Writing

What we built · 2 October 2026

Curiosity Is Not a Trait. It Is a Condition You Can Build.

On what actually produces curious, creative people — and what we built in Cameroon because of it

Mbifanse M.K. Vuchi — Founder, Uweza

Walk into a classroom in Bamenda, or Buea, or anywhere else, and ask a teacher which of the children are curious. The answer comes quickly, and with confidence. That one. Not that one. The curious ones are known, the way the tall ones are known.

Then watch the same children outside the classroom. The quiet one who asked nothing all morning is taking apart a radio. The one who could not sit still for the lesson on percentages is arguing about why the price of a bag of cement moved. The curiosity did not arrive with the child. It arrived with the conditions — and left when they did.

This is the claim I want to make, and then show you what we built on it:

Curiosity and creativity are not properties people have. They are states that conditions produce. Change the conditions and you change the output, in the same person, within minutes. Which means they can be engineered — and that a school which treats them as inborn is not measuring talent. It is measuring whose conditions were already good.

I. The distinction that decides everything

There is a version of this argument that is wrong, and it is worth killing early, because it sounds like mine and leads somewhere else entirely.

The wrong version says: curiosity is driven by external incentives, so pay for it. Offer prizes, points, badges, places on a leaderboard, and curiosity will follow.

Fifty years of research says the opposite. When people are offered a reward for doing something they already found interesting, their interest in it tends to fall once the reward is removed — the overjustification effect, replicated for decades since the early work of Deci and of Lepper and colleagues in the 1970s. Teresa Amabile's experiments on creativity found the same shape: people working for a contracted reward produced work judged less creative than people doing the same task without one. The reward does not add to the motive. It replaces it, and it is a weaker motive than the one it displaced.

So I am not arguing for incentives. I am arguing for something that is easy to confuse with them and is in fact their opposite.

The external factors that produce curiosity are not rewards. They are conditions. What is placed in front of a person. What they already know when they meet it. Whether their not-knowing is safe to admit. Whether anyone will ever see what they make. Whether the question is theirs or was handed to them already answered.

Rewards act on the person and crowd out their motive. Conditions act on the situation and create the motive. The research tradition that matters here is not the one about what to pay people. It is the one about what to put in front of them.

II. What the evidence actually supports

Curiosity is manufactured by a gap, not by a temperament

In 1994, the economist George Loewenstein published what is still the most useful account of curiosity we have. His argument was that earlier theories had failed in a specific way: they could not say why people voluntarily seek out information that gives them no material benefit, and they had almost nothing to say about the situational determinants of curiosity — the circumstances that produce it.

His own account: curiosity is a cognitively induced deprivation that arises from the perception of a gap in your knowledge. It behaves less like a personality trait and more like an itch. You feel it when attention becomes focused on something you do not know but could.

The practical consequences are sharp, and three of them matter enormously for teaching:

A gap needs a floor. Curiosity requires prior knowledge. Complete ignorance produces nothing, because you cannot perceive a gap in a field where you know nothing at all. This is why "that student just isn't curious about physics" is so often a misreading: there is no gap because there is no floor, and no floor because nobody built one.

A gap needs a lid. It must seem closeable. A gap too large reads as impossible and the mind lets it go. There is a size at which the gap bites: big enough to open, small enough to seem worth opening.

A gap is destroyed by an answer given too early. It is the question that produces the state. Being told the answer quells curiosity before it starts. Every lesson that opens with the conclusion has spent its own best asset in the first thirty seconds.

The state is physically real, and it improves learning of everything nearby

This is not merely a nice framing. In 2014, Gruber, Gelman and Ranganath had people rate their curiosity about trivia questions while in a scanner. States of high curiosity produced increased activity in the dopaminergic midbrain and the hippocampus, and stronger interaction between them.

The result that should change how anyone designs a lesson is the second one: while in a state of high curiosity, people also learned better the things they were not curious about. Unrelated faces, shown between the questions, were remembered better when the person happened to be curious at that moment.

