What the Marshmallow Test Really Teaches About Self-Control

One of the most influential modern psychologists, Walter Mischel, addresses misconceptions about his study, and discusses how both adults and kids can master willpower.

Charles Dharapak / AP

The image is iconic: A little kid sits at a table, his face contorted in concentration, staring down a marshmallow. Over the last 50 years, the “Marshmallow Test” has become synonymous with temptation, willpower, and grit. Walter Mischel’s work permeates popular culture. There are “Don’t Eat the Marshmallow!” t-shirts and Sesame Street episodes where Cookie Monster learns delayed gratification so he can join the Cookie Connoisseurs Club. Investment companies have used the Marshmallow Test to encourage retirement planning. And when I mentioned to friends that I was interviewing the Marshmallow Man about his new book, The Marshmallow Test: Mastering Self-Control, nobody missed the reference.

It began in the early 1960s at Stanford University’s Bing Nursery School, where Mischel and his graduate students gave children the choice between one reward (like a marshmallow, pretzel, or mint) they could eat immediately, and a larger reward (two marshmallows) for which they would have to wait alone, for up to 20 minutes. Years later, Mischel and his team followed up with the Bing preschoolers and found that children who had waited for the second marshmallow generally fared better in life. For example, studies showed that a child’s ability to delay eating the first treat predicted higher SAT scores and a lower body mass index (BMI) 30 years after their initial Marshmallow Test. Researchers discovered that parents of “high delayers” even reported that they were more competent than “instant gratifiers”—without ever knowing whether their child had gobbled the first marshmallow.

But there’s been criticism of Mischel’s findings too—that his samples are too small or homogenous to support sweeping scientific conclusions and that the Marshmallow Test actually measures trust in authority, not what he says his grandmother called sitzfleisch, the ability to sit in a seat and reach a goal, despite obstacles. I met with Mischel in his Upper West Side home, where we discussed what the Marshmallow Test really captures, how schools can use his work to help problem students, why men like Tiger Woods and President Bill Clinton may have suffered “willpower fatigue”—and whether I should be concerned that my five-year old devoured “the marshmallow” (in his case, a small chocolate cupcake) in 30 seconds.

Jacoba Urist: I have to tell you right off, my son is in kindergarten and he flunked the Marshmallow Test last night.

Walter Mischel: First, it’s important that I say “the test” in quotes, because it didn’t start out as a “test” but a situation where we were studying the kinds of things that kids did naturally to make self-control easier or harder for them. Four-year-olds can be brilliantly imaginative about distracting themselves, turning their toes into piano keyboards, singing little songs, exploring their nasal orifices.

Urist: The problem is, I think he has no motivation for food. In our house, dessert isn’t a big deal. Could the kids who wait for the marshmallow just not care that much about treats? Maybe their families didn’t use food as a reward system so they didn’t respond to it as a motivator?

Mischel: You have to understand, in the studies we did, the marshmallows are not the ones presented in the media and on YouTube or on the cover of my book. They were these teeny, weeny pathetic miniature marshmallows or the difference between one tiny, little pretzel stick and two little pretzel sticks, less than an inch tall. It’s really not about candy. Many of the kids would bag their little treats to say, “Look what I did and how proud mom is going to be.” The studies are about achievement situations and what influences a child to reach his or her choice. In some cases, we even used two colored poker chips versus one.

Urist: How important is trust then? Some critics claim that a 2012 University of Rochester study calls the Marshmallow Test into question. Children in a reliable environment (where they could trust that the delayed reward would materialize) waited four times longer than children in the unreliable group. Were the kids in your test simply making a rational choice and assessing reliability? And wouldn’t that factor be outside the scope of the original Marshmallow Tests?

Mischel: This is another thing the media regularly misses. Before the marshmallow experiments, I researched trust in decision-making for adults and children. Trust is a tremendous issue. Therefore, in the Marshmallow Tests, the first thing we do is make sure the researcher is someone who is extremely familiar to the child and plays with them in the playroom before the test. It’s also important to realize, it’s not a matter of if somebody will come back with the two little marshmallows. They are all right there on the tray. It’s all out in the open, so there’s no trust issue about whether the marshmallows are real.

