HSW-0227 · How Studying Works
Most study advice treats a wrong answer as a cost to be minimised. That is sensible when the stakes are real, the error is dangerous, or the learner has no reliable correction. But low-stakes learning creates a different possibility: sometimes a student can be asked to generate an answer that is deliberately wrong, then compare it with the correct explanation and repair the difference.
This sounds perverse. Why practise being wrong when the goal is to become right?
The direct answer is: a deliberate error can force the learner to represent the concept, construct an alternative, notice a boundary and then process the correction against something they personally generated. Under carefully controlled, low-stakes conditions, that contrast may produce learning that is at least as durable as ordinary retrieval practice and, in some studies, may support stronger transfer. It is not a universal replacement for retrieval practice, worked examples or correct practice. It is a specialised learning move whose value depends on correction, task type, prior knowledge and safety.
Deliberate erring is not ordinary failure
Four situations can all contain a wrong answer while doing very different psychological jobs.
- Accidental error: the learner tries to be correct and fails.
- Erroneous example: the learner studies or diagnoses someone else’s incorrect solution.
- Productive failure: the learner attempts a difficult problem before receiving full instruction and may generate unsuccessful solutions.
- Deliberate error: the learner intentionally constructs an incorrect response, knowing that the next job is to identify and correct it.
The distinction matters because the learning mechanisms differ. A deliberate error is not a teacher leaving a misconception uncorrected. It is not praise for carelessness. It is not “make random mistakes.” The learner is temporarily asked to create a plausible wrong alternative so the correction has something explicit to defeat.
What the 2026 experiments found
Xiuyun Qiang and Xiaofeng Ma reported two experiments with college students in the British Journal of Educational Psychology. Their 2026 paper, “The advantages of deliberate errors in promoting college students’ memory retention and transfer”, compared deliberate erring with retrieval practice and copying under equal-time learning conditions.
Experiment 1 used conceptual terms with 96 participants. Experiment 2 used text passages with 103 participants. Both examined immediate or delayed retention, transfer and learners’ metacognitive evaluations.
The pattern was nuanced. In the first experiment, immediate retention did not provide strong evidence of a difference between deliberate erring and retrieval practice, while delayed retention favoured deliberate erring. Transfer favoured deliberate erring. In the second experiment, retention did not significantly differ between deliberate erring and retrieval practice, although both outperformed copying; deliberate erring again produced better transfer. Learners tended to underestimate the deliberate-error condition in their judgments.
The authors describe potential transfer advantages over retrieval practice, but they also state that this comparison needs further replication. That is the right level of confidence. Two experiments with college students are interesting evidence, not permission to replace ordinary study with a blanket “get it wrong first” rule.
Why making the error yourself may matter
A correction is easiest to ignore when it arrives without a competing model. The student reads the right answer, recognises that it looks reasonable and moves on. Recognition can be shallow because nothing inside the learner has been forced to collide with the explanation.
Deliberate erring can create that collision deliberately. The learner has to ask, “What would a wrong version look like?” To answer, they must represent at least part of the rule. When the correction arrives, the learner can compare two structures: the constructed wrong case and the valid one.
Several mechanisms are plausible rather than conclusively established. Generating the wrong alternative may increase elaboration. The contrast may sharpen discriminating features. The self-generated response may create a stronger episodic trace that makes the later correction distinctive. The requirement to explain why the deliberate error is wrong may support abstraction beyond the original item.
These are theoretical explanations. The experiments demonstrate performance differences under their conditions; they do not isolate one single mechanism as the cause.
A worked example: build the wrong rule, then break it
Consider an illustrative vocabulary lesson on the difference between infer and imply. A conventional study sequence gives definitions, examples and a quiz. A deliberate-error sequence can add one extra move.
The learner is asked: “Write a sentence that deliberately swaps the roles of infer and imply. Make the sentence sound plausible.” The student writes: “From her silence, the speaker inferred that she was unhappy.” That sentence is actually correct, so the tutor says the deliberate error has not yet been created. The learner tries again: “The evidence implied that the detective should imply the suspect was lying.” Now the student has produced a mixed structure that can be inspected.
The correction is not merely “wrong.” The learner identifies the roles: evidence or a speaker can imply; a reader or listener infers. The tutor then asks for one new correct sentence and, after a delay, a different context with no cue about which word belongs.
The educational job is boundary learning. The deliberate error is useful only because the learner must locate exactly what made it invalid and then prove that the corrected distinction transfers.
The error must be plausible enough to teach from
Random nonsense is usually a poor deliberate error. If a student studying fractions writes “one half plus one quarter equals seventeen,” the mistake may not activate the structure of fraction addition at all. There is little useful contrast.
A stronger deliberate error violates a meaningful decision boundary. For example, “one half plus one quarter equals two sixths” can reveal an overgeneralised rule: adding numerators and denominators independently. The correction can then expose why denominators represent unit size and why equivalent fractions are needed before combining quantities.
This is also why deliberate erring should not be introduced indiscriminately to a learner with almost no domain knowledge. If the student cannot tell what kind of wrong answer is structurally plausible, generating errors may become guessing. Some prior representation is needed before error construction can become informative.
