you’re invited to doubt, check, and outgrow.