Here is a story you have probably heard. Put a four-year-old in a room with a marshmallow. Tell them they can eat it now, or wait fifteen minutes and get two. The ones who hold out, the story goes, grow up to be better students and better-regulated teenagers. One small test, one big destiny.
The marshmallow test is one of those findings that gets repeated so often it stops feeling like a finding and starts feeling like a law of nature. So I went looking. What I found is a much more interesting story than the parable, and a more honest one.
A note before going further: I am not a psychologist or a clinician, and nothing here is advice about your child or yourself. This is one curious reader working through the research, which is observational, with results that describe patterns across groups rather than predictions about any single person.
What the original test seemed to show
The experiments themselves were real and carefully run. Walter Mischel ran his delay-of-gratification studies at Stanford’s Bing Nursery School, and the famous version of the result came from a 1990 follow-up by Shoda, Mischel, and Peake, which reported that children who waited longer as preschoolers had higher SAT scores years later.
Those correlations were not small. The link between preschool delay time and later SAT performance was strong enough to travel far beyond the lab. A fifteen-minute test that appears to predict your teenager’s exam scores is exactly the kind of clean cause-and-effect we love to retell.
There was a quieter problem underneath. The follow-up rested on children originally drawn from Stanford’s Bing Nursery School, and the later SAT analysis was based on only 94 parent-reported scores, with some key correlations coming from smaller subgroups.
The authors themselves urged caution writing “We must emphasize the need for caution in the interpretation of the total findings” and “given the smallness of the sample, the obtained coefficients could very well exaggerate the magnitude of the true association”. The culture around the study did not.
What the 2018 replication actually found
In 2018, Tyler Watts, Greg Duncan, and Haonan Quan published a conceptual replication in Psychological Science using a far larger and more diverse sample, roughly 900 children rather than fewer than a hundred in the original SAT follow-up. The team’s results, Watts told reporters, show that “once background characteristics of the child and their environment are taken into account, differences in the ability to delay gratification do not necessarily translate into meaningful differences later in life.”
The result is the part that should make anyone pause. The raw link between waiting at age four and achievement at fifteen turned out to be about half the size reported in the original work. Once the researchers added controls for family background, early cognitive ability, and the home environment, the link shrank by roughly two-thirds. Most of what looked like willpower predicting success may have been background circumstances predicting both.
This is one replication, not a closed case, and worth holding as a strong clue rather than a verdict.
Why the distinction matters, and what waiting really reflects
Confidence and correctness feel identical from the inside, and the marshmallow test is a fair example of the two coming apart. For decades the finding was confidently held and widely repeated. The 2018 paper is the moment that confidence met a larger, more diverse sample.
Perhaps, the more interesting question is what the waiting was measuring all along. The sociologist Jessica McCrory Calarco, writing in The Atlantic, offered an explanation that reframes the whole thing. A child’s choice, she argued, depends on trust: “A second marshmallow seems irrelevant when a child has reason to believe that the first one might vanish.”
For a child whose world is unpredictable, grabbing the treat now is not a failure of self-control. It is a rational read of the room. As Calarco put it, for some children “daily life holds fewer guarantees: There might be food in the pantry today, but there might not be tomorrow, so there is a risk that comes with waiting.” This is her interpretation rather than a proven mechanism, but it’s something to keep in mind. The kid eating the marshmallow may not lack patience. They may have learned, correctly, that promises do not always hold.
What to take from this as a parent or a person
None of this means self-control is fake or that teaching kids to wait is pointless. It probably means self-control is less like eye color and more like a plant: it grows or it doesn’t depending heavily on the soil. A child who can count on the people around them learns that waiting pays off, because for them it usually has.
So the marshmallow works less like a verdict and more like a thermometer pointed at a child’s circumstances, dressed up for years as a test of their character. Self-control is real, and the soil it grows in shapes how much a child has to draw on.
If you are reading this while worrying about your own child, or your own past, a good clinician or counsellor will be far more useful than any single study or essay, including this one.
