My Child Can Start a Mathematics Question but Gets Stuck Halfway

Part I — Why a Correct Start Can Still Collapse Halfway

Starting a Mathematics question correctly proves only that the learner recognised the opening structure. Multi-step Mathematics asks for something harder: after every successful move, the problem changes. A new value appears, a constraint becomes usable, a diagram gains information, an equation changes form, or one branch becomes more promising than another. The learner must repeatedly read the new state and decide what to do next.

The halfway stall is therefore often a state-update problem, a subgoal problem, a method-handover problem or a monitoring problem. It may also expose algebra, notation, units, working-memory load or overdependence on tutor prompts. The correct intervention depends on which link breaks.

Every useful step should change what the student knows. Every new state should change what the student considers next.

Thirty-five halfway-stall mechanisms

Correct first formula, no next move

Signal. The learner can substitute correctly but treats the result as an endpoint.

Likely bottleneck. Ask what the new value represents and which remaining unknown it can unlock.

Practice. Train a one-line state update after every major intermediate result.

Verification. Verify on a changed problem where the same kind of result leads to a different second method.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Correct first subpart, later subpart ignored

Signal. The learner does not treat earlier answers as reusable information.

Likely bottleneck. Mark reusable results clearly and scan later parts for dependencies.

Practice. Practise multi-part continuation with earlier outputs already available.

Verification. Verify that later parts use prior results without tutor reminders.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Main target disappears

Signal. The learner becomes absorbed in local calculations.

Likely bottleneck. Keep the final target visible and restate it after major subgoals.

Practice. Use a target ladder: final target, next required quantity, current operation.

Verification. Verify that the learner can explain how each intermediate result serves the final goal.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Topic recognised, next action unclear

Signal. The learner knows ‘this is trigonometry’ or ‘this is algebra’ but cannot plan locally.

Likely bottleneck. Ask what quantity or relationship must be created next.

Practice. Replace topic labels with specific subgoals.

Verification. Verify on several questions from the same topic with different second steps.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Continuous calculation without direction

Signal. The learner equates activity with progress.

Likely bottleneck. Insert short route checks: what did this step produce, and what is now possible?

Practice. Use micro-plans before the next few lines.

Verification. Verify that irrelevant calculations decrease.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Messy algebra after a good setup

Signal. Representation cost becomes the bottleneck.

Likely bottleneck. Return to the last clean line and simplify, factor, substitute or relabel before continuing.

Practice. Practise continuation from deliberately messy partial solutions.

Verification. Verify fewer complete restarts.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Negative signs are lost midway

Signal. High-risk algebraic handovers contaminate later work.

Likely bottleneck. Slow only the transitions involving brackets, subtraction and substitution.

Practice. Use selective sign checks rather than narrating every line.

Verification. Verify across several topics where the same sign mechanism appears.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Fractions derail the chain

Signal. The mathematical idea is intact but fraction execution consumes attention.

Likely bottleneck. Repair the exact fraction operation separately, then reinsert it into the multi-step route.

Practice. Allow side-line simplification where helpful.

Verification. Verify fraction control inside another topic.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

No bridge quantity is identified

Signal. The final target requires an intermediate quantity the learner has not named.

Likely bottleneck. Work backward from the final step: what would make that step possible?

Practice. Then work forward from the givens and look for overlap.

Verification. Verify on geometry, rate and modeling questions with hidden bridge quantities.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Intermediate answer is not interpreted

Signal. The learner computes but does not update meaning.

Likely bottleneck. Require a short statement: ‘This value is ___, so now I can ___.’

Practice. Use especially for gradients, rates, lengths, probabilities and parameters.

Verification. Verify correct use of the result later in the same solution.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Diagram is not updated

Signal. Derived facts remain outside the representation.

Likely bottleneck. Add justified angles, lengths, coordinates or labels as soon as they are found.

Practice. Use the richer diagram to reduce memory load.

Verification. Verify on questions requiring several derived facts.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Equation is correct but directionless

Signal. Legal transformations occur without movement toward the target.

Likely bottleneck. Ask what form or variable the equation ultimately needs to produce.

Practice. Choose transformations that reduce distance to that goal.

Verification. Verify with equations where several legal but differently efficient routes exist.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

A plausible branch becomes unproductive

Signal. The learner cannot tell when to revise a plan.

Likely bottleneck. After two or three moves, ask whether new structure is appearing.

Practice. Return to the branch point and choose another representation or method for a reason.

Verification. Verify on problems with one tempting inefficient route.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Sunk cost traps the learner

Signal. Time already invested makes the student persist on a dead branch.

Likely bottleneck. Review what the last minute actually produced.

Practice. Practise preserving valid work and abandoning only the unproductive branch.

Verification. Verify lower stall duration under time.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Every doubt causes full restart

Signal. A local uncertainty becomes a global reset.

Likely bottleneck. Locate the earliest uncertain transition rather than erasing everything.

Practice. Preserve confirmed values and relations.

Verification. Verify more local corrections and fewer complete restarts.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Wrong intermediate value contaminates everything

Signal. The learner cannot isolate where the error entered.

Likely bottleneck. Trace the dependency chain and identify which later lines depend on the wrong value.

Practice. Correct locally where possible and preserve unaffected reasoning.

Verification. Verify on multi-part tasks with planted intermediate errors.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Too many quantities are held mentally

Signal. Working-memory load rises until values are confused.

Likely bottleneck. Externalise values in a table, diagram or labeled list.

Practice. Keep important intermediate values visible.

Verification. Verify lower transcription and substitution error rates.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Too many symbols are introduced

Signal. Notation itself becomes cognitive load.

Likely bottleneck. Use meaningful variable names or concise annotations and remove unnecessary symbols.

Practice. Keep definitions visible in modeling work.

Verification. Verify fewer symbol-confusion errors.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Exact and rounded values are mixed

Signal. Precision state becomes unclear.

Likely bottleneck. Label approximations and preserve exact forms where useful.

Practice. Delay rounding until the appropriate stage.

Verification. Verify on multi-stage numerical problems.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Units disappear

Signal. Intermediate values lose physical meaning.

Likely bottleneck. Carry units or dimensional meaning through high-risk steps.

Practice. Use units as a route and checking cue.

Verification. Verify on rates, mensuration and applications.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Representations do not connect

Signal. Diagram, equation and graph are treated as separate tasks.

Likely bottleneck. Ask what feature in one representation corresponds to another.

Practice. Practise explicit translation between forms.

Verification. Verify on problems requiring representation switching midway.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Methods are known but cannot be chained

Signal. Individual tools exist but handovers do not.

Likely bottleneck. Practise two-method chains before longer sequences.

Practice. State explicitly how the output of Method A becomes the input of Method B.