Curiosity is not just a motive to pursue one fact. It is a state of the brain in which learning in general goes better — and it is a state that the design of a lesson can switch on or fail to switch on. The dull material in your curriculum is not doomed. It is waiting for better company.

Creativity is produced by a social environment, not only by a person

Teresa Amabile's componential theory holds that creative output comes from three things inside the person — domain skills, creativity-relevant processes, and task motivation — and a fourth component outside: the social environment. The environment can act on all three, but its most immediate effect is on motivation.

Her laboratory work made the mechanism concrete. Expected evaluation, surveillance, competition for prizes, constrained choice: each reliably depressed the judged creativity of work produced under it. Not because the people were less able that day. Because the situation had changed what they were doing it for.

And the later meta-analytic work (Byron and Khazanchi, 2012) adds the nuance that saves this from being a counsel of despair: external factors are not uniformly poison. Rewards tied to performance quality — given informationally, in an environment that already supports autonomy — can raise creative output. It is control that kills, not contact with the world. Feedback that tells you something is help; feedback that tells you you are being watched is not.

Not-knowing has to be survivable

One more condition, from a different literature. Amy Edmondson's work on psychological safety found that teams differ in whether members believe they can admit ignorance or error without being thought less of — and that this belief, more than ability, predicts whether they ask questions, report problems and learn.

Curiosity, remember, is a form of admitted deprivation. It begins with the public acknowledgement that there is something you do not know. In a room where not knowing is dangerous, the rational move is to conceal the gap. You have not produced an incurious child. You have produced a child doing the sensible thing.

III. What this means a school is actually for

Put the four findings together and a school's job changes.

If curiosity is a state produced by gaps, and creativity is an output shaped by the social environment, then a school is not a place that finds the curious children and advances them. A school is a machine for manufacturing the conditions — and its results tell you about the machine at least as much as about the children.

Most schools, including the ones many of us were proud to attend, are built almost exactly backwards:

  • They begin with the answer, which destroys the gap before it forms.
  • They sort by speed of recall, which rewards those who already have a floor, built elsewhere, usually at home.
  • They evaluate constantly and publicly, which is the precise condition Amabile found most reliably suppresses creative work.
  • They make not-knowing dangerous, which teaches concealment.
  • They score everything, substituting a weak extrinsic motive for the strong one it displaces.
  • They end with the exam, so nothing a student makes outlives the term, and no one outside the room will ever see it.

Then, on the far side of all that, they say: some children simply are not curious.

IV. What we built, and why each piece is there

Uweza is the attempt to build the opposite machine. Every mechanism below exists because of one of the findings above. I am describing what runs, not what is planned.

1. The question comes first, and the answer is withheld

A scholar on Uweza does not begin with a subject. They begin with a question — "Why is Africa poor?" is the first — and work across several subjects at once because the question needs them.

The design rule is Loewenstein's: open the gap, do not fill it. Our first economics lesson does not open by defining GDP. It opens with a fact: in April 2014, Nigeria's economy was measured at 270 billion dollars one day and over 500 billion the next. Nothing in the country had changed overnight. Not one road, not one farm, not one hour of anybody's work.

That is a gap with a floor (every scholar knows what money is) and a lid (the answer is reachable in one lesson) and no answer offered. The lesson then earns its way to the definition of GDP, which arrives as a resolution rather than as a premise.

2. A lesson cannot be finished without crossing subjects

To complete a lesson, a scholar must read with Amara from two different subjects that meet on the same question. Not two chapters of one book — the system counts distinct subjects and refuses two economists.

The economics reading explains how national income is counted. The history reading explains whose ledger was being used, and why. Each reading names the point where it meets the question, and Amara's final station makes the scholar carry the idea across.

This is a deliberate manufacture of the most productive gap there is: the one that opens when two accounts of the same thing do not fit together. A single subject can be learned without ever noticing a gap. Two subjects, forced to meet, produce one reliably.

