The marshmallow test: Delayed gratification isn't just a matter of willpower

© 2018 Gwen Dewar, Ph.D., all rights reserved

Sometimes the smart thing is to reject an immediate reward in order to wait for something better. But this isn’t always the case, and delayed gratification isn’t always a matter of willpower.

Studies show that children’s choices depend a lot on our own behavior. When adults appear unreliable – or downright untrustworthy – kids choose instant rewards over future benefits. Here are the details.


If you’ve read about self-control and delayed gratification in children, you’ve probably heard of the marshmallow test. Sit a child down at a table, offer him a marshmallow, and make the following promise:

“You can eat this now if you want, but if you wait 15 minutes until I come back, and I see you haven’t eaten it, I will give you another one. You’ll end up with two marshmallows."

What do kids do? Some show great powers of delayed gratification, not touching that marshmallow for the entire 15 minutes. Others give in to temptation after only a few minutes.

And it seems to matter. When researchers have followed up on the preschoolers who’d participated in the first marshmallow experiments of the 1970s, they have found that a child’s performance on the test was a predictor of many later outcomes.

Kids who’d waited the longest went on to score higher on scholastic achievement tests. They were also more likely to finish college and end up with lower body mass indices, or BMIs.

So the marshmallow test has gotten a lot of attention as a measure of self-control and a predictor of life success.

But is it really? Can we assume that kids who do poorly on the marshmallow test – and real-world equivalents of the marshmallow test– are suffering from a special deficit of self-control? Or is it possible that these seemingly “impulsive" kids are responding to the cues around them and making smart choices?

Some kids have learned hard lessons about the world. The adults they know don’t keep promises, and nobody seems to enforce fairness. When these kids get something nice, they know that somebody bigger may come along and take it away.

That’s what struck Celeste Kidd in 2012, when she was a student earning her Ph.D. in Brain and Cognitive Sciences at the University of Rochester.

She was watching children at a homeless shelter-- children who lived in a dog-eat-dog environment, where theft was common, and adults rarely intervened.

How would these kids behave in a marshmallow test? The answer seemed clear. "All of these kids would eat the marshmallow right away."

So she designed a clever new version of the marshmallow experiment, and got some astonishing results. If you manipulate a child’s trust in the adult, you radically change his or her performance on the marshmallow test (Kidd et al 2013).

Learning when to wait

The experiment worked this way:

Before introducing any marshmallows, Kidd and her colleagues had preschoolers work on an art project. Each child was seated at a table in an “art project room" where there was a tightly sealed jar of used crayons. A friendly adult told the child she could use those crayons now, or she could wait until the adult returned with some nicer, brand-new crayons.

And then one of two things happened:

  • In the reliable condition, the adult returned after a couple of minutes with the new crayons.
  • In the unreliable condition, the adult came back empty-handed and apologized. “I’m sorry, but I made a mistake. We don’t have any other art supplies after all…"

This was repeated a second time with a promise of fancy stickers. Again, some kids were rewarded for waiting. Other kids waited only to get an apology that the stickers couldn’t be found.

After this warm-up, the kids were finally offered the marshmallow and given the choice. Eat one now, or wait and get two later.

And the results were remarkable.

Children in the reliable condition – who had previously received the promised rewards – waited four times as long their counterparts did.

Moreover, kids in the reliable condition were more likely to wait the full 15 minutes. Nine of the 14 children in the reliable condition waited the full 15 minutes, but only 1 of the 14 kids in the unreliable condition did so.

As coauthor Richard Aslin notes, these are dramatic differences for an experiment of this kind. Usually when researchers report they’ve found an effect, the effect is statistically significant, but rather small. Here we have a quite a big difference – and one resulting from a brief intervention.

What must things be like for children who are exposed to unreliable conditions day after day? At home or elsewhere?

As Kidd and her colleagues noted, children must be experiencing radically different views of the world depending on their home life. A child living with parents who “reliably promise and deliver small motivational treats" is going to have reason to wait for her marshmallow. But for a child “accustomed to stolen possessions and broken promises, the only guaranteed treats are the ones you’ve already swallowed."

