In 2014, three researchers at the University of California, Davis put adults inside an fMRI scanner and asked them trivia questions. The trick was that the researchers had asked each person earlier which questions they found interesting and which they did not. Then they showed the questions, one at a time, and gave the person a few seconds to wonder before revealing the answer.
What happened in the brain during those few seconds is what the rest of this post is about. When the person was curious about the question, the dopaminergic reward circuit fired in the same circuit that fires when the brain anticipates food, sex, or money. The hippocampus, the structure that forms new long-term memories, fired alongside it. After the scan, when the researchers tested memory for the answers, the curious-question answers were remembered significantly better than the not-curious ones. Even incidental information shown during the curious moments was remembered better.
This is Matthias Gruber, Bernard Gelman, and Charan Ranganath’s 2014 paper in Neuron. The title is dry. The finding is not. Curiosity is not a personality trait. It is a brain state, briefly entered, that the dopaminergic system rewards and the memory system records into. When the curiosity state is on, learning is efficient. When the curiosity state is off, learning is what every adult who ever sat in a boring lecture remembers — possible, with effort, but slow and shallow.
The folk model treats curiosity as a fixed individual trait. The curious kid versus the not-curious kid. Some kids love to learn; others have to be made to. The Gruber finding, and the body of work that has followed it, says the folk model is wrong. Curiosity is a state. The state is engineerable. The implications for how childhood is structured run deep.
What the Gruber Experiment Actually Found
Each adult participant was shown trivia questions inside the scanner. Each question had been pre-rated by that person on a scale of how curious it made them. After the question appeared, there was a several-second wait before the answer revealed. That waiting window is where the action was.
During those waiting seconds, the fMRI captured activation in two regions simultaneously. The first was the dopaminergic midbrain, the structures the brain uses to signal that a reward is coming. The second was the hippocampus, the structure that forms new long-term memories. The two were synchronized. The brain was, in real time, preparing to encode what came next.
The post-scan memory test showed two findings. Curious-question answers were remembered better than not-curious-question answers, and the dopaminergic activation predicted the memory advantage. More striking: faces shown to participants during the curious-question windows were remembered better than faces shown during the not-curious windows — even though the faces had nothing to do with the questions. The curiosity state opened a memory-encoding window, and whatever the attention landed on during that window got recorded with the boost.
This is the part that reshapes how we ought to think about learning. Curiosity is not just a motivational signal that helps a kid pay attention to one specific thing. It is a state of the brain, lasting seconds to minutes, during which the encoding system is more efficient at recording everything the attention lands on. A teacher who produces a curious moment about volcanoes does not just teach volcanoes better. She produces a window during which whatever the student attends to gets encoded with the boost.
One caveat. The Gruber participants were adults, not children. Pediatric fMRI is harder to run, so direct evidence in children leans on behavioral studies rather than imaging. The mechanism is plausibly the same across the lifespan, and the next move in the science is to bring the test into childhood directly.
The Replication in Real Classrooms
Some of the replication has happened in classrooms. In 2024, the economists Sule Alan and Ipek Mumcu published a paper in the American Economic Review that took the curiosity mechanism and tested it on elementary-school children in Turkey.
Two cluster-randomized trials. The intervention trained teachers to shift from lecture toward inquiry — wondering aloud, asking open questions whose answers they did not know, tolerating the lesson going somewhere the plan did not predict. The kids in the engineered classrooms paid more attention, retained more, and were more willing to do work to find out things they did not yet know. The effects persisted three years later, through teacher changes and through pandemic school closures.
What had been treated as personality (this kid is curious, that one is not) turned out to be a state the environment installs or suppresses. The intervention did not select for naturally curious children. It produced curiosity, in the kids who were already there, by changing what the adults around them were doing.
What This Means for Childhood as Currently Structured
If curiosity is the learning state, then the variable that gates childhood learning is how often the curiosity state is on. Not how many hours of instruction. Not how many practice problems. Not how many enrichment activities. How often the kid is, by the brain’s measurement, in the neurological state that opens the memory window.
The dominant model of childhood, in school and at home, has been organized around content delivery. The adult presents the information; the child receives it; the child is tested on whether they received it. The Gruber-Alan mechanism inverts the model. The child’s curiosity is the gating variable. Content delivered in the absence of the curiosity state is poorly encoded, no matter how much of it gets pushed across the desk (or when or by whom). Content delivered during the curiosity state is encoded efficiently and laterally, including content the adult was not trying to teach.
This explains a great deal of what teachers and parents already observe. The kid who cannot focus during the math drill but who can build a sustained engineering project on his own. The teen who cannot remember the chemistry vocabulary but who absorbed the entire production credits list of an obscure band. The first-grader who learned to read in the bedroom after lights-out because she wanted to know what happened next in the story.
The kid is not deficient. The instructional model is misaligned with how their brain encodes information. The content was being delivered in the wrong state.
The parent’s job, per the curiosity-state research, is not to instruct or even to teach. It is to engineer the conditions under which the curiosity state is more likely to turn on.
Try This / Consider This
TRY THIS: The Wonder-Out-Loud Week
For one week, run three small moves the curiosity-state research suggests. The point is to engineer conditions, not to deliver content.
Move one: wonder out loud. When a question comes up that you do not know the answer to, do not silently move on. Say it: “Huh, I do not actually know why that happens. I am curious now.” The curiosity state is contagious. Kids who watch adults inhabit it become more likely to enter it themselves. This is the easiest of the three moves and the one that compounds fastest, because it makes wondering modeling instead of performance.
Move two: tolerate the tangent. The conversation in the car was supposed to be about Saturday’s plans. The kid took it to a question about whether birds have nightmares. The instinct is to steer back. The instinct is wrong. The bird-nightmare tangent IS the curiosity state turning on. Redirecting away from it is closing the memory-encoding window the Gruber paper measured. Let the tangent run for a few minutes. You can come back to the schedule.
Move three: follow the attention. When your kid plants themselves in front of the construction equipment at the zoo instead of moving toward the elephants (true story), what you are looking at is the curiosity state activating on the construction equipment. The elephants will be there next time. The two hours of attention on the backhoe IS the learning that happened that day.
At the end of the week, notice one thing: did you find yourself fighting your own instinct to redirect more than once? The instinct is the variable. The kid is not the variable.
CONSIDER THIS
When was the last time you, as an adult, felt your own curiosity state turn on?
What conditions produced it? A book you picked up without an agenda. A conversation that went sideways into something neither of you had planned to talk about. A walk during which you noticed something you had walked past a hundred times. The conditions matter, because if you cannot produce the state in your own life, producing it in your kid’s life will feel like work you do not want to do. Whatever in your week consistently turns the state OFF in you is also a clue about what is turning it off in your kid.
Gabriele and I write more about parenting and developmental science at our Functional Child Development Substack.
Want the research behind this post? Read the Dive Deeper: Curiosity as Brain State — five studies that anchor what we wrote, what they found, and where the evidence has limits.
Every Functional Child Development post comes with a Dive Deeper companion. Subscribe for the full picture.