Play is one of childhood’s natural operating systems. Children repeat actions, test limits, negotiate rules, imitate others, build objects, compare quantities and discover what happens when a plan fails. That makes play educationally attractive. It is tempting to assume that if an activity feels playful and contains numbers, mathematical learning will follow.
The evidence asks for more precision. The Education Endowment Foundation’s ONE Programme trial, completed in September 2026, tested a 12-week early-years programme that combined practitioner development with short guided-play activities designed to support numeracy and executive functions. The programme was well received, changed practitioners’ understanding, and was feasible enough to run in many settings. Yet children in programme settings made no additional average progress in numeracy compared with controls, and executive-function findings were similarly unconvincing overall. A disadvantaged subgroup showed a positive maths estimate, but the evidence was less secure.
That result is useful because it separates two claims that are often merged: play can contain mathematically valuable activity and a specific play-based programme will reliably raise measured mathematics outcomes. The first can be true while the second remains uncertain.
The practical question is therefore not “Is play good for learning?” It is: what has to be designed into guided play so that children actually notice, practise and generalise mathematical relationships rather than merely enjoy the activity?
The 50-second answer
Guided play sits between free play and direct instruction. The adult preserves meaningful child choice, but the environment, materials, prompts or rules are deliberately arranged around a learning goal. For early numeracy, that goal might involve counting, comparing, composing numbers, recognising patterns, spatial relations or choosing strategies.
The mechanism is not entertainment. It is repeated, motivated contact with a mathematical relation under conditions that require the child to use it. Executive functions such as attention control, remembering rules, switching strategies and inhibiting impulsive responses can also be recruited by games, but recruiting a process is not the same as producing a general improvement in that process.
The 2026 EEF ONE trial is a caution against overselling the category. Its average numeracy effect was essentially zero. Guided play should therefore be judged by the mathematics children can later demonstrate, not by engagement alone. The strongest designs make the target relation visible, use adult prompts sparingly, vary examples, check transfer and connect playful representations back to ordinary mathematical language and symbols.
1. Guided play needs a learning target
Free play can generate rich learning, but guided play earns the adjective guided because somebody has identified a relationship worth noticing. Without a target, adults may celebrate activity while being unable to say what mathematical capability is changing.
A concrete example: a shop game is designed around comparing quantities and making totals, not merely pretending to buy objects.
The caveat: The target should not script every child action. Too much adult control removes the decision-making that gives play its exploratory value.
2. Choice is useful when the mathematical constraint stays real
Children can choose materials, order, roles or routes while still facing a common mathematical demand. Bounded freedom keeps agency without letting the target disappear.
A concrete example: children choose which block tower to build, but every tower challenge requires matching a stated height or using exactly ten blocks in more than one arrangement.
The caveat: If some choices let children avoid the mathematical relation, the activity is only superficially differentiated.
3. Rules create cognitive demand
Games recruit memory and inhibition because players must remember what counts, wait, update and sometimes change strategy. Those demands can support self-regulation inside the task.
A concrete example: a number-card game asks children to remember the current target and resist grabbing the first visible match.
The caveat: Harder rules are not automatically better. If rule management consumes all attention, the mathematical idea may receive too little processing.
4. Counting needs one-to-one correspondence
Young children can recite number words without coordinating each word to exactly one object. Play can make that coordination necessary.
A concrete example: a delivery game requires one parcel for each toy house, with the child checking whether any house is missing or doubled.
The caveat: Speedy chanting can hide weak correspondence. Adults should watch the mapping, not only the final number.
5. Cardinality must become more than the last word said
Children need to understand that the final count word names the quantity of the whole set. Play offers many chances to ask for the amount after counting rather than restarting every time.
A concrete example: after counting seven counters, the adult asks, ‘So how many are there altogether?’ then rearranges them without changing quantity.
The caveat: Repeated recounting may indicate that cardinality is not secure. The game should reveal this rather than simply move on.
6. Comparison becomes mathematical when the child must justify it
Words such as more, fewer, greater and less encode relationships. Guided play can make comparison consequential rather than verbal decoration.
A concrete example: two toy animals receive different piles of food and the child must decide who has more and explain how they know.
The caveat: Visual size can mislead when objects differ in spacing or physical size. Vary representations so children compare number, not appearance.
7. Composition of number needs flexible part-whole thinking
Knowing that eight can be made as five and three, six and two, or four and four supports later arithmetic. Games can expose multiple decompositions naturally.
A concrete example: children build a train of eight cubes using two colours and find as many colour splits as they can.
