Wait, what? If a system works now, how do we know which part is actually necessary?

One powerful scientific move is to imagine a controlled change: remove one part, block one pathway, replace one material, or hold one factor constant while changing another. Then ask what should happen if our model is correct.

This preserved Hougang Primary 5 Science URL now owns one precise job: counterfactual intervention and causal testing. The old duplicated tuition advertisement, locality claims, grade promises and unrelated image stack have been removed.

This page is distinct from the existing Hougang P5 guides on mechanisms, bottlenecks, trade-offs, feedback loops, flow and scale. Those pages explain how systems are organised. This page asks:

If I change one part of the system and keep the rest relevantly the same, what should happen—and what would that tell me about the part’s causal role?

Counterfactual means “what would happen if…”

A counterfactual asks about a situation that is not the current one.

• What if this pathway were blocked?
• What if this component were removed?
• What if this material were replaced?
• What if this input doubled?
• What if this organism disappeared from the food web?

The point is not imaginative storytelling. The point is to use the scientific model to predict how the system changes under a controlled alternative.

Hold the other relevant conditions constant

Counterfactual reasoning becomes useful when the change is isolated.

If we remove a leaf and simultaneously change light, water and temperature, we cannot tell which change produced the outcome.

A cleaner thought experiment is:

Imagine everything relevant stays the same except X.

This is the reasoning core of controlled comparison.

Remove one part to test necessity

If a component is necessary for a system function, removing it should disrupt that function under the stated conditions.

Examples of the reasoning pattern:

• remove a required circuit connection → path becomes incomplete → current cannot pass through the intended complete path;
• block a transport pathway → movement to downstream parts is reduced or stopped;
• remove a required environmental input → the dependent process decreases or stops.

The exact Science depends on the system. The reasoning question is stable:

Does the function fail when this part is absent?

Necessary does not mean sufficient

A part can be required without being enough by itself.

A battery may be necessary for a simple circuit, but a battery alone is not sufficient if the path is incomplete. Water may be necessary for plant processes, but water alone does not guarantee growth if other required conditions are absent.

The counterfactual tests are different:

• Necessity test: remove X. Does the function fail?
• Sufficiency test: provide X alone. Is the function guaranteed?

Primary 5 students do not need formal logic terminology to benefit from the distinction.

Replace one component to test which property matters

Replacement questions are powerful because they can isolate a property.

Suppose a component is replaced with another of the same shape but a different material.

If the system behaviour changes, the material property becomes a candidate explanation.

Ask:

• What stayed the same?
• What property changed?
• What outcome changed?
• Which mechanism links the property to the outcome?

Replacement reasoning is especially useful in materials and design questions.

Block a pathway to test flow

Flow systems are naturally tested by blockage.

A general pattern is:

pathway blocked → less or no transfer through that route → downstream process changes

The student should identify:

• what is moving or being transferred;
• through which pathway;
• which downstream part depends on it;
• whether an alternative route exists.

This connects counterfactual reasoning to flow and bottleneck reasoning without duplicating them.

Alternative pathways change the prediction

If the system has two routes, blocking one may reduce the effect rather than stop it completely.

Ask:

• Is this the only route?
• Is there a backup or parallel pathway?
• Can another component compensate?
• Does the question provide evidence that the alternative route is sufficient?

Counterfactual predictions must respect the topology of the actual system.

Direct effect versus indirect effect

A changed component may affect one process directly and another only later.

Example structure:

• remove component X;
• process A decreases directly;
• output from A falls;
• process B receives less input;
• whole-system outcome changes later.

Students should separate:

• direct consequence: first process affected by the intervention;
• indirect consequence: later effect propagated through dependencies.

This prevents causal leaps.

Immediate effect versus delayed effect

Some counterfactual changes do not produce the final outcome instantly.

Ask:

• What changes immediately?
• What requires time?
• What stored resource may temporarily hide the effect?
• When should the final consequence become observable?

Time ordering matters in systems with stores, reserves or multiple stages.

The counterfactual should preserve the model boundary

Do not imagine impossible conditions that break the syllabus model itself.

A useful counterfactual changes one relevant condition while leaving the rest of the model meaningful.

Ask:

• Is this intervention physically or biologically meaningful?
• Does the model still apply after the change?
• Am I changing so many things that the comparison loses interpretability?

A thought experiment should simplify causality, not destroy it.

Counterfactual reasoning in circuits

Circuits support clear intervention questions.

• What if one connection is removed?
• What if a working bulb is replaced by a broken one?
• What if a parallel branch is opened or closed?
• What if the power source is removed?

Each intervention should be traced through connectivity and the circuit model rather than through memorised visual patterns.

Counterfactual reasoning in plants and body systems

Ask what happens if one input, structure or pathway becomes unavailable.

The learner should trace:

• which process depends on the part;
• what output decreases;
• which later process depends on that output;
• what whole-system effect follows.