Verification. Verify on new topic combinations.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Correct second method is chosen too late

Signal. Selection latency consumes time.

Likely bottleneck. Use partial-solution continuation drills and time only the next-target decision.

Practice. Keep calculations simple during diagnosis.

Verification. Verify faster continuation with stable accuracy.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Someone else’s partial solution cannot be continued

Signal. The learner depends on a personally familiar path.

Likely bottleneck. Give valid partial solutions at different stages and ask the learner to continue.

Practice. This trains state reading rather than first-step generation.

Verification. Verify across different solution styles.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

One tutor hint leads to another

Signal. The hint solves only the immediate gap.

Likely bottleneck. After each hint, require the learner to state what changed and what new options are now available.

Practice. Do not supply the next bridge automatically.

Verification. Verify falling sequential hint counts.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Two-step questions work, four-step questions fail

Signal. Coordination load rises with chain length.

Likely bottleneck. Increase chain length gradually while keeping individual methods familiar.

Practice. Track where the first failure occurs.

Verification. Verify later first-error points and more complete chains.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Substitution creates chaos

Signal. Brackets and new expressions overwhelm structure.

Likely bottleneck. Write the substitution completely, then simplify immediately.

Practice. Use clear line separation around the substitution handover.

Verification. Verify in formulas, simultaneous equations and calculus.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Factorisation is found but not used

Signal. The learner treats factorisation as a final product.

Likely bottleneck. Ask what factorisation now makes possible: cancellation, zero-product solving or structure recognition.

Practice. Practise factorisation as an intermediate move.

Verification. Verify in algebraic fractions and quadratics.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Gradient is found but not used

Signal. The learner does not connect gradient to line equations, rates or interpretation.

Likely bottleneck. Require a ‘gradient now tells me…’ statement.

Practice. Use coordinate and graph continuation tasks.

Verification. Verify complete chains on new line problems.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Angle is found but not used

Signal. The new angle is not added back into the geometric state.

Likely bottleneck. Update the diagram and ask what triangle or theorem is now unlocked.

Practice. Practise angle-as-bridge problems.

Verification. Verify unfamiliar geometry continuation.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Length is found but not used

Signal. A new side length remains isolated.

Likely bottleneck. Label it on the diagram and ask which relation now has the required inputs.

Practice. Use composite geometry and mensuration.

Verification. Verify correct method handover.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Derivative is found but not used

Signal. Differentiation is treated as the end of the question.

Likely bottleneck. Ask what the derivative represents and what the final task requires.

Practice. Practise derivative-to-tangent, stationary-point and optimization handovers.

Verification. Verify across changed calculus applications.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Antiderivative is found but not used

Signal. Integration technique is disconnected from area, limits or context.

Likely bottleneck. Restate whether the task requires area, accumulation or a family of functions.

Practice. Use integration-as-bridge questions.

Verification. Verify changed applications.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Variables lose meaning in modeling

Signal. Symbols drift away from the original situation.

Likely bottleneck. Keep variable definitions and units visible.

Practice. Translate important algebraic results back into context at checkpoints.

Verification. Verify on longer modelling questions.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

The learner cannot tell when enough information exists

Signal. All needed quantities are available but the student keeps searching.

Likely bottleneck. Ask what the final step needs and compare that list with current knowns.

Practice. Train stopping rules alongside continuation rules.

Verification. Verify quicker completion after long setup phases.

The important training point is the handover after a successful step. Instead of asking the tutor for ‘the next step’, the learner should update the problem state, name a smaller target and choose a justified next move from the new information.

If the same pattern appears in several chapters, treat it as one cross-topic process weakness. That is usually higher value than re-teaching each chapter separately.

Part I handoff

These mechanisms explain why a student can look capable at the beginning of a problem and still fail to complete it. Part II should now train continuation directly: partial-solution drills, missing-middle tasks, branch comparison, representation changes, error containment, prompt fading and transfer to full multi-step questions.

Part II — How to Train the Middle of a Solution

Thirty-five continuation drills

State snapshot

Pause after a major step and list current knowns, unknowns and constraints.

Why it helps. This trains the learner to read the new problem state instead of operating from memory of the original question.

How to use it. Use only on longer questions at first, then fade to a brief mental check.

Success condition. Success means the learner can update state without tutor prompting.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Target ladder

Write the final target, the quantity needed immediately before it, and the current local target.

Why it helps. This converts a long problem into a short chain of subgoals.

How to use it. Use when the learner cannot identify what to find next.

Success condition. Success means subgoals are generated independently.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Backward bridge

Work one step backward from the final operation.

Why it helps. This is useful when the end of the problem is clearer than the middle.

How to use it. Ask what would make the last step possible, then search the givens for a route to that bridge.

Success condition. Success means backward and forward reasoning meet without tutor rescue.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Forward inventory

Generate only justified facts available from the givens.

Why it helps. This can reveal relationships when the route is initially unclear.

How to use it. Avoid producing every possible fact; select information likely to matter to the target.

Success condition. Success means local deductions become purposeful rather than exhaustive.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Partial-solution continuation

Give the first half of a valid solution and ask the learner to continue.

Why it helps. This isolates middle-route reasoning from first-step recognition.

How to use it. Vary where the partial solution stops so the learner reads different problem states.

Success condition. Success means continuation works across multiple starting points.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Missing-middle task

Provide the opening and final line with one or more missing intermediate steps.

Why it helps. The learner reconstructs the bridge that must connect them.

How to use it. Use in algebra, proofs, geometry and modeling.

Success condition. Success means the learner can infer necessary intermediate relationships.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Branch comparison

Present two plausible next moves and ask which is more productive.

Why it helps. This trains route evaluation rather than blind persistence.

How to use it. Ask what each branch is likely to produce before calculating fully.

Success condition. Success means the learner can justify branch choice.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Dead-end diagnosis

Show an unproductive partial route and ask where it stops creating useful structure.

Why it helps. This teaches evidence for abandoning a branch.

How to use it. Require the learner to identify the last productive step.

Success condition. Success means dead branches are detected earlier in independent work.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Last-clean-line recovery

Ask the learner to mark the most recent unquestionably valid line.

Why it helps. This creates a local restart point after an error.

How to use it. Use when the student tends to erase entire solutions.

Success condition. Success means corrections restart locally.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Representation switch

Move from words to diagram, graph to equation, equation to table or vice versa.

Why it helps. This is useful when the current form hides the relationship.

How to use it. Ask what information becomes easier to see in the new representation.

Success condition. Success means representation change is chosen for a reason.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Intermediate labeling

Require clear names for derived lengths, rates, gradients, probabilities or parameters.

Why it helps. This reduces later substitution errors and memory load.

How to use it. Use especially in long applied questions.

