What the Marshmallow Really Measured

A single marshmallow sits on a plate. A four-year-old stares at it, kicks the table leg, hums a tune, presses both hands over her eyes. The rule was explained a moment ago and it was simple enough for a preschooler to grasp: the researcher is leaving the room, and if the marshmallow is still there when he returns, she gets two. If she eats it, she gets nothing more. Fifteen minutes stretch out like an eternity. Some children last only seconds. A few hold out to the end, trembling with the effort of not tasting sugar that sits inches from their mouths.

This scene, filmed and re-filmed in nursery schools across the world, became the most recognizable experiment in the history of psychology. It launched a thousand parenting articles and nearly as many TED talks. It gave the culture a tidy, hopeful story: that the small child who could wait was destined to become the successful adult, and that the trait doing the predicting was willpower. Teach a child to resist the marshmallow, the reasoning went, and you were teaching her to resist every temptation that would ever stand between her and a good life.

It was a beautiful theory. It was also, in the form most people came to believe it, wrong. When later scientists tried to reproduce the famous result, the magic mostly evaporated. And when they looked closely at what the children were actually doing on that plastic chair, they found something more interesting than willpower. They found children making a shrewd calculation about whether the adult in front of them could be believed at all.

The Nursery at Stanford

The man behind the test was Walter Mischel, an Austrian-born psychologist who fled the Nazi occupation of Vienna as a boy and eventually built a career studying human character.¹ In the late 1960s he was running experiments at the Bing Nursery School on the Stanford campus, and he was interested in a deceptively ordinary human ability: delayed gratification, the capacity to give up something good now for something better later.

The design was elegant in its plainness. A researcher would sit a child at a table, place a treat in front of her (a marshmallow, a pretzel, a cookie, depending on the child’s preference), and explain the deal. Wait until the researcher came back, and you get two. Ring the bell to summon him early, or simply eat the treat, and you get one. Then the adult left, and a hidden camera recorded what happened next. Around six hundred children passed through these early Stanford trials over the years.²

What Mischel noticed first was not a moral difference between children but a difference in technique. The successful waiters were not simply gritting their teeth against desire. They were distracting themselves with visible ingenuity. They sang songs, invented games, turned away from the plate, talked to themselves, and in some cases seemed to pretend the marshmallow was not really there. The ones who caved tended to keep staring at the very thing they were trying to resist. Mischel drew a lesson that has aged well: attention was the lever. “If you think about the marshmallow,” he liked to say, “you can’t resist it.” The children who lasted had learned, however crudely, to think about something else.

That observation about mental strategy was subtle and probably correct. It is not the part of the marshmallow test that conquered the world. The part that conquered the world came later.

The Follow-Up That Made It Famous

Years after those first trials, Mischel and his colleagues went looking for the children who had once fidgeted in the Bing Nursery. They were teenagers now, and the researchers gathered reports on how they were doing. The pattern they described was striking. The children who had waited longer as preschoolers appeared to be flourishing as adolescents. Their parents rated them as more competent. And in a much-repeated statistic, the long-waiters scored, on average, more than two hundred points higher on the SAT than the children who had grabbed the treat quickly.³

Later studies in the same tradition extended the story into adulthood, linking childhood self-control to lower body weight, better stress management, and fewer behavioral problems. A single choice at a nursery table seemed to cast a shadow across an entire life. The conclusion wrote itself: self-control in childhood predicts success far into adulthood. And because self-control felt like something you could coach, the finding carried an irresistible promise. If willpower was destiny, then destiny was teachable.

The intuition was so satisfying that it spread faster than the evidence behind it. Willpower became a kind of secular virtue, and the marshmallow became its emblem. But science holds even its most beloved findings to a single, unglamorous demand. Do it again. See if the result survives.

When the Effect Shrank

In 2018, a team led by the psychologist Tyler Watts, working with Greg Duncan and Haonan Quan, set out to test the marshmallow finding on a scale Mischel had never approached.⁴ Where the original samples counted in the hundreds and drew heavily on the children of Stanford faculty and graduate students, Watts and his colleagues assembled a group of more than nine hundred children drawn from a large, nationally representative American study. Crucially, this sample reflected the real spread of the country: different income levels, different family structures, different educational backgrounds. It was not a bubble of tenured comfort.

That distinction turned out to matter enormously. Mischel’s original participants had been, by and large, privileged children living in an unusually stable corner of the world. Their parents were educated, employed, and financially secure. In a population like that, nearly every child has good reason to believe a promise. The variation the test picked up was variation within a group that already had its future more or less assured.

When Watts’s team ran the numbers on their broader sample, the famous link between waiting and later achievement did not vanish entirely, but it shrank dramatically. And once the researchers statistically accounted for a child’s family background (the parents’ education, the household’s income, the home environment) the predictive power of the marshmallow test nearly disappeared. The seconds a four-year-old could hold out told you very little about her academic future that you did not already know from her ZIP code.⁴

What had looked like a measurement of character was, to a large degree, a measurement of circumstance. Children from wealthier and more stable homes tended to wait longer, and they also tended to do better later. But the waiting was not causing the success. Both were downstream of the same thing: a secure environment. As one summary of the reinterpretation put it, the marshmallow was measuring the child’s world, not the child’s willpower.

A Rational Bet on Trust

If environment mattered so much, the next question was mechanical. What exactly about a stable environment made a child better at waiting? A clever experiment had already suggested an answer, and it reframed the entire test.