The caveat: If adults praise only the first successful split, the child may miss the flexibility that matters.
8. Spatial language can carry mathematical structure
Words such as above, between, turn, edge, corner, near and far help children represent space. Guided construction makes these words operational.
A concrete example: one child gives instructions for placing a block while another builds without seeing the model.
The caveat: Language prompts help only when they refer to relationships the child can inspect. A vocabulary list detached from action is weaker.
9. Pattern games should require prediction
Repeating a pattern is less demanding than identifying the unit and predicting what comes next. Play can move from copying to rule detection.
A concrete example: beads follow red-blue-blue, red-blue-blue, and the child must continue the pattern and explain the repeating unit.
The caveat: A visually attractive sequence can be copied without understanding. Change colours or objects while keeping the rule.
10. Manipulatives are representations, not magic objects
Blocks, counters and shapes can make relations visible, but children need help connecting physical actions to mathematical ideas.
A concrete example: a child combines two groups of counters, then matches the action to a simple number sentence.
The caveat: Manipulatives can become toys disconnected from the target. The adult must occasionally name the relation and bridge to symbols.
11. Adult prompts should extend, not commandeer
The best prompts preserve the child’s ownership while directing attention toward structure. Questions such as ‘How do you know?’ or ‘Can you make it another way?’ invite reasoning without supplying the answer.
A concrete example: after a child builds six as four and two, the adult asks whether another split is possible.
The caveat: Constant questioning can turn play into an oral test. Adults should watch, wait and intervene selectively.
12. Modelling can open the game
A short demonstration can show how materials work or what the rule means, especially when the task is novel.
A concrete example: the adult plays one round aloud, naming the decision point, then hands control to the child.
The caveat: Overmodelling can pre-solve the strategy. Show enough to enter, not enough to remove discovery.
13. Peer interaction can create mathematical language
Children explain, negotiate and correct one another during games. That talk can make reasoning visible and provide repeated use of comparison and quantity language.
A concrete example: two children must agree which container holds more before they can move on.
The caveat: One confident child can dominate decisions. Roles or turn-taking may be needed so both learners do the thinking.
14. Executive function inside a game is task-specific evidence
Waiting, switching and remembering rules demonstrate regulation in that context. It is tempting to generalise from game performance to broad executive-function growth.
A concrete example: a child successfully switches sorting rules when the card changes from colour to shape.
The caveat: The ONE trial did not show secure overall executive-function gains. Practising control inside one activity does not guarantee broad transfer.
15. Engagement is a condition, not an outcome
Enjoyment can increase time on task and willingness to repeat. That is valuable because practice opportunities matter.
A concrete example: children voluntarily replay a counting game several times, giving the target relation repeated exposure.
The caveat: High engagement can coexist with low learning if the game rewards speed, luck or decoration rather than the intended mathematical reasoning.
16. Repetition should vary the surface
Learning strengthens when the same relation appears across different materials and situations. Variation helps children see what stays invariant.
A concrete example: the idea of five appears with fingers, counters, steps, dots and toy animals.
The caveat: Too much novelty at once can raise interface cost. Keep the relation stable while varying one feature at a time.
17. Feedback should preserve play
Children need accurate information, but correction can be embedded in the activity rather than delivered as constant judgement.
A concrete example: a game outcome reveals that one token is missing, prompting the child to recount and repair.
The caveat: If every error stops the game for a mini-lecture, motivation and flow can collapse.
18. Play can reveal misconceptions early
Because children act, teachers can see how they count, compare, group and interpret rules. This creates diagnostic information that worksheets may hide.
A concrete example: a child consistently counts a moved object twice, revealing unstable tracking.
The caveat: Observation should lead to targeted teaching. Simply recording that the child ‘struggled’ is too vague.
19. The mathematical language should be precise
Everyday words are useful, but children also need stable terms for relations. Guided play gives those words a meaningful setting.
A concrete example: the adult contrasts more with bigger when a physically larger object belongs to the smaller set.
The caveat: Precision should not become correction of every casual sentence. The goal is conceptual clarity, not policing speech.
20. The game should connect to later independent evidence
The strongest test is whether the child can use the idea outside the original playful surface.
A concrete example: after a sharing game, the child solves a simple picture problem involving equal groups without the game board.
The caveat: Transfer can lag. Failure on one new item does not erase learning, but it tells the teacher not to assume generalisation.
21. Short activities can be enough when the target is sharp
The ONE programme used activities lasting roughly five to ten minutes. Early-years attention does not require long formal sessions when the relationship is repeatedly revisited.
