Skip to main content

Back online

Concept Explainer

The Marshmallow Test Was Wrong

The famous experiment got the wrong conclusion. Discover what really predicts long-term success — it's not just willpower.

12:58 listenAudio + TranscriptUpdated Feb 2026
Listen to the Episode
0:000:00

Explore this idea

Think through the marshmallow test was wrong with AI...

Free account · 30 seconds to start

Want to explore this idea without listening first?

Want a deep dive on a different topic?

Full Transcript

# Delayed Gratification

A kid sits alone in a room at Stanford.

The year is 1968.

On the table in front of them: one marshmallow.

The deal is simple. Eat it now... or wait fifteen minutes and get two.

Except Walter Mischel—who'd spent years studying how kids in Trinidad resisted temptation differently than kids in America—just designed a test that would follow these children for forty years. A test that would spark a debate still burning today about willpower... about poverty... and about whether any of us are really in control.

Most people think they know this story.

Kid waits, kid succeeds. Kid eats, kid struggles.

Clean narrative. Nice life lesson about self-control.

Not even close.

Mischel tracked six hundred kids.

The ones who waited—who covered their eyes, who sang songs, who turned away from temptation—they did have better outcomes decades later. Higher SAT scores... by an average of two hundred and ten points. Better health. More stable relationships.

When he published in 1972, it felt like he'd found the master key. Delayed gratification as the skill that unlocks everything else.

But here's where it gets strange.

*Why* could some kids wait... and others couldn't?

What was happening in those fifteen minutes that echoed for decades?

Fast forward to 2004.

Samuel McClure and his team at Princeton get people into an fMRI machine. They offer choices: twenty dollars now... or a hundred in a month.

And they watch what lights up.

When someone reaches for the immediate reward, it's the ventral striatum firing. Ancient. Emotional. The same circuit that lights up when rats get cocaine.

When they choose to wait? The prefrontal cortex kicks in. That's the part that does long-term planning. That imagines futures. That can say "not yet."

Two systems.

One brain.

Constant negotiation.

And here's what nobody tells you: the limbic system is *faster*.

It gets first dibs on every decision. The prefrontal cortex has to catch up... like a parent chasing a toddler who's already grabbed the candy.

By the time your rational brain shows up, your emotional brain has already made the call.

Daniel Kahneman called them System One and System Two. Fast and slow.

Delayed gratification is basically System Two trying to talk System One off the ledge.

Every. Single. Time.

Which sounds exhausting.

And it is.

Roy Baumeister figured this out in 1994 with his radish experiment.

He brought hungry students into a room filled with the smell of fresh-baked cookies. Half got to eat the cookies. The other half? They had to eat radishes... while staring at those cookies.

Then both groups tried to solve impossible puzzles.

The radish-eaters gave up twice as fast.

They'd burned through their self-control budget resisting cookies... and had nothing left for math.

He called it ego depletion. Willpower works like a muscle—you can strengthen it, but you can also wear it out. Make a bunch of hard decisions in a row and your ability to resist that next temptation drops.

Not moral failure.

Depletion.

The story gets messier, though.

Frequently asked questions

Was the marshmallow test debunked?
Not entirely. A 2018 replication at NYU and UC Irvine with a bigger sample found the predictive power dropped by two-thirds once family income and home environment were controlled for. The effect is real but much smaller than advertised.
Why would a kid rationally eat the marshmallow right away?
If adults have broken promises before, or the electricity gets shut off without warning, waiting is a bad bet. Celeste Kidd showed this at Rochester: kids who'd just experienced a broken promise waited about 3 minutes, while kids who saw promises kept waited 12.
How did the kids who waited actually manage it?
They didn't white-knuckle it — they got strategic. One pretended the marshmallow was a cloud, another sang the Sesame Street theme, another covered it with a cup and forgot it. Mischel called these "cool strategies," and teaching them pushed wait times from six minutes to fourteen.
What happens in the brain during a now-versus-later choice?
In Samuel McClure's 2004 fMRI work, immediate rewards lit up the ventral striatum — the same circuit that fires when rats get cocaine — while choosing to wait engaged the prefrontal cortex. The limbic system is faster, so it gets first dibs on every decision.
If willpower runs out, what should I do instead?
Design systems so the delayed choice is the default, as Harvard's David Laibson argues: opt-out retirement savings, round-up saving apps, commitment devices. And turn waiting into a game — hit a specific savings number or track how long you can go without your phone.
MindCast — Audio learning that helps you think clearly