Success condition. Success means intermediate results remain usable several steps later.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Dependency map

Draw arrows from earlier results to later targets.

Why it helps. This makes multi-step dependence explicit.

How to use it. Use when one wrong intermediate value contaminates many later lines.

Success condition. Success means the learner can identify which later work depends on which result.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Method handover sentence

After completing one method, write one sentence explaining how its output becomes input to the next.

Why it helps. This strengthens chaining.

How to use it. Use selectively on representative problems.

Success condition. Success means handovers become internalised and no longer require writing every time.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Micro-plan

Before the next three or four lines, state the local goal.

Why it helps. This prevents long stretches of unstructured manipulation.

How to use it. Keep the plan short enough that it does not slow routine work.

Success condition. Success means the student’s calculations become visibly goal-directed.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Progress check

After a high-cost step, ask whether the target is closer and what new information was gained.

Why it helps. This helps detect dead routes early.

How to use it. Use in longer problems and proofs.

Success condition. Success means plan revision happens before large time losses.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Plausibility checkpoint

Use sign, units, magnitude, graph sense or domain after a risky intermediate result.

Why it helps. This catches contamination before it spreads.

How to use it. Do not check every line; target high-cost transitions.

Success condition. Success means fewer long chains built on impossible values.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Substitution checkpoint

Plug an important intermediate value back into the defining relationship.

Why it helps. This is useful when later work depends heavily on that value.

How to use it. Keep the check cheap.

Success condition. Success means high-cost local errors are caught early.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Simpler-case probe

Reduce numerical or structural complexity while preserving the relationship.

Why it helps. This can reveal a hidden pattern or method.

How to use it. The learner must then return to the original problem.

Success condition. Success means insight transfers back to the full case.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Boundary probe

Test zero, equality, extreme or limiting cases.

Why it helps. This can expose impossible branches or misinterpreted formulas.

How to use it. Use only where the mathematics supports it.

Success condition. Success means the learner uses boundary sense as a diagnostic tool.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Diagram refresh

Redraw a cluttered figure after several derived facts.

Why it helps. This reduces visual overload and makes the new state visible.

How to use it. Use in geometry, trigonometry and coordinate problems.

Success condition. Success means the learner can create a clearer representation independently.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Table refresh

Reorganise changing values or cases in a table.

Why it helps. This helps with rates, sequences, probability and data.

How to use it. Use when relationships are spread across prose or several lines.

Success condition. Success means the table reveals the next mathematical move.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Variable reset

Restate what each symbol means.

Why it helps. This is useful in modeling, functions and long algebraic derivations.

How to use it. Use units and constraints where helpful.

Success condition. Success means symbols keep stable meaning through the full chain.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Unit reset

Write or check units on current intermediate quantities.

Why it helps. This can expose invalid operations and missing conversions.

How to use it. Use in applied mathematics.

Success condition. Success means dimensional sense guides route decisions.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Exact/approximate split

Mark which values remain exact and which are rounded.

Why it helps. This prevents cumulative precision loss.

How to use it. Use in trigonometry, surds and multi-stage numerical work.

Success condition. Success means rounding happens at the right point.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Prompt-fading continuation

Tutor asks broad continuation questions first, then removes them over successive problems.

Why it helps. This targets tutor dependence directly.

How to use it. Track how many prompts are needed after the first step.

Success condition. Success means the chain continues silently.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Timed continuation

Provide a correct partial solution and measure time to identify the next subgoal and method.

Why it helps. This isolates continuation speed from opening recognition.

How to use it. Keep arithmetic modest during diagnosis.

Success condition. Success means faster next-step selection with stable quality.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Mixed continuation

Use partial solutions from several topics in one set.

Why it helps. This tests whether middle-route skill transfers beyond one chapter.

How to use it. Begin only after the continuation routine is stable.

Success condition. Success means the learner can read states without topic cues.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Error-in-the-middle task

Insert one incorrect intermediate line into an otherwise valid solution.

Why it helps. Ask the learner to identify the first wrong decision and continue correctly.

How to use it. This combines checking with continuation.

Success condition. Success means local error detection becomes faster.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Handover narration

Explain only the sentence linking one method to the next.

Why it helps. This captures causal structure without overexplaining routine arithmetic.

How to use it. Use in two-method and three-method problems.

Success condition. Success means the learner can later internalise the handover.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Subgoal generation drill

Give a final target and the current state, then ask for the next useful unknown.

Why it helps. This directly trains planning.

How to use it. No calculation is required at first.

Success condition. Success means the learner proposes productive intermediate targets reliably.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Subgoal rejection drill

Offer several candidate intermediate targets, some irrelevant.

Why it helps. Ask which target actually reduces the remaining problem.

How to use it. This develops prioritisation.

Success condition. Success means the learner rejects mathematically valid but strategically useless calculations.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Continue-from-wrong-start drill

Give a plausible but inefficient start and ask whether to repair, switch or continue.

Why it helps. This trains route evaluation.

How to use it. Require one reason for the decision.

Success condition. Success means the learner does not confuse effort already spent with route quality.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Continue-from-right-start drill

Give a good opening and several possible next moves.

Why it helps. This tests whether the learner can recognise the best continuation.

How to use it. Use changed contexts to prevent memorised sequences.

Success condition. Success means the student sees what the first step has made possible.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Silent tutor drill

Tutor observes without asking guiding questions for a defined interval.

Why it helps. This reveals whether the learner truly owns the middle decisions.

How to use it. Record where the student would normally look for help.

Success condition. Success means longer independent chains.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

One-hint maximum drill

Allow at most one hint in a multi-step problem.

Why it helps. The hint should target the current state, not supply the next formula.

How to use it. This creates pressure to use the hint as information rather than invitation for more help.

Success condition. Success means the learner completes the rest independently.

The drill should be temporary. Once the learner can perform the middle decision reliably, remove the scaffold and return to complete unfamiliar questions.

Twenty worked continuation cases

Geometry chain

Mid-solution problem. A student finds one angle correctly, then stops.

Training plan. Require the angle to be written into the diagram and ask what theorem or triangle now has enough information.

Verification. Verify on a new diagram where the same angle plays a different intermediate role.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Trigonometry chain

Mid-solution problem. A student finds a length but does not see the second triangle.

Training plan. Label the length visibly and identify which remaining triangle now contains enough knowns.

Verification. Verify with a different composite figure.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Coordinate chain

Mid-solution problem. A student finds gradient but cannot reach a line equation.

Training plan. Use a method-handover sentence connecting gradient plus point to line relation.

Verification. Verify on an unseen coordinate problem.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Probability chain

Mid-solution problem. A student builds a tree but cannot answer the final event.

Training plan. Restate the event in terms of branches and decide whether paths should be added or multiplied.