A concrete example: a six-minute comparison game appears several times across the week with new materials.
The caveat: Short duration is not a virtue if setup consumes most of the time or the target is never revisited.
22. Continuous provision can carry mathematics
Early-years settings already contain blocks, pretend shops, sand, water, movement and sorting opportunities. Guided play can make the mathematical possibilities more deliberate.
A concrete example: a construction area includes challenge cards that invite symmetry, length comparison and counting.
The caveat: Turning every play area into a teacher task can reduce genuine exploration. Some spaces should remain less directed.
23. Practitioner knowledge matters
Adults need to recognise the mathematics inside children’s actions so prompts arrive at useful moments. Professional learning can improve that noticing even when outcome effects are uncertain.
A concrete example: a practitioner sees that a child’s tower-building problem is really about unit iteration and comparison.
The caveat: Training that changes adult knowledge is promising but should not be confused with demonstrated child attainment.
24. Implementation fidelity is not identical to rigid delivery
A programme may specify activities while allowing contextual adaptation. The important question is whether the mechanism survives the adaptation.
A concrete example: a nursery substitutes local materials for programme blocks while preserving the same comparison rule.
The caveat: Changing rules, target concepts and sequence simultaneously can make it impossible to know what is being implemented.
25. The null average result deserves respect
The EEF trial found no additional average numeracy progress. That should lower certainty around claims that play-plus-executive-function training automatically raises maths attainment.
A concrete example: a setting continues to use a game because children enjoy it but does not advertise it as a proven route to higher maths scores.
The caveat: A null average effect does not mean every activity is useless. It means local value should be demonstrated rather than presumed.
26. Subgroup signals need caution
The disadvantaged subgroup in the ONE trial showed a positive estimate, but the sample was smaller and the finding less secure.
A concrete example: schools may view the signal as a reason for further study rather than as a guarantee of two months’ progress.
The caveat: Subgroup results are especially vulnerable to overinterpretation when numbers are small.
27. Opportunity cost matters even in play
A ten-minute guided activity replaces another possible experience. Schools should ask what children are not doing during that time.
A concrete example: compare the value of a guided numeracy game with high-quality shared reading, outdoor movement or direct language teaching for the same group.
The caveat: No educational activity is free simply because it is enjoyable.
28. Not every mathematical idea belongs in a game
Some conventions, vocabulary and procedures may be taught more efficiently through brief explicit explanation followed by practice.
A concrete example: the adult directly names the equal sign meaning before using games to explore equivalence.
The caveat: Forcing every concept into play can create unnecessary complexity and hide the structure.
29. Guided play can prepare, practise or consolidate
The same format can serve different phases. Before teaching, it can reveal prior knowledge; during teaching, it can rehearse a relation; afterward, it can consolidate and vary examples.
A concrete example: a balance game first exposes intuitive comparison, later rehearses equality, then applies it in new contexts.
The caveat: Teachers should know which phase they are using because the prompts and success criteria differ.
30. Families can use playful mathematics without buying kits
Everyday objects create abundant mathematical opportunities when adults focus on relationships rather than products.
A concrete example: sorting laundry by size, sharing fruit, counting steps and comparing containers all recruit early numeracy.
The caveat: Parents should not be made to feel that ordinary home play requires constant instructional commentary.
31. The child should eventually be able to explain the game mathematically
As language grows, asking children to name what made a move legal or successful helps connect playful action to explicit knowledge.
A concrete example: a child explains, ‘I won because I made two groups of four, so both sides were equal.’
The caveat: Verbal explanation is not the only evidence for very young children. Gesture and action can show understanding before language catches up.
32. A strong design has an exit from the game
If the mathematical relation only exists inside one game, learning remains context-bound. The teacher should deliberately move toward diagrams, stories or symbols.
A concrete example: after several rounds of a sharing game, children solve a simple draw-and-share task on paper.
The caveat: Moving to abstraction too quickly can discard the support before the concept is stable.
33. The best question is what the child learned to notice
Guided play is powerful when it changes attention. Children begin to see number, pattern, space and relation where previously they saw only objects and actions.
A concrete example: a child spontaneously notices that two different block arrangements still contain eight pieces.
The caveat: If the activity produces only compliance with game rules, the mathematical noticing may not have developed.
Practical route for teachers
Choose one mathematical relation at a time. Design the game so success requires that relation rather than luck or speed. Watch children before prompting. Use questions that expose reasoning, vary examples across the week, and finish with one non-game transfer check.
Keep a light record of what children can now notice or do independently. If a game remains highly engaging but produces no change in mathematical language, strategy or transfer, redesign it or use the time elsewhere.
