HSW-0306 · How Studying Works
A learner practises five items from a chapter and leaves the other fifteen untouched.
What should happen to the fifteen?
A familiar warning from memory research is that selective retrieval can make related unpractised material harder to retrieve. That phenomenon is real and already has a place in the eduKateSG estate.
But it is not the whole story.
Under some conditions, retrieving part of a learned set can improve later memory not only for the practised material but also for material that was never directly retrieved.
This article owns that counter-direction: retrieval-induced facilitation. It asks when selective practice can reactivate a broader memory set, strengthen neighbouring material or reset its accessibility, and why students should not assume that selective retrieval always narrows memory or always spreads benefit.
The neighbouring owner Relational Mindset and Retrieval-Induced Forgetting covers the risk that selective retrieval can disadvantage related unpractised material. Retrieval Interference covers competition among similar memories. This page owns the opposite empirical possibility and the study decision it creates.
Quick Answer
A paper published online on 8 September 2026 in Psychonomic Bulletin & Review tested 216 participants learning common and rare Chinese characters. After a selective retrieval phase, recall improved not only for retrieved characters but also for nonretrieved characters. The authors also reported that retrieval restarted forgetting functions, and that the gains were similar for high- and low-prior-knowledge items even though common characters were remembered better overall. See Huang, Cao, Li, Woodman, Zhang and Wang (2026).
That is one recent experimental result, not a universal law that “testing some material teaches the rest.” The wider literature contains robust cases of retrieval-induced forgetting, output interference and cue competition. The useful conclusion is therefore conditional:
Selective retrieval can reshape a memory neighbourhood. Sometimes the neighbours become easier to reach; sometimes they become harder. The direction depends on what retrieval reactivates, competes with, organises and reconnects.
1. Selective Practice Changes More Than the Selected Item
Students often imagine retrieval practice as a local operation:
I tested item A, so only item A changed.
Memory is not always that modular. Retrieving one fact can reactivate its category, context, neighbouring examples, competing alternatives, source episode or associated structure. The retrieval event therefore has the potential to alter access to more than the target itself.
That broader activation is precisely why selective retrieval can generate both beneficial and harmful spillovers.
2. Retrieval-Induced Forgetting Is Real—but It Is Not the Only Spillover
Classic retrieval-induced forgetting work shows that practising some members of a category can reduce later access to related unpractised members. One proposed mechanism is competition: retrieving the target requires suppressing or overcoming competitors, leaving those competitors temporarily less accessible.
That matters educationally. If a student repeatedly practises only one formula, one definition or one subtype, nearby alternatives can become harder to discriminate or retrieve.
But research also shows circumstances in which selective retrieval can improve memory for other items. The same operation that selects one target can sometimes reactivate the shared study context or broader set strongly enough to help neighbours.
3. Think in Memory Neighbourhoods
A useful educational metaphor is a neighbourhood.
- The target item is one house.
- Category membership creates nearby streets.
- Temporal context links events learned in the same session.
- Meaning links related concepts.
- Shared cues create common entrances.
- Competition can make one route dominate another.
Retrieving one house can strengthen the path to that house. It can also reactivate the map around it—or make neighbouring routes less accessible if the search repeatedly excludes them.
4. The 2026 Study: What It Actually Showed
The September 2026 study used Chinese characters that differed in prior familiarity: common characters represented higher prior knowledge and rare characters lower prior knowledge. Participants were assessed either directly or after selective retrieval.
Three findings matter here:
- retrieved items benefited;
- nonretrieved items also benefited;
- high-prior-knowledge items were remembered better overall, but the retrieval-related gains and forgetting rates were similar across knowledge levels.
The authors interpret the result as evidence that selective retrieval can act as a domain-general memory-stabilising phenomenon rather than depending entirely on semantic expertise.
Do not overextend that conclusion. Chinese-character learning is not the same as a Mathematics chapter, a literature essay or a Science mechanism. The study does not prove that testing five questions from any school topic will automatically strengthen the untested questions.
5. Context Reactivation Can Carry Benefit Beyond the Target
One route to facilitation is context reactivation. Retrieving an item can bring back aspects of the episode in which several items were learned together.
