Direct Answer: Learning from expert modelling works when a learner observes not only what an expert does, but also the decisions, cues, checks and reasons that control the performance. A polished demonstration can hide the thinking that made it possible. Experts often skip steps mentally because those steps are already automatic. Novices need those invisible decisions made explicit. Strong modelling therefore combines demonstration with carefully chosen thinking aloud, highlights the cue that triggers each move, shows errors and recovery where useful, and then reduces the model so the learner has to carry progressively more of the task. The final receipt is not that the learner watched an expert perform. It is that the learner can reproduce the relevant decision process, adapt it to a changed case and explain why the route works.
HOW LEARNING WORKS · EXPERT MODELLING
The expert’s hands are visible. The expert’s judgement usually is not.
Modelling earns its place when it makes the hidden decision process inspectable and then gives that process back to the learner.
The simplest definition
Expert modelling is the deliberate demonstration of a task together with the cues, reasoning, monitoring and decisions that make the task work.
The model can be a teacher, tutor, peer, video or worked demonstration. What matters is whether the learner can see the structure of the performance rather than only its finished surface.
The expert-modelling mechanism
TASK → EXPERT ORIENTS → RELEVANT CUE IDENTIFIED → DECISION VERBALISED → ACTION DEMONSTRATED → CHECK EXPOSED → LEARNER PREDICTS NEXT MOVE → PARTIAL REPRODUCTION → FEEDBACK → SUPPORT FADES → INDEPENDENT PERFORMANCE → TRANSFER
1. Demonstration can hide expertise
Experts act quickly because many sub-decisions have become compressed. A Mathematics teacher sees a factorable quadratic immediately. An experienced reader notices a contradiction before a novice knows which sentence matters. A skilled writer rejects a sentence because it violates tone without consciously listing every rule.
If the learner sees only the final action, the most valuable cognition remains invisible.
2. Thinking aloud reveals the decision layer
Thinking aloud means verbalising selected thought processes while performing the task.
The EEF’s 2026 construction guidance describes modelling thinking aloud as a way to make expert judgements and rationale explicit to novices, linking it to its metacognition recommendations. EEF: Thinking aloud in construction.
The key word is selected. An expert cannot and should not narrate every mental event. The useful narration identifies the decisions the learner must eventually make.
3. Model cues, not only procedures
“First I do this, then this” gives sequence. “I choose this because the question contains this relationship” gives strategy.
The learner needs to know what feature triggered the action. Otherwise the procedure can become a script attached to the demonstration rather than a method attached to the problem structure.
4. Model checks as part of the performance
Experts do not only act; they monitor.
Show the reasonableness check, the unit check, the rereading of the question, the comparison with evidence, the moment a sentence is rejected, or the decision to change strategy.
If modelling shows only flawless forward motion, learners can mistake expertise for never needing to verify.
5. Model uncertainty honestly
Some tasks are genuinely uncertain.
An expert may say, “Two interpretations are plausible. I prefer this one because the evidence is stronger here.” That teaches judgement better than pretending every complex task has one obvious route from the start.
6. Model error recovery
A carefully chosen error can reveal monitoring.
“I nearly used this formula, but the units show it cannot fit.” “I assumed the pronoun referred to the nearest noun, but the sentence meaning contradicts that.”
The learner sees that good performance includes detecting and repairing a wrong route.
7. Expert modelling and worked examples overlap—but are not identical
A worked example is a completed solution or explanation the learner studies. Expert modelling is the performance of the decisions behind that solution, often in real time.
A static worked example can be excellent. A live model can expose decision points that static notation hides. The choice depends on what the learner needs to see.
8. Observational learning is broader than expert modelling
People learn from observing others in many social contexts. Expert modelling is a deliberate instructional use of observation with the important structure highlighted.
The educational advantage is control: the teacher can select the task, cue, commentary and handover.
9. Novices need more explicit modelling than experts
A novice may need the teacher to identify the cue, model the decision and explain the check.
A more advanced learner may benefit from a sparse model that highlights only a strategic fork. Too much narration can become redundant once the learner already owns the basic route.
10. Ask the learner to predict the next move
Pause the model before a key decision.
“What would you do next?” “Which evidence matters now?” “Would you continue or switch?” Prediction turns observation into active model reconstruction.
11. Ask the learner to explain the model
After the demonstration, ask for the decision rule rather than the sequence alone.
“Why did I choose that method?” “What would have made the other method better?” “Which check caught the possible error?”