Urist: When it comes to correlations between the Marshmallow Test and indicators of success later in life, some people say the marshmallow tests are based on too small a sample to draw meaningful conclusions, that you originally studied over 500 children, but you only tracked down 94 of the participants’ SAT scores?

Mischel: We didn’t want parental reports of SAT scores. We actually wanted to be able to contact the organization that administered the SAT at the time and therefore had to use a subset of the children. But the correlations were sufficiently strong that the smaller sample size isn’t relevant. To me, the real problem was that we were dealing with an incredibly homogenous sample, either children of Stanford faculty or Stanford graduate students—and we still saw strong correlation. But it was an unbelievably elitist subset of the human race, which was one of the concerns that motivated me to study children in the South Bronx—kids in high-stress, poverty conditions—and yet we saw many of the same phenomena as the marshmallow studies were revealing.

Urist: Are some children who delay responding to authority? Could waiting be a sign of wanting to please an adult and not a proxy for innate willpower? Presumably, even little kids can glean what the researchers want from them.

Mischel: Maybe. They might be responding to anything under the sun. But it’s how they respond. The most interesting thing, I think, about the studies is not the correlations that the press picks up, but that the marshmallow studies became the basis for testing all kinds of adults and how adults deal with difficult emotions that are very hard to distance yourself from, like heartbreak or grief.

[Ed. note: Mischel’s book draws on the marshmallow studies to explore how adults can master the same cognitive skills that kids use to distract themselves from the treat, when they encounter challenges in everyday life, from quitting smoking to overcoming a difficult breakup.]

Urist: I have to ask you about President Clinton and Tiger Woods, both mentioned in the book. I’ve heard of “decision fatigue”—are their respective media scandals both examples of adults who suffered from “willpower fatigue?” Men who could exercise enormous self-discipline on the golf course or in the Oval office but less so personally?

Mischel: No question. People experience willpower fatigue and plain old fatigue and exhaustion. What we do when we get tired is heavily influenced by the self-standards we develop and that in turn is strongly influenced by the models we have. Bill Clinton simply may have a different sense of entitlement: I worked hard all day, now I’m entitled to X, Y, or Z. Confusion about these kinds of behaviors [tremendous willpower in one situation, but not another] is erased when you realize self-control involves cognitive skills. You can have the skills and not use them. If your kid waits for the marshmallow, [then you know] she is able to do it. But if she doesn’t, you don’t know why. She may have decided she doesn’t want to.

Urist: So for adults and kids, self-control or the ability to delay gratification is like a muscle? You can choose to flex it or not?

Mischel: Yes, absolutely. That’s a perfectly reasonable analogy.

Urist: In the book, you advise parents if their child doesn’t pass the Marshmallow Test, ask them why they didn’t wait. What should I be trying to elicit from my son about why he grabbed the first little cupcake? When I asked, he just shrugged and said, “I don’t know.”

Mischel: It sounds like your son is very comfortable with cupcakes and not having any cupcake panics and I wish him a hearty appetite. Whether the information is relevant in a school setting depends on how the child is doing in the classroom. If he or she is doing well, who cares? But if the child is distracted or has problems regulating his own negative emotions, is constantly getting into trouble with others, and spoiling things for classmates, what you can take from my work and my book, is to use all the strategies I discuss—namely making “if-then” plans and practicing them. Having a whole set of procedures in place can help a child regulate what he is feeling or doing more carefully.

Urist: One last question. After all these years, why a book now?

Mischel: Well, there are two reasons. First, so much research has exploded on executive function and there have been so many breakthroughs in neuroscience on how the brain works to make it harder or easier to exercise self-control. It’s an enormously exciting time within science for understanding in a much deeper way the relationships between mind, brain, and behavior and to ask the important questions: How can you regulate yourself and control yourself in ways that make your life better? Second, there have been so many misunderstandings about what the Marshmallow Test does and doesn’t do, what the lessons are to take from it, that I thought I might as well write about this rather than have arguments in the newspapers.