3. Reading is guided, enforced, and never surrendered

Amara builds a path through each text: four to seven passages, each quoted word for word from the text itself — the quotes are checked against the source before the path is stored, and invented ones are discarded. At each passage the scholar must do something: restate the claim in their own words, name the weakest assumption, say what evidence would change the author's mind, or apply it to a real problem.

The next passage opens only when this one is done.

Help escalates, so struggle comes first. On the first wrong attempt Amara says only what kind of thing is missing. On the second she names the sentence to read again. On the third she asks one question. Only on the fourth does she explain it fully — and the system records that the scholar needed it, so their teacher can see. Resending the same answer is refused and does not count as an attempt, so nobody can click their way to the explanation.

Why so careful? Because the moment just before understanding is the moment the dopaminergic state is live. An answer handed over early does not just skip a step; it spends the exact condition that makes the next hour of learning better.

4. "I understand" is a claim, and it gets checked

When a scholar tells Amara they have got it, she does not accept it and move on. She asks one question that only real understanding can answer — apply it to a new case, predict what changes if one thing changes, explain why rather than what.

The feeling of understanding and the ability to use an idea are different things, and a teacher who accepts the feeling teaches the feeling.

5. No points. None. Anywhere.

Uweza had XP, streaks, badges and points on Passport entries. We removed all of it, and then we enforced the removal in the database itself: a rule zeroes any points value on write, whichever of our fifty services is writing. Not a policy a future developer can forget — a floor they cannot fall through.

This was the decision that followed most directly from the evidence. A scholar who finishes a reading because it is worth 20 points has been given a weaker motive in place of a stronger one, and when the points stop, so does the reading.

What replaces them is not nothing. It is evidence: what you did, and how it was verified.

6. Not-knowing is made safe, deliberately

Under every scene is a quiet line: Not clear? Ask Amara. Not a banner, not a prompt, no one else in the room. A scholar can say "I don't understand" to an AI teacher at two in the morning with no social cost whatsoever.

That is psychological safety by construction. It does not depend on a teacher's temperament or a classroom's culture, both of which are hard to change and unevenly distributed.

But because the thing providing that safety is an AI, and because we know AI has led young people toward harm: every one of those conversations is screened by fixed phrase rules in English, French and Pidgin — before Amara sees the message and before her reply is sent — and reviewed by a separate model. Where something is found, a person at Uweza reads it. Opening a scholar's conversation requires a typed reason, which is logged. A flag closes only with a named reviewer and a note. Safety that is not accountable is just surveillance.

7. What a scholar makes outlives the term, and is seen

This is the piece I care most about, because it addresses the condition schools most reliably destroy: nothing a student makes has an audience or a future.

On Uweza, scholars write Works: their own pieces, holding prose, mathematics, chemical equations, code, and photographs of work done with pen and paper — because most real working in Cameroon is done with pen and paper, and a system that cannot accept it is excluding the working, not the student.

Works cite one another. Each shows what it builds on and what has built on it. That is not decoration; it is the structure that turns a pile of student essays into a body of knowledge that a later scholar can enter.

To count on a Passport, a work passes four stages:

  1. Amara's review, against a rubric.
  2. The author's defence — three questions only the person who did the work can answer well, in their own words. Review alone cannot tell whether a scholar understands what they submitted. This can.
  3. A teacher's sign-off, with a sentence the author can learn from.
  4. Seven days open to challenge by other scholars on the same topic, in which any substantive objection must be answered by the author before the work is verified.

That last stage is the one people ask about. It is there because it is what actual scholarship rests on: not a mark from an authority, but exposure to people who know the topic and are permitted to object. A sixteen-year-old whose argument has survived a week of her peers trying to break it has done something no exam can certify.

And it is seen. Works are discoverable by question, topic and subject across the platform. With a teacher's approval, a work can be shared outside Uweza entirely — to a university, a competition, a mentor — as a page anyone can read.

8. Scholars build together, and each carries a part

Projects on Uweza have an Ubuntu board: each member names the subsystem they carry and publishes the figures they produce, which everyone else can see and build on. A bare number is refused — "12 of what?" — so a signal is always a measured thing with a unit and a note.

