What a Four-Year-Old Likes Is Not a Curriculum: The Case for Structure

This piece argues a position, and it is fair to state it at the top. Unschooling in its strong form — follow the child's interests, supply no curriculum, wait for readiness — has a structural flaw that no amount of trust in the child fixes. Homeschooling, done with real resources, does not have that flaw. The disagreement is not about whether children are curious. It is about what curiosity can act on.

We publish a good deal of sympathetic material on self-directed learning, including what unschooling actually is, and we are not withdrawing it. Unschooling gets several things right that conventional schooling gets badly wrong. But the version that treats structure itself as the enemy is the version that fails, and it fails predictably.

Interest can only choose from what a child has already met

A child cannot develop an interest in number theory, Carnatic music, structural engineering or Ambedkar's writing without first encountering them. Interest is a selection mechanism, and it operates only on the menu it has been shown. A household that supplies no menu has not liberated the child's preferences — it has quietly narrowed them to whatever happened to be lying around: the family's existing hobbies, the parents' own education, the algorithm on the nearest screen.

This is why "we follow her interests" and "we provide wide exposure" are not the same policy, though they are often spoken as if they were. The second is real work — deliberately putting things in front of a child that she did not ask for and could not have asked for. Done well, it is the engine that makes child-led learning produce anything. Skipped, "child-led" collapses into "chance-led", and chance in an Indian home usually means the parents' comfort zone.

Which brings us to the four-year-old. A four-year-old's stated preferences are worth listening to and worth very little as a planning instrument. They are a snapshot of a nervous system that has met almost nothing yet. Optimising a childhood around them is not respect for the child; it is a category error about what a preference is at that age.

What the research says about learning with minimal guidance

The claim that children learn best when left to discover has been tested, and it has not held up well. In one of the most cited papers in instructional psychology, Kirschner, Sweller and Clark reviewed the evidence for constructivist, discovery, problem-based, experiential and inquiry-based teaching, and concluded that minimally guided instruction is less effective than approaches that provide explicit guidance. Their argument is not ideological — it rests on how human working memory, cognitive load and the gap between novices and experts actually behave.

Why Minimal Guidance During Instruction Does Not Work: An Analysis of the Failure of Constructivist, Discovery, Problem-Based, Experiential, and Inquiry-Based Teaching.

That is the title of the paper itself — Kirschner, Sweller and Clark, Educational Psychologist, 2006. The finding that matters for unschooling is the novice–expert distinction: methods that work beautifully for someone who already knows a domain work poorly for someone who does not. An expert exploring a field is doing something cognitively different from a beginner being left to explore it, even though from the outside both look like curiosity.

Critical thinking is not a subject you can wait for

The most common defence of unschooling is that it produces independent, critical thinkers rather than exam-passers. It is the right goal. The evidence just does not support getting there by removing content.

Daniel Willingham's work on this is unambiguous: critical thinking is not a free-floating skill that can be trained in the abstract and then applied anywhere. What counts as good thinking differs by domain, and the ability to reason well inside a domain depends heavily on knowing a lot about it. Advice like "look for hidden assumptions" does almost nothing for a reader who lacks the background knowledge to understand the argument in the first place. His practical conclusion is the inconvenient one for interest-only approaches: if you want a child to think critically, make sure they have plenty of domain knowledge.

Independent thinking, in other words, is downstream of substance. A curriculum is one way of guaranteeing the substance arrives. Waiting for a child to request it is not.

Not everything is equally important, and the clock is not symmetric

A genuine flaw in strong unschooling is the implied claim that all learning is equivalent — that a year spent deep in Minecraft and a year spent becoming fluent in reading and arithmetic are simply different paths. They are not, because some capabilities are load-bearing for everything that comes after. Fluent reading is what makes every later subject accessible. Number sense is what makes secondary mathematics survivable. These are not more important because society says so; they are more important because everything else stacks on them.

The Indian data should sober everyone in this debate, including the pro-school side. ASER 2024, surveying 649,491 children, found that only 44.8% of Std V children in government schools could read a Std II level text, and 59.3% in private schools. Only 30.7% of Std V children nationally could solve a three-digit-by-one-digit division problem. Structure is clearly not sufficient — most Indian classrooms have plenty of it and still miss. But that is an argument for better structure, not for less of it. An approach that supplies still less scaffolding needs a serious answer for how it beats those numbers, and "children learn naturally" is not one.

What unschooling gets right, and we should not pretend otherwise

• Motivation is real and fragile. A child who has been drilled into hating reading has lost something a syllabus cannot give back, and unschoolers take that risk seriously when schools do not.
• Deschooling after a bad school experience genuinely helps. A period without imposed academics is often the right medical response to burnout.
• Age-graded lockstep is arbitrary. A child ready for algebra at nine should do algebra at nine, and a child not ready at thirteen is not broken.
• Teenagers are a different case. By fifteen or sixteen, a young person with foundations in place and a genuine specialism should be given real autonomy — that is where self-direction earns its keep.

Every one of those is compatible with structure. None of them requires abandoning it.

Homeschooling works — on one condition

Homeschooling with a plan, materials, an adult who teaches, and an honest external check on progress is a strong option in India, and the certification route is genuinely open: NIOS registers Secondary candidates at 14, including home learners with no school record, and its certificates carry the same recognition as other boards. The condition is resources — a capable adult's sustained time, or bought teaching where that time is missing.

That condition is the honest dividing line, and it has nothing to do with philosophy. A household with a parent who can teach and the time to do it can homeschool superbly. A household without either is choosing between structured alternatives — an online school, a co-op, tutoring, a hybrid arrangement — and calling the resulting vacuum "unschooling" does not fill it. Our comparison of structured homeschooling and self-directed learning works through what each actually demands.

What we would actually do

1. Guarantee the foundations. Reading fluency and number sense get taught deliberately and early, whether or not the child requests them.
2. Treat exposure as your job, not the child's. Put unfamiliar things in front of them on purpose and regularly; interest cannot select from an empty menu.
3. Let interest drive depth, not coverage. Once the base is secure, follow the child hard into whatever they love — that is where self-direction pays.
4. Check externally. A yearly independent assessment tells you whether the approach is working while there is still time to change it.
5. Widen autonomy with age. Real decision-making power should grow steadily through the teens, ending close to total by seventeen.

The position is not "children need school". It is that children need exposure, foundations and someone taking responsibility for both — and that a four-year-old's preferences, however delightful, are not a plan for a childhood.

Is this saying unschooling always fails?

No. It says the strong form — interest-led with no deliberate exposure or structure — has a flaw that trust in the child does not fix, and that outcomes then depend heavily on how rich the home happens to be in books, conversation and adult attention. Many self-described unschooling families already supply substantial structure; the critique lands on the doctrine, not on them.

What about research showing unschooled adults do fine?

Survey work on grown unschoolers, such as Gray and Riley's, reports largely positive outcomes and is worth reading. Treat it carefully: these are self-selected respondents from committed, resource-rich, often highly educated households, which is exactly the population where informal exposure is abundant. It is weak evidence that the method transfers to households without those advantages.

At what age should a child get real say?

Increasingly, and early, over what to go deep into — but not over whether to become literate and numerate. Broad autonomy over the direction of learning makes sense once foundations are secure, typically from the early teens, and should be close to complete by the end of school.

Does this mean we should buy a boxed curriculum?

Not necessarily. Structure means someone is deliberately deciding what gets covered, checking that it lands, and widening exposure — a curriculum is one way to do that, not the only one. What does not work is nobody holding that responsibility.