HSW-0206 · How Studying Works
You watch somebody else choose.
You see what happens.
You never touch the controls.
Later, you remember the episode almost as accurately as if you had made the choice yourself.
That sounds like passive learning. It is not necessarily passive cognition.
Vicarious learning occurs when a person updates knowledge or behaviour from another person’s actions and outcomes rather than only from direct action.
A 2026 study adds a useful nuance: in a reward-learning task, recognition accuracy for incidental images did not significantly differ between direct and observational learning, even though participants felt more confident about memories from their own experience. The objective and subjective memory stories were not identical.
This article owns the narrow study question of how observing another person’s choices and outcomes can influence episodic memory. It does not claim that watching is generally equivalent to doing. The Enactment Effect remains the owner of memory benefits from physically performing actions. The Locus of Generation owns who performs the generative cognitive work when a tool or other person supplies an explanation. Here the object is observation of another actor’s decisions and consequences.
The 50-Second Read
- Observation can be cognitively active. A learner can predict, simulate, compare and update while watching another person act.
- Direct experience is not automatically superior on every memory measure. In a 2026 reinforcement-learning study, recognition accuracy did not significantly differ between experiential and observational trials.
- Subjective memory still differed. Participants reported greater confidence in memories from their own experiences.
- Observed choices mattered. Images associated with observed “play” choices were remembered better than those associated with observed “pass” choices, paralleling direct experience.
- The result is task-specific. It does not prove that watching a worked solution produces the same mathematical skill as solving it.
- Observation should prepare action. Predict what the model will do, explain why, compare the outcome, then perform independently.
- The practical loop: observe → anticipate → explain → compare → imitate selectively → perform alone → return later.
1. Learning Without Direct Contact
Human learning would be impossibly expensive if every lesson required personal trial and error.
We learn by watching:
- a teacher solve a problem;
- a parent repair something;
- a peer make a mistake;
- a scientist demonstrate a procedure;
- a speaker handle a difficult question;
- a player choose under uncertainty.
Observation allows one person’s experience to become information for another.
But “watching” contains several possible cognitive states. A learner can stare at a demonstration with almost no prediction or explanation. Or the learner can internally simulate each decision, compare it with an alternative, notice the outcome and update a model.
The visible behaviour—watching—is the same. The learning operation is not.
2. The 2026 npj Science of Learning Study
Woitow, Jang, Eppinger and colleagues published Shared effects of one’s own and others’ experiences during reinforcement learning on episodic memory in npj Science of Learning on 28 February 2026. See Woitow and colleagues, 2026.
Seventy-nine young adults alternated between two kinds of trials.
- Experiential learning: participants made a play-or-pass decision themselves.
- Observational learning: participants watched a simulated other player make a comparable choice.
Images appeared during the learning task and were later tested unexpectedly. This allowed the researchers to examine how the learning context shaped incidental episodic memory.
3. Accuracy and Confidence Told Different Stories
Recognition accuracy did not significantly differ between the experiential and observational conditions.
That is striking because the common intuition is that “doing it yourself” should always leave a clearly stronger memory.
But subjective confidence did differ. Participants were more confident in memories associated with their own experience.
This separates two questions:
- How accurately did the person discriminate old from new information?
- How strongly did the person feel that the memory was present?
Direct experience may enrich conscious recollection or contextual detail even when one recognition-accuracy measure does not show a large condition difference.
4. Observed Choices Were Not Neutral
In both learning conditions, images from trials involving a play choice were remembered better than images from pass trials.
Crucially, this pattern appeared even when participants were observing somebody else’s play choice.
One interpretation is that people closely represent observed actions when those actions matter to an unfolding decision environment. The observer is not merely receiving pixels. The observer may be anticipating, evaluating and simulating the other person’s action.
5. Observation Can Contain Prediction
Imagine watching a teacher solve an equation.
Before each line, you can ask:
- What will the teacher do next?
- Why that move?
- What alternative would fail?
- What outcome should follow?
Now observation contains a sequence of internal predictions and updates.
That is different from waiting for the finished solution and thinking, “Yes, that makes sense.”
6. Reward Prediction Errors Add a Timing Lesson
The 2026 study also examined reward prediction errors—mismatches between expected and received reward-related information.
Positive prediction errors occurring when the memory image was presented were associated with enhanced memory in both experiential and observational play trials. By contrast, prediction errors at later feedback did not show the same memory relationship.