The correction is the centre of the method
The phrase “learn from mistakes” can hide the most important step: an error does not teach merely by existing. It must be corrected accurately, and the learner must process the relation between the error and the correction.
A deliberate-error cycle should therefore include four explicit moves:
- Generate: create one plausible incorrect response on purpose.
- Locate: identify the exact rule, assumption or relationship that makes it wrong.
- Repair: produce the correct version and explain the difference.
- Transfer: solve a changed case where the same distinction must be selected independently.
If the cycle stops after generation, the learner has rehearsed an error. If it stops after receiving the right answer, the learner may have watched the correction instead of owning it. The transfer step tests whether the corrected rule can guide a fresh decision.
How this differs from existing error-learning owners
eduKateSG already has several specialist owners around mistakes, and their jobs should stay separate. How Erroneous Examples Work focuses on diagnosing someone else’s incorrect solution. How Productive Failure Works examines difficult pre-instruction problem solving where unsuccessful attempts can prepare later learning. How the Hypercorrection Effect Works addresses what can happen when a learner confidently makes an error and then receives corrective feedback.
Deliberate erring is narrower: the wrong response is intentionally generated as part of the study procedure. That deliberate first-person construction is the mechanism being tested.
Where deliberate errors are a bad idea
The technique has clear boundaries. Do not use deliberate error generation when an incorrect action could cause physical, legal, financial, medical or safety harm. Do not ask a novice to practise a dangerous procedure incorrectly. Simulation can sometimes create safe contrast, but the safety owner must govern the design.
It is also a poor fit when the learner cannot reliably access corrective feedback, when the misconception is already highly persistent, or when the task is so elementary that constructing a plausible wrong alternative adds administrative effort without useful discrimination.
Finally, deliberate error work should be low stakes. If students believe every generated wrong answer will be graded as incompetence, they are being asked to take a learning risk inside a punishment system. That changes the task.
Why learners may dislike a method that helps
Qiang and Ma found a metacognitive mismatch: learners tended to underestimate deliberate erring relative to other strategies. This is educationally plausible. Copying feels fluent. Retrieval practice feels recognisably effortful and legitimate. Intentionally writing something false can feel like moving backwards.
That feeling should not be dismissed. A method that produces confusion, embarrassment or mistrust may be implemented badly even if the laboratory mechanism is promising. Students need to know why the error is being generated, that correction is immediate and mandatory, and that the wrong answer is a temporary contrast case rather than something to memorise.
A small diagnostic protocol for students
Use deliberate errors selectively on concepts with confusable boundaries. Pick one concept and ask:
- What is a plausible wrong version of this rule?
- Why would someone be tempted by it?
- What single feature makes it invalid?
- What is the corrected principle?
- Can I use the corrected principle in a different example tomorrow?
The last question is the real test. A learner who can explain yesterday’s deliberate error but cannot recognise the same boundary in a fresh problem has learned the episode more strongly than the principle.
For parents and tutors: control the risk, then ask for the reason
A tutor can use deliberate errors as short contrast drills after a concept has an initial foundation. “Give me the tempting wrong rule.” “Make one example that breaks the rule.” “Now repair it.” “Now show me a case where the original rule really does apply.” This makes misconceptions inspectable without treating every mistake as evidence of low ability.
Parents should not turn the technique into a game of catching the child out. The learning value comes from the child generating and repairing the contrast. If the adult supplies the trap, laughs at the mistake and then announces the answer, the cognitive and emotional job has changed.
The delayed and independent check
After a day or longer, test the target distinction with new material. Do not remind the learner of the deliberate error. If the task was conceptual, ask for a new application. If it was procedural, change the surface while preserving the decision boundary. If it was vocabulary or language, place the distinction in a new sentence or passage.
Record three things separately: retention of the correct answer, explanation of why the tempting alternative is wrong, and transfer to the new case. Deliberate erring is most interesting when it changes the third measure, not merely when the original correction is remembered.
What remains uncertain
The 2026 study adds to earlier deliberate-erring research, but the evidence base remains much smaller than the literature on retrieval practice. Samples have often been university students, tasks are controlled, and implementation details matter. We should not assume equal effects for young children, every subject, every error type or long real-world retention intervals.
The evidence also does not mean errors are inherently beneficial. Uncorrected errors can persist. Repeated exposure to false information can create familiarity. High-stakes domains can punish incorrect action severely. The scientific job is therefore to understand the conditions under which an intentional, corrected error creates useful contrast—not to romanticise being wrong.
The return: the wrong answer is useful only if it sharpens the right one
Studying normally tries to increase the number of correct responses. Deliberate erring asks a different question for a moment: can constructing the wrong response make the boundary of the right response clearer?
Sometimes, under controlled conditions, the answer appears to be yes. But the error is not the achievement. The achievement is the corrected distinction that survives delay and transfers to a problem the learner has not seen before.
Research and further reading
- Qiang, X., & Ma, X. (2026). The advantages of deliberate errors in promoting college students’ memory retention and transfer. British Journal of Educational Psychology, 96, 922–944.
- Wong, S. S. H. (2023). Deliberate Erring Improves Far Transfer of Learning More Than Errorless Elaboration and Spotting and Correcting Others’ Errors. Educational Psychology Review, 35, 16.
- How Studying Works | Numbered Series Reading Index
- How X Works | Master Hub