Do not jump straight from “part removed” to “organism dies”. The scientific value lies in the dependency chain and in recognising possible compensation.

Counterfactual reasoning in food webs

Removing one population can produce cascading effects.

But the correct prediction depends on the actual web:

• What eats the removed organism?
• What did the removed organism eat?
• Are alternative food sources shown?
• Which effect is direct?
• Which later effect depends on a population change first occurring?

Counterfactual food-web reasoning should be one edge at a time.

The intervention matrix

Intervention	Direct effect	Indirect effect	Alternative pathway?	Prediction
Remove X	?	?	?	?
Block Y	?	?	?	?
Replace Z	?	?	?	?

This scaffold helps the learner avoid one-step predictions in multi-part systems.

The reversal counterfactual

If removing X reduces Y, does adding more X necessarily increase Y?

Not always.

X may be necessary only up to a sufficient level. Another factor may then become limiting.

This is a useful check against symmetric reasoning:

“Removing hurts” does not automatically mean “adding more always helps”.

Counterfactuals should respect bottlenecks and trade-offs.

Counterfactual reasoning can test competing explanations

Suppose two explanations fit the current observation.

Ask:

• If Explanation A were true, what would happen after intervention X?
• If Explanation B were true, what would happen instead?
• Which intervention would produce different predictions?

A discriminating intervention turns “both sound possible” into a testable difference.

Counterfactual reasoning can expose hidden assumptions

When a prediction changes unexpectedly, ask which assumption the original model depended on.

For example:

• Did we assume there was only one pathway?
• Did we assume another resource was unlimited?
• Did we assume all samples were identical?
• Did we assume the effect was immediate?

Interventions are good assumption detectors.

Do not invent consequences beyond the evidence

A counterfactual answer should follow the supplied system, not every real-world possibility.

If a food web shows only selected relationships, use those. If a circuit diagram defines only selected connections, do not invent hidden wiring. If a plant experiment controls several conditions, do not assume an unmentioned pest or disease.

Good “what if?” reasoning is bounded, not imaginative without limit.

Five Primary 5 counterfactual failure modes

1. Everything-changes thinker

The imagined scenario changes several conditions at once. Repair by isolating one intervention.

2. Necessary-means-sufficient thinker

A required part is assumed to guarantee the whole function. Repair with remove-versus-provide tests.

3. Direct-only thinker

The student stops at the first effect. Repair by tracing downstream dependencies.

4. No-alternative-route thinker

Blocking one route is assumed to stop the entire system. Repair by checking topology and compensation.

5. Reverse-symmetry thinker

If less X hurts, more X is assumed to help forever. Repair with limiting-factor and trade-off checks.

A Phase 4 Primary 5 counterfactual lesson

• Model: map the current system.
• Intervene: remove, replace, block or alter one condition.
• Hold: keep other relevant conditions constant.
• Direct: identify the first affected process.
• Propagate: trace indirect downstream effects.
• Alternate: check for compensating pathways.
• Time: separate immediate and delayed effects.
• Test: ask whether the prediction distinguishes competing explanations.
• Bound: avoid inventing unsupported consequences.
• Reverse: test whether the opposite intervention really produces the opposite result.

Why small groups help counterfactual reasoning

Three students may predict three different outcomes after the same intervention.

• Which student changed more than one condition?
• Who forgot an alternative pathway?
• Who stopped at the direct effect?
• Which prediction depends on an unstated assumption?

The disagreement makes the system model visible.

What parents can practise at home

• Ask “what if we remove this part?”
• Ask what stays the same in the imagined comparison.
• Ask for the first direct effect, then the later indirect effect.
• Ask whether another route can compensate.
• Ask whether adding more of a necessary factor always helps.
• Ask which intervention would distinguish two explanations.

How to tell whether counterfactual reasoning is improving

• One variable is changed at a time.
• Necessary and sufficient conditions are distinguished.
• Direct and indirect effects are separated.
• Alternative pathways are checked.
• Time delays are considered.
• Predictions remain bounded by the provided model.
• Counterfactuals are used to discriminate explanations.
• Reverse interventions are not assumed to be symmetrical.

How this page fits the Hougang Science network

This eduKateSingapore page owns counterfactual intervention and causal testing. It complements constraints and bottlenecks, feedback loops and cascading effects, trade-offs and competing constraints, and mechanism debugging.

For the complete P3-to-PSLE map, use Hougang Primary Science Learning Library.

Official curriculum reference

The Ministry of Education’s Science Teaching & Learning Syllabus: Primary Three to Six develops prediction, investigation, analysis and explanation within Systems and Interactions. Counterfactual reasoning is used here as an age-appropriate scaffold for testing causal dependencies.

---

Primary 5 Science gets stronger when “what if?” becomes a controlled scientific test. Change one part, hold the rest relevantly steady, trace the direct and indirect effects, check for alternative routes, and ask what the result would reveal about the system’s causal structure.