Marshmallow test of self control: An African grey parrot performs as well as preschoolers.

A Parrot Passes the Marshmallow Test

A Parrot Passes the Marshmallow Test

Wild Things
Slate’s animal blog.
Sept. 9 2014 9:00 AM

A Parrot Passes the Marshmallow Test

This African grey parrot has more patience than you.

Photo by Ronaldo Schemidt/AFP/Getty Images

Can your kid pass the “marshmallow test”? And what does it mean if he can’t, but a parrot can?

The marshmallow test is pretty simple: Give a child a treat, such as a marshmallow, and promise that if he doesn’t eat it right away, he’ll soon be rewarded with a second one. The experiment was devised by Stanford psychologist Walter Mischel in the late 1960s as a measure of self-control. When he later checked back in with kids he had tested as preschoolers, those who had been able to wait for the second treat appeared to be doing better in life. They tended to have fewer behavioral or drug-abuse problems, for example, than those who had given in to temptation.


Most attempts to perform this experiment on animals haven’t worked out so well. Many animals haven’t been willing to wait at all. Dogs, primates, and some birds have done a bit better, managing to wait at least a couple of minutes before eating the first treat. The best any animal has managed has been 10 minutes—a record set earlier this year by a couple of crows.

The African grey parrot is a species known for its intelligence. Animal psychologist Irene Pepperberg, now at Harvard, spent 30 years studying one of these parrots, Alex, and showed that the bird had an extraordinary vocabulary and capacity for learning. Alex even learned to add numerals before his death in 2007. Could an African grey pass the marshmallow test?

Adrienne E. Koepke of Hunter College and Suzanne L. Gray of Harvard University tried the experiment on Pepperberg’s current star African grey, a 19-year-old named Griffin. In their test, a researcher took two treats, one of which Griffin liked slightly better, and put them into cups. Then she placed the cup with the less preferred food in front of Griffin and told him, “wait.” She took the other cup and either stood a few feet away or left the room. After a random amount of time, from 10 seconds to 15 minutes, she would return. If the food was still in the cup, Griffin got the nut he was waiting for. Koepke and colleagues presented their findings last month at the Animal Behavior Society meeting at Princeton.

At every time period tested, Griffin successfully waited at least 80 percent of the time, even at the maximum 15 minutes. That’s the best performance ever seen in an animal, comparable to Mischel’s original results with preschoolers.


Human children faced with the test use a variety of strategies to distract themselves from the tempting first marshmallow. Similarly, the bird didn’t always just sit in peaceful anticipation of receiving his treat. Sometimes he tossed the lesser food away or gave it a taste, or he distracted himself with preening. (Koepke and colleagues produced this video directly comparing Griffin to children put through the test.)

What does it mean? Well, we already know that Griffin is a pretty smart bird. He has a large vocabulary, knows the names of dozens of objects, and can recognize colors, shapes, and numbers. Pepperberg has compared his intelligence that of a five- or six-year-old child. Perhaps it shouldn’t be all that surprising that he could master the marshmallow test.

However, Gray noted, “I don’t believe Griffin was unique.” Members of other bird species would probably be able to accomplish the same thing, given the right circumstances, she said.

Goffin’s cockatoos were put to the test by researchers at the University of Vienna last year and could manage only a bit over a minute at best. But these birds had to hold the reward in their beaks, rather than look at it on the table. “It’s a little unfair,” Koepke said. How many of us would be able to resist eating something tasty if it were put in our mouths? If these birds were tested in the setup used for Griffin, they, too, may have succeeded.

But the marshmallow test isn’t necessarily one about smarts. A couple of years ago, researchers discovered that trust was a key factor. When the experiment was altered so that kids had no reason to trust the experimenter telling them to wait, they often didn’t bother. Griffin, Gray noted, lives in a trustworthy environment—he has no reason to doubt that the promise of a treat will be fulfilled.

But wouldn’t it be interesting to watch what would happen, Gray speculated, if he were confronted with an experimenter he didn’t trust? Perhaps he’d be just as unwilling to wait as anyone else.