Verification. Verify on a different event structure.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Statistics chain

Mid-solution problem. A student calculates a mean but cannot interpret comparison.

Training plan. State what the mean represents and what decision the question still asks for.

Verification. Verify with a new dataset.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Rates chain

Mid-solution problem. A student calculates one rate but loses units before the second stage.

Training plan. Use a table and unit reset.

Verification. Verify with a multi-stage rate problem.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Mensuration chain

Mid-solution problem. A student calculates one component area but cannot combine the figure.

Training plan. Use a dependency map for component and total quantities.

Verification. Verify on a different composite shape.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Algebra chain

Mid-solution problem. A student sets up a correct equation then becomes lost in manipulation.

Training plan. Use micro-plan and last-clean-line recovery.

Verification. Verify on an equation with similar structure but different coefficients.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Function chain

Mid-solution problem. A student evaluates one function but cannot continue into composition.

Training plan. Write the output as the next input explicitly.

Verification. Verify on a new composition problem.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Calculus chain

Mid-solution problem. A student differentiates but does not connect to stationary points.

Training plan. Require a statement of what derivative zero means in this context.

Verification. Verify on a new optimization or graph question.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Modeling chain

Mid-solution problem. Variables are defined but no governing equation appears.

Training plan. Use target ladder and relationship inventory.

Verification. Verify on a different word problem.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Proof chain

Mid-solution problem. A useful fact is found but the proof stops.

Training plan. Use backward bridge from the target and compare with the current facts.

Verification. Verify on a changed proof statement.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Bounds chain

Mid-solution problem. Lower and upper bounds are written but combined incorrectly.

Training plan. Use monotonic reasoning and ask how each variable affects the target expression.

Verification. Verify on another derived quantity.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

A-Math algebra chain

Mid-solution problem. Conceptual setup is correct but algebra overwhelms the continuation.

Training plan. Repair shared algebra separately and reinsert into the advanced question.

Verification. Verify across two A-Math topics.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Multi-part exam chain

Mid-solution problem. Part (a) is correct but the result is not reused.

Training plan. Mark reusable outputs and scan subsequent parts before starting fresh.

Verification. Verify on another linked paper question.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Long word problem chain

Mid-solution problem. The first relationship is found but the learner cannot create the second.

Training plan. Use a table or diagram to update all quantities after the first calculation.

Verification. Verify in a different context.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Graph-to-equation chain

Mid-solution problem. A point or gradient is read correctly but not translated algebraically.

Training plan. Use representation-switch practice.

Verification. Verify with an unfamiliar graph.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Equation-to-graph chain

Mid-solution problem. An equation is simplified but the learner cannot predict graph features.

Training plan. Ask what intercepts, gradient or shape now follow.

Verification. Verify on another equation.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Sequence chain

Mid-solution problem. A pattern is recognised but nth-term or later use stalls.

Training plan. State the general structure explicitly before substitution.

Verification. Verify with a new sequence.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Recovery-after-error chain

Mid-solution problem. One middle error is found but the student restarts the whole problem.

Training plan. Use dependency map to preserve unaffected work.

Verification. Verify with a planted local error.

The important measure is not only whether the final answer becomes correct. It is whether the learner now carries the solution through the middle with fewer prompts and more accurate state updates.

Continuation metrics

A tutor can track a small set of signals: number of prompts after the first step, time to identify the next subgoal, first point of failure, number of full restarts, ability to preserve valid intermediate work, and whether the learner can continue a changed problem after a delay.

The strongest progress signal is a longer independent chain. A student who once needed a prompt after every line may begin to need one after a major handover, then only on unfamiliar branches, and eventually none on routine multi-step work.

Part III — Topic-Specific Halfway Stalls

Fifty topic-specific continuation guides

Linear equations

Typical halfway stall. The learner starts rearranging correctly but gets lost after fractions or brackets appear.

Continuation move. Use the last-clean-line method and make the local target explicit: isolate the variable while preserving equality.

Verification. Verify on equations where the complexity appears in different positions.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Simultaneous equations

Typical halfway stall. The learner eliminates one variable but does not know how to use the reduced equation.

Continuation move. State that the elimination has created a single-variable equation and solve that subgoal before back-substitution.

Verification. Verify on a pair where substitution is the better first method.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Quadratics

Typical halfway stall. The learner reaches factorised form but does not connect it to roots, or reaches the quadratic formula but mishandles substitution.

Continuation move. Make the handover explicit: product equals zero leads to factor roots; formula setup leads to two candidate values.

Verification. Verify on mixed quadratic forms.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Inequalities

Typical halfway stall. The learner manipulates correctly until a negative multiplication or division appears.

Continuation move. Pause at the sign-sensitive step and restate the order relation before continuing.

Verification. Verify on inequalities with the negative step in different positions.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Indices

Typical halfway stall. The learner applies one law then cannot simplify the new expression.

Continuation move. Ask what common base or exponent structure now exists after the first transformation.

Verification. Verify with varied exponent forms.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Surds

Typical halfway stall. The learner simplifies one radical but cannot combine the remaining terms.

Continuation move. Use factor structure and like-surd recognition as the next state update.

Verification. Verify with changed coefficients and denominators.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Functions

Typical halfway stall. The learner evaluates an inner function but forgets that the output becomes the next input.

Continuation move. Write the intermediate output explicitly and feed it into the outer function.

Verification. Verify on several compositions and inverse chains.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Graphs

Typical halfway stall. The learner reads a point or gradient but cannot use it later.

Continuation move. Translate the graph feature into an algebraic statement immediately.

Verification. Verify on a new graph where the feature supports a different final target.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Coordinate geometry

Typical halfway stall. The learner finds gradient but stalls before the line equation or intersection.

Continuation move. Treat gradient as an intermediate object and combine it with a known point or second line.

Verification. Verify across several coordinate goals.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Sequences

Typical halfway stall. The learner identifies a pattern but stalls when asked for a general term or distant value.

Continuation move. Convert the observed pattern into an explicit rule before further substitution.

Verification. Verify with a sequence requiring a different representation.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Ratio

Typical halfway stall. The learner forms a ratio correctly but stalls when the total, scale or unknown share enters.

Continuation move. Translate the ratio into units or parts, then identify which quantity the next step requires.

Verification. Verify in sharing, scale and mixture contexts.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Rates

Typical halfway stall. The learner calculates one rate but cannot continue across a second interval or unit system.

Continuation move. Update the table with the new value and standardise units before the next relation.

Verification. Verify in multi-stage rate problems.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Percentages

Typical halfway stall. The learner finds one percentage change but stalls on repeated or reverse change.

Continuation move. State the new base explicitly after the first change.

Verification. Verify with different original and final quantities.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Direct proportion

Typical halfway stall. The learner identifies direct proportion but cannot use the constant later.

