In the late 1960s, Walter Mischel and his colleagues at Stanford offered preschoolers a choice between one treat immediately or a larger reward if they waited — and early follow-ups found that the children who waited longer tended to earn higher SAT scores and cope better as teenagers, although later, larger studies showed the test was far less predictive of adult success than the famous story suggests.

The marshmallow test became famous because it seemed to offer a clean story about childhood self-control. A preschooler waits for a larger reward, and years later that child appears to do better in school, stress and social life. It is an irresistible story, which is part of the problem.

The actual evidence is more interesting and less tidy. Walter Mischel and colleagues were studying how children manage temptation, attention and delay, not trying to build a fortune-telling device for adult success. This is not parenting advice or a diagnostic tool. It is a reading of a research history that became much simpler in public memory than it was in the data.

The core task grew out of Stanford work on delay of gratification. In the classic setup, a young child was offered a smaller treat immediately, or a larger reward if they could wait until the experimenter returned. The details varied across studies, including whether the reward was visible, whether children were given distraction strategies, and what treat each child preferred.

What the original work actually showed

The early Stanford papers were mainly about attention. In a 1970 Journal of Personality and Social Psychology paper, Mischel and Ebbe Ebbesen examined how seeing or not seeing the rewards affected waiting. The finding cut against a simple willpower story: rewards sitting in front of children often made waiting harder because the children had to keep looking at what they were trying not to take.

A later paper by Mischel, Ebbesen and Antonette Raskoff Zeiss, published in 1972 in the same journal, pushed further into the cognitive strategies children used. Thinking about the reward in a “hot” way, as something delicious and immediate, tended to make waiting harder. Turning attention elsewhere, or thinking about the treat in a cooler, less appetizing way, helped some children delay longer.

That is already a different lesson from the popular version. The test did not reveal a fixed amount of virtue inside a child. It showed that waiting was shaped by the situation, by attention, by strategy and by how the child represented the reward in mind.

The follow-ups made the story famous

The larger public myth came from follow-up work. In a 1988 Journal of Personality and Social Psychology paper, Mischel, Yuichi Shoda and Philip Peake reported that preschoolers who had waited longer were later described by parents as more competent adolescents. Those ratings included qualities such as coping with frustration, handling stress, concentrating and planning.

Then came the SAT result. In a 1990 Developmental Psychology paper, Shoda, Mischel and Peake reported that delay time predicted adolescent cognitive and self-regulatory competencies under certain diagnostic conditions. The paper included SAT scores and became part of the famous account: wait longer at age 4, score higher years later.

But those early follow-ups were not large national samples. They drew from children at Stanford’s Bing Nursery School, a selective and relatively advantaged setting. The findings were correlational. They showed that waiting time went along with later outcomes in that group, not that waiting caused those outcomes, and not that the test could separate a child’s inner self-control from family context, trust, language, cognition or classroom experience.

The famous story was too clean

The marshmallow test became shorthand for a moral lesson: delay gratification and success follows. That version removed too much. It treated one laboratory episode as if it were a stable trait, and it treated the child as if the situation did not matter.

Even Mischel’s own work does not fit that flattening. The original experiments showed that children could wait longer or shorter depending on reward visibility and mental framing. Later research also asked whether children believed the adult would keep a promise. In a 2013 Cognition study by Celeste Kidd, Holly Palmeri and Richard Aslin, children waited longer after experiencing a reliable adult than after an unreliable one. That does not make the original work wrong. It shows that waiting can be a reasonable response to trust, not just a measure of self-control.

This matters because many children do not live in environments where waiting has always paid off. If promised rewards are uncertain, taking the smaller reward now may be practical rather than impulsive. A laboratory task can capture behavior without capturing the full history that makes the behavior sensible.

The bigger replication changed the weight of the claim

The clearest challenge came from Tyler Watts, Greg Duncan and Haonan Quan. In their 2018 Psychological Science conceptual replication, they used a much larger and more diverse sample than the Stanford follow-ups and examined links between delay of gratification around age 4 and outcomes at age 15.

The result was not that waiting meant nothing. The authors found that an additional minute of waiting predicted a small gain in achievement. But the bivariate association was about half the size of the original reports, and it shrank by about two thirds after controls for family background, early cognitive ability and home environment. Associations with behavioral outcomes were much smaller and rarely statistically significant.

That is the sober version of the replication. It did not turn the marshmallow test into nonsense. It made the simple destiny story much harder to defend. Waiting time still carried information, but much of that information overlapped with things that were already true about the child’s early environment.

Adult outcomes made the limit clearer

The later adult-outcomes work sharpened the same point. In a 2024 Child Development paper, Jessica Sperber, Deborah Lowe Vandell, Duncan and Watts followed the question into adulthood and concluded that marshmallow-test performance did not reliably predict adult functioning.

That conclusion is important because the popular story was not really about SAT scores. It was about adulthood: wealth, health, achievement, character, success. The newer work suggests that the test is far weaker for those claims than the cultural memory of the experiment implies.

The more accurate lesson is narrower. Delay of gratification in a preschool task can be associated with later outcomes, especially achievement, but those associations are smaller in broader samples and heavily entangled with children’s early cognitive skills, home environments and family circumstances.

What remains useful

The marshmallow test still matters as a study of self-regulation under temptation. The original experiments showed that attention and framing can change behavior. Children who looked away, distracted themselves or transformed the reward mentally often waited longer. That is a real psychological finding, and it helped make self-control less mysterious.

What should be retired is the notion that a preschooler alone with a treat reveals a life trajectory. A child waiting at a table is not only demonstrating self-command. The child is also navigating trust, instructions, hunger, temperament, past experience, language, class context and the immediate design of the task.

The story is therefore not that the marshmallow test was debunked and should be forgotten. It is that a good experiment became a bad folk theorem. The science was about how situations and strategies shape delay. The myth was that one moment of waiting could rank children by future success.