If that reinstated context makes neighbouring memories easier to access, selective retrieval can act like reopening a folder rather than merely opening one file.
This does not mean the entire folder is strengthened equally. Reactivation can be partial, uneven and dependent on how the material was encoded.
6. Relational Encoding Changes the Question
If items were learned as disconnected facts, retrieving one may have little reason to reactivate the others. If they were learned as a coherent causal chain, comparison set or conceptual network, selective retrieval has more routes through which neighbouring information may return.
This is one reason “study the chapter” and “memorise twenty cards” are not cognitively identical even when the same words appear.
7. Facilitation Does Not Mean Transfer
If an unpractised item becomes easier to recall, that is a memory benefit. It does not automatically prove the learner can explain it, apply it, discriminate it from a near neighbour or use it independently.
The transfer law still applies:
Better accessibility is not the same as broader capability.
After selective retrieval, test the unpractised knowledge under changed wording or application before concluding that the whole knowledge network improved.
8. Mathematics: Practising One Form Can Wake a Family—or Narrow It
Suppose a learner practises solving simultaneous equations by substitution. That practice may reactivate the broader system of equations, variables, equivalence and solution checking. This could make neighbouring knowledge more accessible.
But repeated exclusive use of substitution can also narrow method selection. Elimination may become less available, especially if the learner treats “simultaneous equations” as a cue that automatically triggers one route.
The educational question is therefore not whether spillover exists in principle. It is whether the practice set preserves the full decision space the learner will later need.
9. Science: Retrieve One Link, Then Test the Chain
In a causal mechanism, retrieving one step can reactivate surrounding steps. Asking “why does evaporation cool the remaining liquid?” may wake knowledge about higher-energy particles escaping, average kinetic energy and temperature.
That is useful facilitation if the learner can later reconstruct the full chain. It is not useful if the practised sentence becomes a memorised island while other relations remain unavailable.
10. English and Humanities: Selective Evidence Can Distort the Argument
Practising retrieval of one argument or quotation may reactivate a broader text. But it may also make one interpretation dominate simply because it is practised most often.
For essays, selective retrieval should therefore be followed by a coverage check: can the learner still retrieve alternatives, counterarguments, source distinctions and qualifications?
11. The Coverage Trap
A learner tests twenty percent of a topic and performs well. Because selective retrieval may produce some broader reactivation, the learner feels the remaining eighty percent is “probably there.”
That inference is unsafe.
Facilitation is not a licence to under-sample the syllabus. Untested material needs its own evidence before an examination or real performance.
12. The Neighbour Test
After practising one item, deliberately test a related unpractised neighbour.
- If it is easier than before, facilitation may have occurred.
- If it is harder, competition or retrieval-induced forgetting may be involved.
- If nothing changes, the retrieval event may have remained mostly local.
For an individual learner, this is a practical diagnostic rather than a laboratory identification of mechanism. Do not pretend one classroom observation proves which memory process caused the change.
13. The Opposite-Direction Check
Every selective-practice plan should ask two competing hypotheses:
- Facilitation hypothesis: retrieving one part reactivates and strengthens the larger network.
- Competition hypothesis: retrieving one part suppresses, blocks or crowds out alternatives.
The safest study design does not assume which one won. It tests the neighbours.
14. Why Mixed Practice Can Be Protective
If selective retrieval repeatedly favours the same pathway, mixing nearby alternatives can keep the learner’s discrimination system alive.
For example, mix:
- two similar formulas;
- two grammar constructions;
- two scientific mechanisms;
- two historical explanations;
- two problem-solving methods.
The goal is not random variety. It is to ensure that retrieval strengthens selection among relevant neighbours rather than allowing one answer to monopolise a broad cue.
15. Prior Knowledge Raises the Floor Without Guaranteeing the Spillover
The 2026 study found that familiar characters were remembered better overall. That is unsurprising: prior knowledge can make encoding and retrieval easier.
But the authors reported comparable retrieval gains and forgetting rates across high- and low-knowledge conditions. In that experiment, prior knowledge changed overall memory level more than the selective-retrieval benefit.