Continuation move. Treat the constant as the bridge between one known pair and the unknown pair.

Verification. Verify with table and equation forms.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Inverse proportion

Typical halfway stall. The learner identifies inverse proportion but loses the constant-product relation halfway.

Continuation move. Keep the invariant product visible and substitute new values into the same relation.

Verification. Verify in changed contexts.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Pythagoras

Typical halfway stall. The learner finds one side but cannot see the second right triangle.

Continuation move. Insert the new side into the diagram and inspect which triangle is now solvable.

Verification. Verify in composite geometry.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Trigonometric ratios

Typical halfway stall. The learner finds one length or angle but does not relabel for the next triangle.

Continuation move. Update the diagram and redefine opposite/adjacent relative to the next reference angle.

Verification. Verify on rotated diagrams.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Sine rule

Typical halfway stall. The learner gets one angle but cannot continue to a second side or area.

Continuation move. Write the new opposite pair explicitly and inspect which formula now has enough information.

Verification. Verify on multi-step triangle problems.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Cosine rule

Typical halfway stall. The learner finds one side but cannot connect it to another relation.

Continuation move. Update known side set and reassess whether sine rule, area or another cosine step is now available.

Verification. Verify on changed triangle structures.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Similarity

Typical halfway stall. The learner identifies similar triangles but loses correspondence during the ratio chain.

Continuation move. Write corresponding vertices or sides once and keep the order fixed.

Verification. Verify on differently oriented diagrams.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Congruence

Typical halfway stall. The learner establishes congruence but does not use corresponding parts afterward.

Continuation move. State which equal angles or lengths now become available because of congruence.

Verification. Verify on multi-part geometry.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Circle geometry

Typical halfway stall. The learner derives one angle but stalls before the next theorem.

Continuation move. Update the circle diagram and ask what new chord, tangent or cyclic relation is now complete.

Verification. Verify on unfamiliar circle diagrams.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Bearings

Typical halfway stall. The learner constructs the first bearing but loses orientation later.

Continuation move. Redraw north lines at each relevant point and carry the derived angle into the next stage.

Verification. Verify on multi-leg journeys.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Mensuration

Typical halfway stall. The learner finds one area or volume but cannot combine composite parts.

Continuation move. Use a component-total dependency map.

Verification. Verify on subtraction and addition composites.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Transformations

Typical halfway stall. The learner identifies one transformation but stalls on a sequence of two.

Continuation move. Apply the first transformation fully, record the new state, then interpret the second relative to the transformed image.

Verification. Verify on combined transformations.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Vectors

Typical halfway stall. The learner writes one vector relation but cannot reach the target vector.

Continuation move. Treat each vector result as a new route segment and build the target path explicitly.

Verification. Verify on changed point configurations.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Matrices

Typical halfway stall. The learner multiplies one matrix correctly but cannot interpret or continue.

Continuation move. State what the resulting matrix represents and what operation remains.

Verification. Verify in multi-stage matrix tasks.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Sets

Typical halfway stall. The learner shades or identifies one region but stalls on a compound condition.

Continuation move. Translate the next verbal condition into union, intersection or complement relative to the current region.

Verification. Verify on new Venn structures.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Probability

Typical halfway stall. The learner finds one branch probability but stalls on the final event.

Continuation move. Restate the event as a combination of branches and choose addition, multiplication or complement accordingly.

Verification. Verify on trees and tables.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Mean

Typical halfway stall. The learner calculates a mean but cannot compare or infer.

Continuation move. Interpret the mean in context and restate the final decision question.

Verification. Verify with another dataset.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Median and quartiles

Typical halfway stall. The learner locates one positional statistic but stalls on spread or comparison.

Continuation move. Treat the found value as one boundary in the distribution and connect it to the comparison target.

Verification. Verify on cumulative frequency.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Histogram

Typical halfway stall. The learner calculates frequency density but does not convert back to frequency or comparison.

Continuation move. Use area meaning to update the data state.

Verification. Verify with unequal class widths.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Scatter plots

Typical halfway stall. The learner identifies correlation but stalls on prediction or interpretation.

Continuation move. State what the trend supports and what it does not justify.

Verification. Verify in a changed context.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Bounds

Typical halfway stall. The learner finds individual bounds but cannot combine them for a final expression.

Continuation move. Ask which combination maximises or minimises the target.

Verification. Verify on quotients, products and derived measures.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Speed-time graphs

Typical halfway stall. The learner finds one area or gradient but cannot continue across several intervals.

Continuation move. Break the graph into segments and keep a running total or state table.

Verification. Verify on multi-stage motion.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Distance-time graphs

Typical halfway stall. The learner reads one slope but stalls on overall journey interpretation.

Continuation move. Update the story of motion segment by segment.

Verification. Verify on changed graph shapes.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Financial mathematics

Typical halfway stall. The learner applies one rate period but loses the updated principal.

Continuation move. Write the new amount as the next base before applying subsequent change.

Verification. Verify with multi-period changes.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Algebraic fractions

Typical halfway stall. The learner factors correctly but stalls on cancellation or equation solving.

Continuation move. State what factorisation has unlocked and keep restrictions visible.

Verification. Verify across simplification and equations.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Polynomials

Typical halfway stall. The learner uses factor theorem once but cannot complete factorisation or solve.

Continuation move. Use the discovered factor as a bridge to division or lower-degree solving.

Verification. Verify on different polynomial targets.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Logarithms

Typical halfway stall. The learner combines logs but stalls on isolating the unknown.

Continuation move. Convert the new expression into exponential or simpler algebraic form deliberately.

Verification. Verify across varied bases.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Exponentials

Typical halfway stall. The learner finds a common base or takes logs but loses the route after that.

Continuation move. State the new equation in the simpler variable/exponent form and solve that subgoal.

Verification. Verify on growth and algebra contexts.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Differentiation

Typical halfway stall. The derivative is found but the final application stalls.

Continuation move. Translate derivative into gradient, stationary-point or rate language immediately.

Verification. Verify across tangent and optimization questions.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Integration

Typical halfway stall. The antiderivative is found but limits, area or interpretation are not completed.

Continuation move. State what the integral represents in the question and apply the remaining condition.

Verification. Verify on different applications.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Kinematics

Typical halfway stall. One variable is found but sign convention or next equation becomes unclear.

Continuation move. Update the known-variable list after every stage and keep direction convention visible.

Verification. Verify on multi-stage motion.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Optimization

Typical halfway stall. The objective is defined but variable reduction stalls.

Continuation move. Use the constraint to eliminate one variable before calculus.

Verification. Verify across geometric and applied models.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Related rates

Typical halfway stall. The relationship is built but differentiation or substitution order breaks.

Continuation move. Differentiate first with respect to time, then substitute the instant values.

