The Marshmallow Test: What It Really Predicts

The Marshmallow Test: What It Really Predicts

Most people meet the marshmallow test twice. The first time, it is a charming story: a preschooler alone at a table, one marshmallow on a plate, and a promise — wait, and you get two. The second time, it arrives as a verdict: the children who waited supposedly went on to better grades, better jobs, better lives, and so the capacity to wait must be the thing that separates people who thrive from people who struggle.

That second version travelled far beyond psychology. It shows up in parenting advice, productivity seminars — and in the private way many adults talk to themselves. If you have ever fought with impulses around food, spending, screens, or drinking, you may have quietly filed yourself under “couldn’t wait for the second marshmallow.”

This article is part of a short series on famous psychology experiments and what modern research did to them. The marshmallow test is a good place to start, because what survives scrutiny turns out to be more useful — and considerably kinder — than the legend. (The second piece in the series, on learned helplessness, is the same kind of story.)

What Was the Original Marshmallow Test?

In the late 1960s and early 1970s, psychologist Walter Mischel and his colleagues ran a series of delay-of-gratification studies with preschoolers at Stanford’s Bing Nursery School. A child sat with a treat and a choice: eat one now, or hold out — usually up to about fifteen minutes, alone in the room — and get two.

The original purpose was narrower than the legend suggests. Mischel was studying how children manage to wait: covering their eyes, singing, turning away from the plate. The famous predictive claims came later, in follow-up work — including Mischel, Shoda and Rodriguez’s 1989 paper in Science and a 1990 follow-up by Shoda, Mischel and Peake in Developmental Psychology — where the team reported that preschoolers who waited longer were later rated as more competent adolescents and scored higher on the SAT.

Two details rarely make it into the retelling. The children came largely from Stanford’s own campus community, which is not a cross-section of anything. And the long-term follow-ups rested on much smaller subsamples of that already unusual group — far fewer families than anyone designing the study today would accept.

Did the Marshmallow Test Get Debunked?

Not exactly — but the famous version of it did not survive contact with a larger, fairer dataset.

In 2018, Tyler Watts, Greg Duncan and Haonan Quan published a conceptual replication in Psychological Science. Instead of a university nursery school, they drew on more than nine hundred children from a large American longitudinal study, deliberately more diverse in income and background, and followed them to age fifteen.

Two findings changed the picture. First, the raw association between waiting at age four and later academic achievement was roughly half the size reported in the original follow-ups. Second, and more important: once the researchers controlled for family background, early cognitive ability and the home environment, about two-thirds of what remained disappeared, leaving a small association. Much of the predictive information also sat at the very start of the task — the difference between children who could wait even briefly and those who could not carried most of the weight, and holding out for the entire period added little.

“Debunked” is still the wrong word, and the researchers did not use it. The behaviour is real, and it remains weakly related to later outcomes. A 2020 analysis by Falk, Kosse and Pinger compared the 1990 and 2018 studies directly and argued they are statistically more compatible than the headlines suggested — the live dispute is what the family-background controls mean. Waiting at age four travels together with the circumstances a child grows up in. The real question is whether those circumstances are a nuisance to be subtracted, or the actual story.

What Actually Shapes Whether a Child Waits?

A small, elegant experiment made a strong case that circumstances are the story. In 2013, Celeste Kidd, Holly Palmeri and Richard Aslin published “Rational snacking” in Cognition. Before running the marshmallow task, an adult made each child a small promise about better art supplies. For half the children, the adult kept it. For the other half, the adult came back empty-handed.

Children who had just watched an adult keep their word went on to wait, on average, about four times as long as children who had watched a promise get broken. Same treat, same table, same task — a different two minutes of evidence about whether adults can be trusted.

Seen from the child’s side of the table, grabbing the treat in an unreliable world is not weakness; it is arithmetic. If the adults around you do not follow through, a marshmallow in the hand beats a promised one that may never arrive. Waiting is what children do when their experience says waiting pays.

That is the takeaway that survives the replication debate, and both sides’ data point toward it: the capacity to delay is not a little engine of willpower installed at birth. It develops inside an environment — its stability, its scarcity, and whether the people in it keep their promises.

What Does This Mean If You Struggle With Self-Control?

The legend of the marshmallow test lives on in adult self-talk: I can’t stick to anything. I’m weak-willed. I’ve always been like this. The research suggests a different reading.

If waiting felt pointless where you grew up — because promises were broken, resources were unpredictable, or good things could be taken away — then acting fast on what was available was an adaptation, not a flaw. People who grew up in chaotic or unsafe homes often carry those adaptations long after their circumstances change, and that history is exactly the territory childhood trauma therapy works in.

The reframe also matters for shame. Struggles with impulses — food, spending, alcohol, gambling, screens — routinely get moralized as failures of character, which stacks a second problem on top of the first. Work on addictions and compulsive habits tends to go further when the question shifts from “why am I so weak?” to “what did I learn about scarcity and trust, and what would help me learn something new?” And when that old verdict has hardened into an identity — a running inner commentary about being undisciplined or defective — the voice itself is workable ground in self-esteem focused therapy.

None of this swings to the opposite extreme. The 2018 findings do not say self-control is meaningless, and they do not say childhood decides everything. They say the deterministic story was oversold in both directions: environments shape the capacity to wait, and environments — along with habits, relationships, and the way you speak to yourself — can change.

The Marshmallow Test: FAQ

Does the marshmallow test predict success?

Weakly — and far less than the legend claims. In Watts, Duncan and Quan’s 2018 data, the association between waiting at four and achievement at fifteen became small once family background and early cognitive ability were accounted for, and associations with behaviour problems largely disappeared under the same controls. It is a modest correlate, not a destiny test.

Should I try the marshmallow test on my child?

It will not tell you what the legend promises. A single wait in an unfamiliar situation mostly reflects that moment — including how hungry your child is and how much they trust the adult making the promise, which is precisely what Kidd’s 2013 study manipulated. Treat it as a party trick, not a measurement.

Was Mischel’s original research bad science?

No. The original studies were careful for their era, and their claims were narrower than the retellings. Mischel himself spent his later years pushing back on the destiny reading — his 2014 book The Marshmallow Test argued that delay skills are learnable and heavily dependent on strategy and context. The legend outgrew the data; that part is on the retellers.

If my environment shaped this, can it change?

The environment finding is the hopeful one. What was learned under one set of conditions can be revisited under better ones. Much of therapy is exactly that: a reliable, consistent relationship in which trust, follow-through and self-direction get practised at a manageable pace.

A Low-Pressure First Step

If the self-control story you carry has hardened into shame, that is a workable therapy topic — not a character verdict. At TEO Counselling, Mohamad Shabib practises as a Registered Psychotherapist (Qualifying) with CRPO and works with clients across British Columbia and Ontario, as well as Alberta, Saskatchewan, Manitoba, and Newfoundland and Labrador, with virtual sessions by video. A free 15 minute consultation is a chance to ask about approach, training, and fees, with no pressure to commit. Book a free consultation through the Jane booking platform, or call (519) 760-5211. Evening and weekend appointments are available.

You don’t have to carry it alone.