Practical route for students
For older children reflecting on early-years style games, the useful habit is to ask what the game is making you practise. Is it counting, comparing, remembering a rule, noticing a pattern or switching strategy? Naming the job helps you transfer it.
When the game ends, try the same idea in a different form. If you can only succeed with the original cards, blocks or rule, the learning may still be tied to the surface.
Practical route for parents and families
Play mathematics does not require expensive resources. Share food, compare containers, sort objects, build equal towers, count steps, notice patterns and ask children to explain choices. Keep questions natural and let the child lead much of the activity.
The goal is not to turn home into school. A few precise mathematical words and occasional prompts can be enough. Enjoyment matters, but the best sign of learning is that the child begins noticing number and relation without being told.
Common failure modes
- Calling an activity mathematical because it contains numbers without requiring mathematical reasoning.
- Adding so many game rules that children spend all their effort remembering procedure.
- Treating engagement as evidence of attainment.
- Using manipulatives without connecting actions to mathematical language or symbols.
- Questioning children continuously until play becomes an oral test.
- Assuming executive-function demand inside a game produces broad executive-function gains.
- Keeping one surface so long that children learn the game rather than the relation.
- Moving to abstract symbols before the concrete relationship is secure.
- Advertising subgroup signals from small samples as settled effects.
- Preserving a popular activity after transfer evidence shows little learning.
Frequently asked questions
Does guided play improve early Mathematics?
It can create mathematically rich practice, but specific programme effects vary. The 2026 EEF ONE trial found no additional average numeracy progress overall, so claims should stay conditional.
Is free play enough?
Free play can support development, but if the goal is a particular mathematical relation, adults may need to shape materials, rules or prompts so that the relation is actually used.
Do games improve executive function?
Games can recruit attention, inhibition and rule switching. Broad improvement beyond the trained activities is a stronger claim and was not securely demonstrated in the ONE trial.
How much adult guidance is appropriate?
Enough to make the target relation available and keep the task productive, but not so much that the adult makes all decisions.
What is the best evidence of learning?
Independent use of the same mathematical idea in a changed context, representation or problem.
The final idea
Guided play is strongest when it preserves two things at once: the child’s agency and the mathematics’ integrity.
The child chooses, experiments and repeats. The adult quietly protects the relation worth learning. Materials make structure visible. Language names it. Variation tests whether it survives. A later non-game task checks whether the idea travelled.
The 2026 evidence is valuable precisely because it prevents us from turning an appealing philosophy into an automatic attainment claim. Play can be rich, meaningful and developmentally appropriate without every packaged programme producing measurable gains.
The rule is simple: do not ask whether the child had fun with mathematics; ask what mathematical relationship became easier to see, use and carry elsewhere.
Sources and further reading
- Education Endowment Foundation — The ONE Programme trial
- Education Endowment Foundation — Early Mathematics evidence
- What Works Clearinghouse — Preparing Young Children for School
Continue exploring on eduKateSG
- How X Works Hub
- Why Small Relational Words Can Unlock Big Ideas | How Concept Vocabulary Works in Early Learning
- How Embodied Learning Works | When Gesture, Movement and Manipulation Carry Meaning
- Why Can One Task Challenge a Whole Class? | How Low-Floor, High-Ceiling Learning Tasks Work
Casebook: six guided-play designs and what to watch
Case A: make ten in more than one way
Give children ten counters and two bowls. The rule is simple: every round must use all ten counters, but the split must change. One child hides one bowl while the other predicts what is inside after seeing the first bowl. The mathematics is part-whole composition; the executive demand is remembering the total and resisting the urge to recount everything from one.
Watch for whether children can infer the hidden part, whether they rely on counting all, and whether they notice commutativity across arrangements. The game is successful when children begin to predict combinations rather than merely move counters.
Case B: the shop with constrained coins
Set up a pretend shop with prices small enough for the age group and give each child a restricted set of tokens. The child chooses what to buy but must decide whether the tokens are sufficient, whether two items can be combined, and what remains.
The adult can vary the mathematical job by changing prices or token sets while keeping the social play familiar. Avoid turning the game into a worksheet performed beside toy food. The value comes from the child needing number relationships to make meaningful choices inside the scenario.
Case C: build the same number differently
Ask children to build towers containing the same total number of blocks but using different shapes, colours or sub-towers. Then compare. Which towers look different but are equal in number? Which are taller despite containing fewer blocks?
This task separates quantity from visual size and creates a reason to use words such as equal, more, fewer, taller and shorter. It also exposes whether children are counting accurately or relying on appearance.