Verification. Verify with changed related quantities.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Proof

Typical halfway stall. One valid statement is established but no bridge to target appears.

Continuation move. Use backward bridge and search for an intermediate statement shared by both chains.

Verification. Verify on a new proof.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Mathematical modelling

Typical halfway stall. A model begins correctly but the learner loses the connection between algebra and context.

Continuation move. At each major result, translate back to what the variable means in the situation.

Verification. Verify in unfamiliar word problems.

The goal is not to memorise a fixed chain for this topic. The student should learn to read the changed state after each major result and decide what the new information makes possible.

Part III handoff

Topic knowledge explains the local tools; continuation skill explains how those tools are linked. The final layer will define learner profiles, parent/tutor responses, final verification standards and the evidence required before this halfway-stall problem can move to maintenance.

Part IV — Independence, Prompt Fading and Final Verification

Thirty learner profiles and the right response

Learner who always asks ‘what next?’

Pattern. The student may know each method but has outsourced subgoal generation.

Tutor response. Tutor response: ask what changed after the last step, then remain silent.

Parent/home response. Home response: avoid supplying the next formula; ask the student to identify the current knowns.

Progress signal. Progress: fewer sequential prompts after the first correct move.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who talks through every step

Pattern. Verbal explanation may help, but can become a substitute for concise mathematical action.

Tutor response. Tutor response: restrict explanation to handovers and decisions.

Parent/home response. Home response: praise clear working rather than long commentary.

Progress signal. Progress: reasoning remains visible while execution becomes more efficient.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who barely writes

Pattern. Mental load may be too high for long chains.

Tutor response. Tutor response: require intermediate values, variable definitions and key transformations on paper.

Parent/home response. Home response: do not demand decorative neatness, only functional clarity.

Progress signal. Progress: fewer lost values and fewer full restarts.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who writes everything

Pattern. Overdocumentation can create time pressure and obscure the key state.

Tutor response. Tutor response: identify which steps can be compressed safely.

Parent/home response. Home response: focus on whether working is auditable, not maximal.

Progress signal. Progress: shorter solutions with stable accuracy.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who never updates diagrams

Pattern. Visual state remains frozen at the start.

Tutor response. Tutor response: require derived facts to be added immediately.

Parent/home response. Home response: ask what new information belongs on the diagram now.

Progress signal. Progress: later methods are selected from the updated representation.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who never updates equations

Pattern. A valid setup remains static while new values appear elsewhere.

Tutor response. Tutor response: substitute and simplify in a clean new line.

Parent/home response. Home response: ask which part of the equation has changed.

Progress signal. Progress: algebraic state remains aligned with problem state.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who gets trapped by one branch

Pattern. Sunk-cost persistence is high.

Tutor response. Tutor response: compare progress per minute across alternative routes.

Parent/home response. Home response: reinforce that changing a plan is mathematical judgement, not failure.

Progress signal. Progress: earlier branch switching with less wasted work.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who abandons too quickly

Pattern. Productive struggle window is too short.

Tutor response. Tutor response: require one justified continuation move before leaving.

Parent/home response. Home response: do not rescue instantly.

Progress signal. Progress: more solvable questions are carried through without long stalls.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who restarts repeatedly

Pattern. Local uncertainty becomes global distrust.

Tutor response. Tutor response: mark last-clean-line and preserve confirmed facts.

Parent/home response. Home response: discourage erasing whole solutions.

Progress signal. Progress: corrections become local.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who refuses to restart

Pattern. The student persists through contaminated working.

Tutor response. Tutor response: teach evidence for invalidating a branch.

Parent/home response. Home response: ask whether the current route is still producing meaningful structure.

Progress signal. Progress: selective restart replaces rigid persistence.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner with strong concepts but weak algebra

Pattern. Mid-solution execution is the bottleneck.

Tutor response. Tutor response: repair shared algebra separately, then reinsert into multi-step contexts.

Parent/home response. Home response: avoid re-teaching the whole topic when the concept is already secure.

Progress signal. Progress: fewer algebraic stalls across different subjects.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner with strong algebra but weak planning

Pattern. Execution is fluent but target generation is weak.

Tutor response. Tutor response: use target ladders and missing-middle tasks.

Parent/home response. Home response: ask what intermediate quantity would make the final step possible.

Progress signal. Progress: more productive subgoals without hints.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner with weak working memory

Pattern. Several intermediate quantities exceed mental capacity.

Tutor response. Tutor response: externalise state in tables, labels and diagrams.

Parent/home response. Home response: accept concise written support rather than demanding mental calculation.

Progress signal. Progress: lower transcription and value-confusion errors.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who loses units

Pattern. Meaning decays during the chain.

Tutor response. Tutor response: use units at handover points and as a check.

Parent/home response. Home response: ask what the intermediate quantity represents.

Progress signal. Progress: fewer impossible units and setup errors.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who rounds too early

Pattern. Precision loss contaminates later values.

Tutor response. Tutor response: mark exact versus approximate state explicitly.

Parent/home response. Home response: ask whether rounding is required now or only at the end.

Progress signal. Progress: fewer accumulated numerical discrepancies.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who cannot use previous subparts

Pattern. Multi-part dependency is not recognised.

Tutor response. Tutor response: mark reusable outputs and scan later parts.

Parent/home response. Home response: ask what part (a) has now made available.

Progress signal. Progress: later subparts begin from earlier results naturally.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who depends on worked examples

Pattern. The route is memorised as a sequence rather than built from state.

Tutor response. Tutor response: change numbers, context and stopping point of partial solutions.

Parent/home response. Home response: hide the model before the learner continues.

Progress signal. Progress: same structure is completed from unfamiliar partial states.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who panics when a route diverges

Pattern. Plan revision feels like failure.

Tutor response. Tutor response: model branch comparison and last-clean-line recovery.

Parent/home response. Home response: normalise trying, evaluating and changing routes.

Progress signal. Progress: calmer local revision of strategy.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who cannot explain what is now known

Pattern. State interpretation is weak.

Tutor response. Tutor response: require one concise state sentence after major steps.

Parent/home response. Home response: ask what the new value means, not what the next formula is.

Progress signal. Progress: more accurate use of intermediate results.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who can state subgoal but not execute

Pattern. Planning exceeds procedural fluency.

Tutor response. Tutor response: isolate the exact method required by the subgoal.

Parent/home response. Home response: avoid adding more planning discussion when the missing piece is execution.

Progress signal. Progress: subgoal statements convert into correct written moves.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Learner who executes but never plans

Pattern. Procedure habit replaces route control.

Tutor response. Tutor response: insert micro-plans only before branch points.

Parent/home response. Home response: ask why the next calculation matters.