But it doesn’t end there.

Kidd’s experiment shows us that children adjust their strategies based on their direct experiences with adults. What about indirect experiences? Might children learn by observing how adults treat other people?

An experiment in dishonesty

Maybe kids don’t have to wait for an adult to let them down personally. To lose faith – and give up on long-term rewards – maybe it’s enough to catch the adult lying to someone else.

That was the guiding hypothesis of Laura Michaelson and Yuko Manakata. So they conducted their own marshmallow experiment on preschoolers in Colorado, this time  replacing promises of art supplies and stickers with an opportunity to observe an adult behaving dishonestly towards another person (Michaelson and Manakata 2016). 

It went like this.

Each participating preschooler began the experiment with a friendly adult – an artist – seated at a table with some modeling clay. The two of them created clay sculptures together while a second adult watched with interest.

Then, when the artist had completed a sculpture of a bird, she left the room for a minute. And what happened next varied by group assignment.

  • Kids randomly assigned to the trustworthy condition saw the adult observer accidentally damage the artist’s sculpture. When the artist returned and asked for an explanation, the observer confessed and apologized.
  • Kids randomly assigned to the untrustworthy condition saw the adult observer break the sculpture on purpose. Then, when the artist returned, the observer lied to the artist, saying “No, I didn’t break your bird. I don’t know how it got broken."

Thus, half the children in this experiment witnessed an adult misbehave and lie to another person. Would these observations have an impact on their willingness to delay gratification?

To answer this question, the researchers had the adult observer administer the marshmallow test. The adult observer gave kids the standard choice: Eat one marshmallow now, or wait and receive two marshmallows later. And children's responses depended on what they had seen the adult do earlier.

Children who’d previously observed the adult behaving honestly were much more inclined to delay gratification. They waited three times longer than the kids who’d seen the adult misbehave and tell a lie.

So preschoolers don't merely remember and respond to our broken promises. They are also capable of observing our bad behavior toward third parties and inferring, this person can’t be trusted. I’d better cut my losses, and go for whatever immediate rewards I can secure right now.

To be sure, there are other factors. It isn't just our personal behavior that influences a child's willingness to wait.

Delayed gratification also appears to depend on the development of brain structures in the frontal cortex -- structures that help us weigh benefits, predict outcomes, and override our impulses (Achterberg et al 2016).

And research in China suggests that kids vary in their willingness to wait as a function of their general outlook on humanity: Kids who express more trust toward people overall tend to wait longer in delayed gratification tests (Ma et al 2018).

But the implications are clear. Delayed gratification is only partly a question of willpower. It's also heavily dependent on a child's environment, and we adults play a crucial role in shaping that environment.

More reading

We can reinforce delayed gratification by behaving in ways that are reliable and trustworthy. What else can we do to help children develop self-control? See these evidence based tips.

And we should keep in mind that adult behavior influences more than whether or not a child goes for immediate rewards. Studies also show adults influence whether or not children tell lies. Read more about that here.


References: Delayed gratification and the marshmallow test

Achterberg M, Peper JS, van Duijvenvoorde AC, Mandl RC, Crone EA. 2016. Frontostriatal White Matter Integrity Predicts Development of Delay of Gratification: A Longitudinal Study. J Neurosci. 36(6):1954-61.

Kidd C, Palmeri H, Aslin RN. 2013. Rational snacking: young children's decision-making on the marshmallow task is moderated by beliefs about environmental reliability. Cognition. 126(1):109-14.

Ma F, Chen B, Xu F, Lee K, Heyman GD. 2018. Generalized trust predicts young children's willingness to delay gratification. J Exp Child Psychol. 169:118-125.

Michaelson LE and Munakata Y. 2016. .Trust matters: Seeing how an adult treats another person influences preschoolers' willingness to delay gratification. Dev Sci. 19(6):1011-1019.

Portions of this article appeared in a previous publication, "Kids fail the marshmallow test when adults are unreliable," written by the same author for BabyCenter in 2012.

title image of waiting toddler by Eduardo Merille /flickr

content last modified 9.18