Case D: pattern detective
Create a repeating movement, sound or object sequence and deliberately stop before the next element. Children predict what comes next, identify the repeating unit and then invent a new surface using the same rule.
The transfer step matters. A child who can continue red-blue-blue but cannot recognise clap-stamp-stamp as the same pattern may have learned the colours rather than the structure. Guided play should gradually make the rule more portable.
Case E: rule-switch sorting
Children sort cards by colour until a cue changes the rule to shape. Later the cue changes again. The activity clearly recruits executive control, but the mathematical learning depends on the classification language and logic built around the switch.
Ask children to explain the current rule and why an item belongs. Do not assume success at switching proves a general executive-function gain. The value is the observed capability inside the task and any demonstrated transfer later.
Case F: spatial instruction relay
One child builds a small block model behind a screen and gives verbal instructions to a partner. The speaker must use position, direction and shape language; the listener must hold and execute the sequence.
Compare the two structures at the end. The discrepancy becomes feedback. Repeat with changed roles and simpler or harder models. The game integrates spatial language, perspective and working memory while still leaving a visible product that can be checked.
A local evaluation plan for guided play
Schools or early-years settings can evaluate one play routine without pretending to run a formal trial. First, state the target relation in observable terms: for example, “children will decompose eight in at least three ways without recounting the whole.” Second, collect a brief baseline using a non-game task. Third, run the game repeatedly over a short period. Fourth, repeat a changed transfer task.
Record engagement separately from mathematical performance. A routine can be popular but weak instructionally, or mathematically useful but poorly delivered. Separating those dimensions improves the next design decision.
Finally, compare opportunity cost. If the same target can be taught more efficiently through a brief explicit explanation plus practice, use that. Guided play earns its place when agency, repetition and embedded decision-making add something that simpler instruction would not.
Three questions to ask before calling a play activity mathematical
First, what relation must the child use to succeed? If the answer is vague—“they are using numbers”—the task is probably underspecified. Name the actual relation: equal groups, one more, part–whole composition, spatial transformation, repeating unit, comparison or order.
Second, what would a misconception look like? A useful activity makes wrong thinking visible enough that the adult can respond. If every child can appear successful by copying a peer or moving objects randomly, the game is poor diagnostic material.
Third, what will count as transfer? Decide before the activity how the child might show the same understanding elsewhere. A changed material, picture, story or symbol is often enough. Without that later check, adults can easily mistake familiarity with the game for mathematical learning.
Why play should sometimes remain play
Not every playful episode needs an adult learning target. Children also need room to invent purposes, repeat actions for their own reasons, negotiate social worlds and explore materials without being constantly steered toward curriculum language.
That does not weaken guided play. It clarifies its place. Guided play is a deliberate instructional mode, not a claim that all play must be instructionalised. Strong early-years provision can contain both: open play whose value includes agency and exploration, and guided play whose value depends on a specific relationship becoming more visible.
The distinction protects children from a day in which every enjoyable activity becomes an assessment opportunity.
A teacher checklist before the game begins
Before introducing a guided-play activity, check six things. Target: can you name the exact mathematical relationship? Entry: can children understand the materials and rules without a long explanation? Decision: does the child have to make a mathematical choice, or can success come from luck? Language: which two or three words will help children name the relation? Feedback: what will happen when a child uses a weak strategy? Transfer: what small later task will show whether the idea survived outside the game?
During play, observe more than correct outcomes. Watch counting behaviour, strategy changes, peer explanations and whether children notice invariants. A child who gets the answer through guessing is in a different state from one who can justify the move.
After play, resist the urge to turn the debrief into a lecture. Ask one or two questions that compress the mathematical idea: “What stayed the same?” “How did you know the groups were equal?” “Can you make the same total another way?” Then move to a changed representation.
The checklist matters because play can hide weak learning unusually well. Smiles, movement and repeated turns create visible activity. The adult’s job is to make sure the activity contains the mathematical work intended.
A parent-facing way to recognise mathematical play
Parents can look for four signs. The child is making decisions rather than simply following a script. The game creates a reason to count, compare, compose, predict or describe space. The adult uses mathematical language naturally rather than constantly testing. And, later, the child notices the same idea in another setting.
A child who says at dinner, “We have six spoons, so if two people are missing we need four,” has carried quantity reasoning beyond the game. That spontaneous transfer is more interesting than how enthusiastically the child completed a branded activity.
The simplest family principle is to notice mathematics already present in ordinary life and give children enough time to reason aloud.