Progress signal. Progress: fewer irrelevant calculations.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Strong learner

Pattern. The main risk is inefficient branching or overcomplicated methods.

Tutor response. Tutor response: compare alternative routes, cost and elegance.

Parent/home response. Home response: avoid adding routine volume merely because the child is strong.

Progress signal. Progress: shorter, more robust multi-step solutions.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Weak learner

Pattern. Too many simultaneous gaps can make long questions unreadable.

Tutor response. Tutor response: reduce chain length and repair high-leverage prerequisites.

Parent/home response. Home response: focus on one bridge at a time.

Progress signal. Progress: accessible chain length grows gradually.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Slow learner

Pattern. Continuation decisions may be accurate but delayed.

Tutor response. Tutor response: time next-target selection separately from execution.

Parent/home response. Home response: avoid blanket ‘be faster’ messages.

Progress signal. Progress: faster subgoal generation with stable accuracy.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Fast learner

Pattern. Impulsive continuation can produce wrong branches.

Tutor response. Tutor response: require one structural reason at key handovers.

Parent/home response. Home response: praise correct route choice, not only quick completion.

Progress signal. Progress: fewer branch errors without major pace loss.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Anxious learner

Pattern. A halfway stall may trigger a global collapse.

Tutor response. Tutor response: use state snapshots and one local next target.

Parent/home response. Home response: keep discussion local to the current problem.

Progress signal. Progress: difficult middle moments create less downstream disruption.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Prompt-dependent learner

Pattern. Tutor cues are part of the route.

Tutor response. Tutor response: protect silent first attempts and record prompt count.

Parent/home response. Home response: avoid asking leading questions.

Progress signal. Progress: longer uninterrupted independent chains.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

A-Math learner

Pattern. Advanced topics magnify algebraic and planning handovers.

Tutor response. Tutor response: separate shared algebra from advanced concept transitions.

Parent/home response. Home response: coordinate E-Math and A-Math workload.

Progress signal. Progress: fewer stalls caused by common algebraic dependencies.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Exam-phase learner

Pattern. Time makes continuation efficiency more important.

Tutor response. Tutor response: use timed partial-solution drills and full-paper verification.

Parent/home response. Home response: avoid teaching new elaborate strategies late.

Progress signal. Progress: shorter mid-solution stalls under paper conditions.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Post-prelim learner

Pattern. The script may show repeated halfway breakdowns.

Tutor response. Tutor response: classify the first failed handover across several questions.

Parent/home response. Home response: focus on recurring mechanism rather than chapter labels.

Progress signal. Progress: later papers show longer intact solution chains.

The profile should not become a permanent label. The purpose is to identify which support is currently carrying the middle of the solution and then reduce that support as the learner takes over the decision.

Twenty tutor and parent guardrails

Do not give the next formula first

Ask what is now known and what the final step still requires.

This preserves the learner’s responsibility for subgoal generation.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not name the chapter as the hint

A chapter label can reveal the method family too directly.

Use relationship or target questions instead.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not correct the first messy line instantly

Observe whether the learner can detect the issue from plausibility or structure.

Intervene when error propagation would otherwise dominate the task.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not rescue every pause

A short pause can be productive planning time.

Distinguish quiet thinking from repeated unproductive action.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not let productive struggle become prolonged helplessness

Use a bounded attempt and then a minimal cue.

Return later to no-cue verification.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not reward only final answers

Notice state updates, valid subgoals and good branch changes.

These are the skills that make long questions independent.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not penalise route revision

Changing a plan after new evidence can be expert behaviour.

Evaluate whether the revision is justified.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not insist on one fixed route

Multiple valid paths may exist.

Compare efficiency and clarity instead.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not make every long question timed immediately

First establish accurate continuation.

Add time pressure after the chain is structurally reliable.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not remove all scaffolds at once

Fade one layer: target prompts, then representation prompts, then silence.

This gives cleaner evidence of what the learner now owns.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not leave scaffolds forever

A support that is still required on familiar questions is part of the problem.

Schedule explicit support-fading.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not overpractice one exact chain

Change stopping points, numbers, contexts and representations.

The goal is continuation skill, not item memory.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not assume every halfway stall is knowledge failure

Compare untimed, supported and changed-condition work.

The issue may be planning, representation or handover.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not assume every halfway stall is exam technique

If the learner cannot continue untimed, a real knowledge or prerequisite gap may exist.

Diagnose before training paper control.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not hide the final target

Long questions become easier to navigate when the destination remains visible.

Use target restatement at major checkpoints.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not hide intermediate meaning

A value should be labelled by what it represents.

This turns numbers into usable state information.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not let units become decoration

Units can guide valid relationships and expose mistakes.

Use them selectively through the chain.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not let calculator output become a black box

Estimate and interpret important results before using them later.

This protects the chain from mechanical contamination.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not let one local error invalidate known-good work

Trace dependencies and preserve unaffected reasoning.

This reduces unnecessary restarts.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Do not close the target after one successful long question

Use delayed, changed and independent verification.

One familiar success may not generalise.

Use the guardrail to preserve student agency. The adult’s job is to reveal the decision process long enough to train it, not to become a permanent bridge between every step.

Ten verification standards

Changed starting point

Test. Give a partial solution ending at a different stage.

Evidence. The learner should reconstruct the next target without relying on a memorised opening.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Changed surface

Test. Change wording, diagram orientation, numbers or context.

Evidence. The continuation skill should survive without exact item familiarity.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Different topic

Test. Test the same state-update routine in another mathematical domain.

Evidence. Cross-topic transfer is strong evidence of genuine process learning.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Delayed attempt

Test. Return after several days.

Evidence. The learner should still generate subgoals without recent tutor cues.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

No-hint attempt

Test. Tutor remains silent after the first step.

Evidence. The learner should carry a meaningful chain independently.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Timed continuation

Test. Measure time to identify next target and method.

Evidence. Speed should improve only after accuracy is stable.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Full-question attempt

Test. Remove the partial-solution scaffold.

Evidence. The learner should build both opening and middle independently.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Paper-section attempt

Test. Place several multi-step questions among routine ones.

Evidence. Continuation should coexist with paper pacing.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Full-paper attempt

Test. Observe mid-solution stalls under realistic conditions.

Evidence. The skill is examination-ready when long questions no longer trigger repeated tutor-like internal dependence.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Maintenance check

Test. Reduce dedicated drills and monitor through normal papers.

Evidence. Reactivate only if the same halfway pattern returns.

A target can move from active repair to fragile, secure and maintenance only when the learner’s continuation remains reliable with less support and more realistic problem conditions.

Final closure: the learner must own the middle

A student who starts well but repeatedly gets stuck halfway does not necessarily need more formulas or more chapter notes. The learner often needs practice in reading what each step has changed, creating intermediate targets, handing one method into the next, protecting useful work and revising a route when new evidence appears.