In 2013, the cognitive scientist Celeste Kidd and her colleagues at the University of Rochester added a single step before the marshmallow ever appeared.⁵ They gave each child an art project and told them they could either use the worn crayons on the table now or wait a moment for a fresh, exciting set of art supplies. Then, for half the children, the researcher delivered exactly what was promised: the wonderful new supplies arrived. For the other half, the researcher came back apologetic and empty-handed. The promised rewards simply never materialized.

Only after this small drama did the classic marshmallow test begin. And the difference was enormous. Children who had just learned that the adult kept her word waited, on average, about four times longer for the second marshmallow than children who had just been let down. Some in the reliable group held out for the full quarter of an hour. Most in the unreliable group ate the marshmallow within minutes.⁵

Read through Kidd’s lens, the marshmallow test looks less like a measure of grit and more like a measure of judgment. A child who has learned that promises are kept has every reason to wait, because waiting pays. A child who has learned that promises are broken is behaving perfectly rationally by eating the marshmallow at once. From that child’s point of view, the second marshmallow is not a reward. It is a gamble against an adult who has already proven unreliable, and the smart move is to secure the sure thing while it is still on the plate. “Why wait,” as the logic goes, “for a second marshmallow that may never come?”

Seen this way, the fidgeting waiters were not morally superior. They were simply lucky enough to live in a world where the future could be trusted. And that trust is not evenly distributed. A child raised in scarcity, where meals are uncertain and adults are stretched thin, learns early that the reliable strategy is to take what is in front of you. That is not a defect of character. It is an accurate reading of the odds.

What the Candy Actually Reflected

Put the three findings side by side and a new picture assembles itself. Mischel’s original work identified a real cognitive skill, the ability to redirect attention away from temptation. But the sweeping claim that grew up around it, the claim that willpower measured in preschool determines success across a lifetime, does not hold once you study a realistic cross-section of children. The correlation was carried, in large part, by wealth and stability. And the mechanism underneath the waiting was not raw self-denial but a reasonable expectation, learned from experience, about whether the world keeps its promises.

The marshmallow test, in other words, was never a clean measure of willpower. It was measuring whether a child could afford to trust the future. Self-control did not cause the good outcomes so much as accompany them, both of them growing out of the same soil of safety. Willpower was less a cause of success than a symptom of security.

It is worth saying that Mischel himself was more careful than the legend that formed around him. In his later writing he warned against reading the results as a verdict on a child’s fixed nature, and he stressed that the skills the test tapped could be taught and that context shaped them heavily.¹ The culture kept the flattering half of his message, that willpower matters, and quietly discarded the cautious half, that willpower is not the whole story and that environment writes most of it.

The Cost of Mistaking Symptoms for Sins

This correction reaches well beyond a plate of candy in a nursery school. The marshmallow story became popular precisely because it fit a comfortable moral instinct, the instinct to credit success to virtue and to read struggle as a failure of will. If discipline is destiny, then the disciplined have earned their good fortune and the impulsive have earned their bad, and the whole unequal arrangement of the world acquires a reassuring gloss of fairness.

But behavior always grows inside a context we rarely see when we sit in judgment of it. A hungry child who grabs food is not exhibiting a weaker soul than a fed child who can wait. A person from a chaotic background who discounts the future is not less virtuous than a person from a stable one who plans for it. They are responding, sensibly, to the worlds they have learned to expect. The candy on the table did not expose a hidden flaw inside the child. It reflected the reliability of everything that had happened to that child before the researcher ever walked in.

That is the quiet lesson buried under decades of willpower talk. What looks like character is often the residue of circumstance, and what looks like weakness is often intelligence adapted to hardship. Before deciding that someone lacks the discipline to wait, it is worth asking a harder and more humane question: what, in the life they have actually lived, were they ever promised, and how often did the promise come true?

Sources

1. Mischel, W., The Marshmallow Test: Mastering Self-Control, Little, Brown, 2014. — https://www.hachettebookgroup.com/titles/walter-mischel/the-marshmallow-test/9780316230865/
2. Mischel, W., Ebbesen, E. B., & Zeiss, A. R., ‘Cognitive and attentional mechanisms in delay of gratification,’ Journal of Personality and Social Psychology, 1972. — https://psycnet.apa.org/record/1972-23530-001
3. Shoda, Y., Mischel, W., & Peake, P. K., ‘Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification,’ Developmental Psychology, 1990. — https://psycnet.apa.org/record/1991-00108-001
4. Watts, T. W., Duncan, G. J., & Quan, H., ‘Revisiting the Marshmallow Test,’ Psychological Science, 2018. — https://journals.sagepub.com/doi/10.1177/0956797618761661
5. Kidd, C., Palmeri, H., & Aslin, R. N., ‘Rational snacking: Young children’s decision-making on the marshmallow task is moderated by beliefs about environmental reliability,’ Cognition, 2013. — https://www.sciencedirect.com/science/article/abs/pii/S0010027712002259
6. Calarco, J. M., ‘Why Rich Kids Are So Good at the Marshmallow Test,’ The Atlantic, 2018. — https://www.theatlantic.com/family/archive/2018/06/marshmallow-test/561779/

Related reading

• The Man in the Gray Coat
• The Basement That Never Was

More from the Mind edition →