Do not generalise this to every kind of school knowledge. Prior knowledge can alter comprehension, cue quality, interference and transfer in many ways. The specific result is valuable precisely because it constrains one claim rather than settling all of them.
16. Forgetting Can Restart After Retrieval
An especially interesting finding in the 2026 paper is that selective retrieval was associated with a restart of forgetting functions. Retrieval can make a memory more accessible now while also placing it back onto a new forgetting trajectory.
This matters for studying because a successful retrieval is not a permanent repair receipt. The memory may need to return again later.
Spacing remains relevant even after a powerful retrieval event.
17. Do Not Count the Benefit Twice
If retrieving A also makes B easier, it is tempting to treat one practice attempt as if it independently trained both A and B.
That can inflate study metrics. B still needs direct verification. Facilitation is a possible bonus, not a bookkeeping shortcut.
18. Study Groups Create an Extra Layer of Spillover
When one learner retrieves aloud, other learners may experience reactivation without producing the answer themselves. That can make the material feel newly accessible.
Do not confuse that social reactivation with individual retrieval practice. After group discussion, require each learner to reconstruct independently.
19. AI Can Create Apparent Facilitation Without Learner Retrieval
An AI explanation can mention one concept and automatically surface several related ideas. The learner may experience a broad sense of activation: “Yes, I remember all of this now.”
But the tool performed the retrieval and selection. Close the tool and test whether neighbouring knowledge can still be generated independently.
20. The Delayed Neighbour Test
Immediate accessibility is not enough. After a delay, test:
- the practised item;
- a related unpractised item;
- a competing alternative;
- a changed example requiring the same relation.
This four-part test distinguishes local strengthening, beneficial spillover, narrowing and transfer.
21. Parent and Tutor Guide
If a learner practises a subset of a topic, do not ask only whether the practised questions improved. Ask:
- Did related unpractised knowledge become easier to retrieve?
- Did alternatives become harder to retrieve?
- Can the learner still choose among methods?
- Does the broader explanation remain coherent?
- Is performance still strong after delay?
- Does the learner succeed on a fresh example rather than the practised wording?
This prevents the practice set from becoming the entire definition of what the learner knows.
22. Evidence Boundary
The September 2026 Psychonomic Bulletin & Review paper is a peer-reviewed laboratory study of 216 adult participants learning common and rare Chinese characters. It provides direct evidence that selective retrieval can improve recall for nonretrieved material under those conditions and that the effect did not depend strongly on the prior-knowledge manipulation used in the study.
It does not establish that selective retrieval facilitates every related memory, every educational subject or every learner. A large literature shows retrieval-induced forgetting and output interference in other paradigms. The direction of spillover depends on retrieval structure, context, competition and encoding relations.
The educational application here is therefore deliberately conservative: selective practice may change unpractised neighbours, so test those neighbours rather than assuming either benefit or harm.
23. The Retrieval-Spillover Protocol
- Choose a meaningful subset for retrieval.
- Record which items were practised and which were not.
- Retrieve the subset without answer-giving support.
- After a short delay, test one or two related unpractised neighbours.
- Test a competing alternative or method.
- If neighbours weaken, widen practice and strengthen discrimination.
- If neighbours improve, treat the gain as provisional until directly verified.
- Space a later return to both practised and unpractised material.
- Use changed cues or examples.
- Require independent performance before calling the network stable.
24. Return: Retrieval Changes the Neighbourhood
The learner who practised five items did not necessarily change only five memories.
Retrieval can reactivate context, sharpen a target, suppress competitors, reorganise a search route and sometimes strengthen information that was never directly practised.
The important study habit is not to memorise a slogan about spillover. It is to test the neighbourhood. If selective retrieval helped the neighbours, use the benefit. If it narrowed access, repair the coverage. Either way, make the conclusion from later independent evidence rather than from the feeling that the whole chapter woke up at once.
Continue through Retrieval-Induced Forgetting, Retrieval Interference, the How Studying Works Numbered Series Reading Index and the How X Works Hub.