The most convincing outcome is not one perfectly coached long question. It is a pattern across unfamiliar work: fewer prompts, fewer full restarts, later first-error points, better use of intermediate results and more complete solution chains. Once those signals persist under mixed and timed conditions, the middle of the solution has become part of the student’s independent Mathematics.

Additional transfer cases

Transfer check — Learner who always asks ‘what next?’

Use a fresh multi-step question in which the student may know each method but has outsourced subgoal generation. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: ask what changed after the last step, then remain silent. Then repeat on a changed problem after a delay.

If progress: fewer sequential prompts after the first correct move., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who talks through every step

Use a fresh multi-step question in which verbal explanation may help, but can become a substitute for concise mathematical action. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: restrict explanation to handovers and decisions. Then repeat on a changed problem after a delay.

If progress: reasoning remains visible while execution becomes more efficient., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who barely writes

Use a fresh multi-step question in which mental load may be too high for long chains. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: require intermediate values, variable definitions and key transformations on paper. Then repeat on a changed problem after a delay.

If progress: fewer lost values and fewer full restarts., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who writes everything

Use a fresh multi-step question in which overdocumentation can create time pressure and obscure the key state. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: identify which steps can be compressed safely. Then repeat on a changed problem after a delay.

If progress: shorter solutions with stable accuracy., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who never updates diagrams

Use a fresh multi-step question in which visual state remains frozen at the start. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: require derived facts to be added immediately. Then repeat on a changed problem after a delay.

If progress: later methods are selected from the updated representation., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who never updates equations

Use a fresh multi-step question in which a valid setup remains static while new values appear elsewhere. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: substitute and simplify in a clean new line. Then repeat on a changed problem after a delay.

If progress: algebraic state remains aligned with problem state., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who gets trapped by one branch

Use a fresh multi-step question in which sunk-cost persistence is high. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: compare progress per minute across alternative routes. Then repeat on a changed problem after a delay.

If progress: earlier branch switching with less wasted work., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who abandons too quickly

Use a fresh multi-step question in which productive struggle window is too short. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: require one justified continuation move before leaving. Then repeat on a changed problem after a delay.

If progress: more solvable questions are carried through without long stalls., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who restarts repeatedly

Use a fresh multi-step question in which local uncertainty becomes global distrust. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: mark last-clean-line and preserve confirmed facts. Then repeat on a changed problem after a delay.

If progress: corrections become local., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who refuses to restart

Use a fresh multi-step question in which the student persists through contaminated working. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: teach evidence for invalidating a branch. Then repeat on a changed problem after a delay.

If progress: selective restart replaces rigid persistence., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner with strong concepts but weak algebra

Use a fresh multi-step question in which mid-solution execution is the bottleneck. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: repair shared algebra separately, then reinsert into multi-step contexts. Then repeat on a changed problem after a delay.

If progress: fewer algebraic stalls across different subjects., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner with strong algebra but weak planning

Use a fresh multi-step question in which execution is fluent but target generation is weak. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: use target ladders and missing-middle tasks. Then repeat on a changed problem after a delay.

If progress: more productive subgoals without hints., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner with weak working memory

Use a fresh multi-step question in which several intermediate quantities exceed mental capacity. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: externalise state in tables, labels and diagrams. Then repeat on a changed problem after a delay.

If progress: lower transcription and value-confusion errors., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who loses units

Use a fresh multi-step question in which meaning decays during the chain. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: use units at handover points and as a check. Then repeat on a changed problem after a delay.

If progress: fewer impossible units and setup errors., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who rounds too early

Use a fresh multi-step question in which precision loss contaminates later values. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: mark exact versus approximate state explicitly. Then repeat on a changed problem after a delay.

If progress: fewer accumulated numerical discrepancies., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who cannot use previous subparts

Use a fresh multi-step question in which multi-part dependency is not recognised. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: mark reusable outputs and scan later parts. Then repeat on a changed problem after a delay.

If progress: later subparts begin from earlier results naturally., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who depends on worked examples

Use a fresh multi-step question in which the route is memorised as a sequence rather than built from state. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: change numbers, context and stopping point of partial solutions. Then repeat on a changed problem after a delay.

If progress: same structure is completed from unfamiliar partial states., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Transfer check — Learner who panics when a route diverges

Use a fresh multi-step question in which plan revision feels like failure. could appear. Do not tell the learner that this is the mechanism being tested. The first evidence should come from the student’s own working.

If the learner needs the same support, keep the target active and narrow the task. Tutor response: model branch comparison and last-clean-line recovery. Then repeat on a changed problem after a delay.

If progress: calmer local revision of strategy., reduce the scaffold and allow ordinary homework, past papers or mocks to provide future maintenance. The purpose is to make the process disappear into competent independent solving.

Final Mid-Solution Verification Appendix

Can the learner state what changed after the first step?

A correct first move is not enough; the learner should identify the new information now available.

If the answer remains vague, keep state-update practice active.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner generate a next subgoal without a hint?

The next target should arise from the final goal and current knowns.

Independent subgoal generation is the core sign that tutor dependence is falling.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner use an intermediate value correctly later?

A derived length, rate, gradient or probability should remain visible and meaningful.

If values are repeatedly lost, strengthen labeling and dependency tracking.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner continue from someone else’s valid partial solution?

This tests whether the learner can read a mathematical state rather than memorise one personal route.

Use several different stopping points.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner detect an unproductive branch?

The student should notice when repeated moves produce no new structure.

Branch revision should be evidence-based rather than emotional.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner return to the last clean line after an error?

Local recovery preserves useful work and reduces full restarts.

This is especially important under time.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner change representation when the current one is unhelpful?

A diagram, table, graph or equation should be chosen because it exposes structure.

Representation change is a continuation skill, not a decorative extra.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner preserve sign, unit and precision state?

Long chains often fail because small local states are forgotten.

Selective checkpoints should reduce contamination.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner hand one method into another?

Output from one method should become purposeful input for the next.

This is the essence of integrated multi-step work.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

Can the learner finish a changed long question with fewer prompts?

The strongest evidence is a longer independent chain on unfamiliar work.

Once this remains stable under mixed and timed conditions, dedicated halfway-stall drills can move to maintenance.

Use a fresh problem rather than the original corrected item. The learner should demonstrate the continuation process from the new mathematical state, not from memory of a previous solution sequence.

If performance is secure, reduce direct scaffolding. If it fails, return to the narrowest missing bridge—state update, subgoal, handover, execution or branch monitoring—and retest after a delay.

The final aim is simple: the student should not need an adult to supply the middle of the solution. Each correct step should generate enough information for the learner to decide what comes next.