What if a child took both marshmallows home? Roasting the Marshmallow Test.

By Anna Roosen and Catherine Burrows

Copyright illustration, Anna Roosen 2022

Picture this: A group of young children take part in an experiment with permission from their parents. The instructions are simple. Each child is left alone in a room with a treat, a marshmallow. They have two choices: they may either enjoy the marshmallow as soon as they like or wait fifteen minutes and get two marshmallows.

One child seems to find the wait easy. They receive the second marshmallow and put both into their pocket to take home. The researchers are astounded. They can’t help but smile as the child quietly leaves the room. This can only be very good news.

Everyone who’s heard of the original ‘marshmallow test’ knows what it means: a child able to resist the urge to eat the first marshmallow will do better in life. We can only imagine the success that a child who resists the second marshmallow will achieve.

But will they really?

What was the Marshmallow Test?

Back in 1972, a Columbia University psychologist named Walter Mischel[1] set out to test ‘delayed gratification’, the ability to resist an immediate reward to gain a greater reward later.

His participants, a group of 93 pre-schoolers from Stanford University’s Bing Nursery School, were given the same two choices as our imaginary child described above. The length of time each child waited without eating the marshmallow was taken to be a definitive measure of their self-control.

The test becomes influential

The test gained interest when Mischel followed up with the participants from the study several years later and found that, on average, the participants who lasted the longest before eating the marshmallow were those who tended to have greater ‘success’ as adolescents.

The study became increasingly influential as other scientists applied it to their fields. Neuroscientists attempted to locate the ‘self-control’ structures in the brain; psychologists developed programs to increase self-control in children; and educators created and promoted new forms of education that boosted self-control. All of us have since likely been impacted in some small way by what has become one of history’s most famous studies.

But can marshmallows really indicate future success?

The ‘marshmallow test’ is still used as a metaphor today for the trait of willpower, with the implication that possessing this trait leads to success. But is this correlation correct? We think there are two problems with this study that make its conclusions questionable. The first regards the set-up and the second relates to assumptions made by the research team.

The set-up

The fact that the children were all attending the same pre-school was presumably intended to act as a control.

However, the age range was three-and-a-half years old to five-and-a-half years old. Despite the enormous differences between a toddler and a school-aged child, reports of the study don’t appear to differentiate based on age. Mischel didn’t seem to differentiate on the basis of gender either, or on social, linguistic, or cultural diversity: this certainly limits the test’s applicability to children everywhere.

We also know that some children (and adults) are more sedentary than others, some are more patient, and some are more content to follow instructions. We don’t think these traits are the same as delayed gratification.

When he went back to the children as adolescents, Mischel looked at their academic success. He found the children who had resisted eating the marshmallow had had greater academic achievement, measured by their SAT results. This seems reasonable, if limited. He also found that those who had waited for the marshmallow were better ‘socially well-adjusted’. When it came to assessing whether a child was ‘socially well-adjusted’, he based this on asking the opinion of the participants’ own parents, hardly an objective measurement.[2][3]

The assumptions

The original purpose of the experiment was to test whether delayed gratification was enhanced when the children thought about a reward. However, the results indicated the opposite: thinking about a reward makes waiting more difficult. The possible link between the ability to wait and willpower was only made later, once the participants were re-assessed as adults. [4]

The marshmallow test findings assume that self-control is a ‘fixed’ trait and that one single act of behaviour at a young age determines whether a child has that trait or not. It doesn’t allow for context or for child development.

The questions we would like to have asked

Mischel also seems to have failed to consider – or even ignored – important issues which might have influenced the children’s behaviour more than willpower.

We wondered what it might have been like to be one of those children, so we came up with a few questions we would like to have asked them:

  • How are you feeling today?
  • What did Mummy/Daddy tell you about today?
  • Do you have a quiet and comfortable place to sleep?
  • Are you hungry?
  • Do you have some place you’d rather be or something you’d rather be doing right now?[5]
  • If an adult promises you something, can you trust them to give it to you?

And, most importantly of all,

  • Do you like marshmallows?

Later studies – The importance of context

Later studies add further to the questions around the marshmallow test. One study, by Kidd, revealed that two-thirds of children, who were living at a shelter and had experienced unreliable circumstances, would wait the fifteen minutes for a second marshmallow when the adult administering the test was someone they trusted. When their peers were administered the test by an adult who had previously deceived them, half of the children ate the marshmallow soon after the adult left. This indicates self-control can be contextual, that is, a response to circumstances, rather than a fixed trait.[6]

Later studies – Willpower vs self-control

We believe that the most revealing study of all is the Dunedin Study, now run by Professor Richie Poulton at the University of Otago, New Zealand. It began in 1975-1976, and today, fifty years later, 961 out of the initial 1,027 participants are still participating in the study.

“There are three things that set the Dunedin study apart…: a high retention rate (94% of the original cohort have stayed), a multidisciplinary approach that gathers an “incredible” breadth of information, and testing and interviewing people face-to-face rather than through questionnaires.”[7]

One way the Dunedin Study is different from the marshmallow test is that it explores ‘self-control’ rather than ‘delayed gratification’. It describes self-control as thinking before speaking or acting and being able to resist temptations or jumping to conclusions. The study has found that self-control predicts success in avoiding obesity, maintaining fitness, sustaining relationships, preventing addiction, resisting spending, and saving for old age.

This is important

We think the apparent simplicity of the study and the charm of the idea is why the marshmallow test continues to be so popular and appealing. Who doesn’t want to watch cute kids trying to resist eating a marshmallow?

But imagine if, as an adult, you were deemed to have a fixed personality trait based on whether you ate or resisted a piece of food on a single occasion? Wouldn’t you object?

The marshmallow test is problematic because it makes assumptions about children and their future based on a single behaviour at a single point in time. This reinforces the belief that behaviour and character are inbuilt and so they can never be changed.

This belief is not just bad for one child, it’s bad for every child. Telling people that some children have extraordinary willpower and others don’t assumes a fixed mindset with no room for growth or change.

Telling children they are exceptional can have unexpected results. It might discourage them from trying harder or it might make them feel anxious about living up to their reputation. Telling other children they have no willpower can be dispiriting and ignores differences in their personalities, their development and their context.

The marshmallow test also reinforces the idea that some children are fundamentally less able than others and this can’t be changed. This, in turn, excuses society from looking at opportunities to improve every child’s life and from helping some children overcome a difficult start.

Right at the beginning of this article, we asked, “What would you think if a child took home both marshmallows?” The marshmallow test is so well-known in our culture that the expectation is that ‘of course’ the child who took the two marshmallows home must have exceptional willpower (and therefore potentially have exceptional success in future life). This would be the natural extension of the study’s findings.

But what if the child simply just didn’t like marshmallows?


[1] We use his name throughout this article to refer to the team who worked on this study.

[2] If you think we are being unfair, try googling, “I don’t understand my teen.”

[3] See also: ‘Think your kid’s gifted? Think again.’ https://www.wired.com/2008/08/think-your-kids/.

[4] Shoda, Yuichi; Mischel, Walter; Peake, Philip K (1990) “Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification: Identifying diagnostic conditions” Developmental Psychology. 26 (6) 978-986

[5] McNicholl, A. 2022. Discussion: Time is the most valuable commodity we have and the one we can’t get back.

[6] Kidd et al., “Rational snacking: young children’s decision-making on the marshmallow task is moderated by beliefs about environmental reliability” Cognition 126, no 1 (2013): 109-114 – PubMed (nih.gov)

[7] https://www.theguardian.com/world/2022/apr/01/the-dunedin-study-at-50-landmark-experiment-tracked-1000-people-from-birth