Four scholars auditing one enterprise, each holding a different part, each dependent on the others' numbers: that is a social environment that rewards contribution and makes concealment useless. It is also, not incidentally, how research actually works.

9. We watch the machine, not just the children

Our Command Centre draws an epistemic map — questions, subject threads, topics — with evidence of mastery on every node, each one clickable into the lessons, the scholars and the checks behind it.

A topic with no lesson is drawn hollow, in a different colour. A check that most scholars fail is ranked at the top of a list titled where they get stuck.

The point of looking at it this way is accountability in the right direction. If a topic is failing, the first hypothesis is not that those children are weak. It is that the conditions there are bad — and the map says exactly where to go and fix them.

V. What I am not claiming

I would rather state the limits than have them found.

This is a design argument supported by evidence, not an evaluated intervention. The findings I have cited are robust. What I have not got is a trial showing that Uweza produces more curious scholars than a good Cameroonian school. We are early, our cohort is small, and I will publish what we find including if it disappoints me.

Conditions are not everything. Temperament and prior experience are real; the distribution of curiosity across people is not flat. The claim is narrower and more useful: the share of the variance that conditions control is very large, largely unexploited, and almost entirely within a school's power. You do not need to win the argument about traits to act on that.

Removing points has a cost I am watching for. Points are a crutch, but a crutch helps someone who cannot yet walk. If what replaces them is weaker for scholars whose conditions are worst, that will show up, and I would rather find it than defend the decision.

Four stages of verification will sometimes be too heavy for a short, good piece of work, and we will hear about it.

VI. What I would say to anyone running a school

You do not need our platform to act on any of this. Most of it is free.

  • Open with the question. Withhold the answer. It costs nothing and it is the single highest-return change available to you.
  • Build the floor before you expect the gap. A child who seems incurious about your subject usually knows too little of it to see what is missing.
  • Make two subjects meet on one question, deliberately, at least once a term. The collision is the lesson.
  • Separate feedback from surveillance. Tell students about their work. Stop telling them they are being watched.
  • Make not-knowing cheap, out loud, repeatedly, by your own example.
  • Stop scoring what you want them to love. Score what must be scored, and leave the rest unpriced.
  • Give their work an audience and a future. Published, cited, challenged, kept. Nothing changes a student's relationship with their own thinking faster than someone else building on it.
  • When a cohort fails, interrogate the conditions first. It is the more likely explanation and the only one you can do anything about.

The reason I built Uweza this way is not pedagogical fashion. It is that the alternative explanation — that the curiosity and inventiveness of a continent's children is simply lower than it needs to be — is both insulting and false, and every year it goes unchallenged it costs us people we cannot afford to lose.

Curiosity is a condition. Conditions are built. We can build them here.

Sources

  • Loewenstein, G. (1994). The Psychology of Curiosity: A Review and Reinterpretation. Psychological Bulletin, 116(1), 75–98.
  • Gruber, M. J., Gelman, B. D., & Ranganath, C. (2014). States of Curiosity Modulate Hippocampus-Dependent Learning via the Dopaminergic Circuit. Neuron, 84(2), 486–496.
  • Amabile, T. M. (1983). The Social Psychology of Creativity: A Componential Conceptualization. Journal of Personality and Social Psychology, 45(2), 357–376.
  • Amabile, T. M. (1996). Creativity in Context. Westview Press.
  • Amabile, T. M., & Pratt, M. G. (2016). The Dynamic Componential Model of Creativity and Innovation in Organizations. Research in Organizational Behavior, 36, 157–183.
  • Byron, K., & Khazanchi, S. (2012). Rewards and Creative Performance: A Meta-Analytic Test of Theoretically Derived Hypotheses. Psychological Bulletin, 138(4), 809–830.
  • Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A Meta-Analytic Review of Experiments Examining the Effects of Extrinsic Rewards on Intrinsic Motivation. Psychological Bulletin, 125(6), 627–668.
  • Edmondson, A. C. (1999). Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly, 44(2), 350–383.