This separates imitation from understanding.
12. Fade the model deliberately
The model should become smaller as the learner becomes stronger.
Full demonstration → partial demonstration → cue only → learner attempt → delayed independent task.
If the teacher continues modelling every step, the learner may become excellent at following and weak at initiating.
13. Mathematics modelling should expose method selection
Do not only solve the equation.
Say what you notice, which method candidates activate, why one is efficient, which condition would invalidate it, and how you check the result.
The learner should eventually reproduce that orientation process on a different equation.
14. Science modelling should expose evidence discipline
Model how an expert distinguishes observation from inference, variable from outcome, mechanism from description, and supported conclusion from overclaim.
A beautiful experiment demonstration can still leave the scientific reasoning hidden if the teacher never verbalises these distinctions.
15. English modelling should expose reader and writer decisions
While reading, model how evidence changes an interpretation. While writing, model how purpose and audience control wording.
“This sentence is grammatical, but I am rejecting it because the tone is too casual for this reader.” That is a decision a learner can later reuse.
16. Model variety matters
If learners see only one expert route, they may infer that good performance always looks the same.
Where legitimate alternatives exist, compare two expert solutions. Ask what each optimises and under which conditions one becomes preferable.
17. The learner must eventually become the model
Ask the learner to think aloud while solving.
The teacher can now inspect whether relevant cues are being noticed, whether method selection is justified and whether checking is active.
This is the handover: expert cognition becomes learner cognition.
18. The final receipt is independent adaptation
Use a fresh case that cannot be solved by copying the exact demonstration.
If the learner notices the right cue, selects a defensible route, checks the result and adjusts under changed conditions, the model has become knowledge rather than theatre.
What expert modelling is not
- A flawless demonstration is not automatically an informative model.
- Thinking aloud should expose selected decisions, not every possible thought.
- Watching is not the final learning receipt.
- Imitating steps is not the same as recognising the cue that made them appropriate.
- Expert modelling should include checking and, where useful, recovery.
- The model must fade as learner control grows.
An expert-modelling diagnostic map
| What adults see | Possible modelling issue | Useful next move |
|---|---|---|
| Can copy steps but not choose method | Decision cues hidden | Model why the method is selected |
| Freezes when example changes | Surface imitation | Change the case and ask for cue recognition |
| Never checks work | Model showed only forward execution | Think aloud through verification |
| Believes experts never make mistakes | Recovery hidden | Model detection and repair of a plausible error |
| Needs teacher demonstration every time | Model dependence | Fade from full model to cue-only support |
| Understands model but cannot explain it | Implicit observation | Require explain-back and learner think-aloud |
A practical expert-modelling cycle
- Name the learning decision.
- Model the relevant cue.
- Think aloud through the decision.
- Demonstrate the action.
- Show the check.
- Pause before the next decision and ask the learner to predict.
- Ask the learner to explain why the route works.
- Give a partial model.
- Move to independent performance.
- Test on a changed case.
For parents
- “What did the tutor notice first?”
- “Why did they choose that method?”
- “What check did they use?”
- “Could you do the same decision on a different question?”
- “What part can you now do before asking to see another example?”
For students
- Watch decisions, not only hand movements or written steps.
- Ask what cue triggered the method.
- Predict the next move before it is shown.
- Explain the model in your own route.
- Notice how the expert checks and recovers.
- Move to a changed case before watching another full demonstration.
How do we know expert modelling is working?
- Learners increasingly identify relevant cues before the teacher names them.
- Method selection becomes explainable.
- Checks appear in independent work.
- Learners predict expert moves accurately for structural reasons.
- Teacher narration can be reduced.
- Changed cases are handled without copying the surface.
- Learners can think aloud through their own decisions.
The complete expert-modelling chain
NOTICE → VERBALISE DECISION → DEMONSTRATE → CHECK → PREDICT → REPRODUCE → FADE MODEL → THINK ALOUD → ADAPT → PERFORM
Research and evidence boundary
Modelling is a central component of explicit instruction and metacognitive teaching. The Education Endowment Foundation’s metacognition guidance recommends teachers model their own thinking, and its July 2026 construction article gives a current applied account of expert thinking aloud. These sources support making decisions and rationale explicit; they do not establish that every narrated demonstration is superior to every worked example or independent task. Modelling should be selected according to the learner’s knowledge and then faded as independent control develops. EEF Metacognition and Self-Regulated Learning; EEF Thinking Aloud, 27 July 2026.