The timing of a learning signal relative to the information being encoded therefore mattered.
This complements—but does not erase—the caution in Prediction Error and One-Shot Memory, which shows that prediction-error interpretations are paradigm-dependent and can be confounded. Different tasks can legitimately produce different evidence.
7. Watching Is Not the Same as Doing
The study does not justify the claim:
Watching somebody perform a skill is just as good as performing the skill yourself.
The researchers measured incidental episodic memory inside a reinforcement-learning task. They did not test whether observation and practice create equal mathematical fluency, laboratory technique, writing skill or motor execution.
Observation can build representations that later support performance. It does not automatically build the performance itself.
8. Vicarious Learning vs the Enactment Effect
The Enactment Effect owns memory advantages that can arise when a person physically performs an action rather than only reading it.
Vicarious learning asks a different question: what can an observer learn from another person’s actions and outcomes?
The two findings can coexist. Direct action can create distinctive motor and self-generated information, while observation can still produce surprisingly rich representations under some conditions.
9. Vicarious Learning vs the Locus of Generation
The Locus of Generation asks who performed the cognitive work that created an explanation, solution or representation.
Watching another person’s explanation can support learning, but a polished model answer may also substitute for the learner’s own generation.
The solution is not “never observe.” It is to design observation so the learner must predict and explain before the model supplies the next move.
10. The Demonstration Illusion
A demonstration can feel easy because the demonstrator carries the hardest decisions.
The learner sees:
- the correct starting point;
- the relevant information;
- the method selection;
- the correct sequence;
- the stopping rule.
When the learner later works alone, all of those must be generated again.
Observation therefore needs a handoff from represented action to self-produced action.
11. Mathematics: Watch the Decision, Not Only the Algebra
When studying a worked solution, pause before each major step.
- Name the current state of the problem.
- Predict the next operation.
- State why it is appropriate.
- Reveal the expert step.
- Compare your prediction.
- Explain any discrepancy.
This turns a worked solution into a series of vicarious decisions rather than a completed object to admire.
12. English: Read a Model Paragraph as a Sequence of Choices
A model essay is useful when the learner sees the decisions inside it.
- Why did the writer begin with this claim?
- Why this evidence?
- Why qualify here?
- Why not use the obvious example?
- What does the final sentence return to?
The learner should then write a new paragraph with the model hidden.
Observation supplies a map. Independent writing proves that the map can be used.
13. Science: Demonstrations Need Prediction and Explanation
Science classrooms rely on demonstration for good reasons: safety, equipment, time and visibility.
But a dramatic observation does not automatically teach the mechanism.
Before the demonstration, ask students to predict. During it, ask what evidence changes. After it, ask which mechanism best explains the observation and what alternative explanation remains possible.
Now the observer is participating in the reasoning even when not physically manipulating the apparatus.
14. The Observer’s Four Jobs
- Anticipate: what should happen next?
- Represent: what state is the actor responding to?
- Evaluate: did the action fit the evidence?
- Update: what changed after the outcome?
If none of these jobs is running, the learner may be exposed to correct performance without constructing a usable model of it.
15. Observation Needs a Comparison Target
Watching is especially informative when the learner has something to compare with:
- their own predicted move;
- a previous attempt;
- an erroneous example;
- a second expert strategy;
- a counterexample.
Without comparison, the observer can encode surface sequence without learning which features controlled the action.
16. The Center-to-Edge Route
- Center: observe one clear expert performance.
- First ring: predict hidden next steps.
- Second ring: explain decision points.
- Third ring: compare with another case where the method changes.
- Edge: perform independently on an unfamiliar case.
The edge test prevents observation from becoming permanent dependence.
17. The School Route: Peer Modelling Can Scale Experience
A classroom contains many experiences that can be shared.
One student’s strategy can become an object for everybody else’s reasoning. One student’s error can become a diagnostic case. One student’s revision can show how an explanation improves.
The educational value rises when observers are given an explicit job rather than simply told to “watch carefully.”
18. The Systems Route: Observation Is a Low-Risk Simulator
Organisations use simulation, shadowing and case review because direct trial can be expensive or dangerous.
Observation lets learners inspect action–outcome structure before paying the full cost of personal error.
But the simulator becomes useful only if it eventually hands control back to the learner.
19. The Financial Route: Borrow Experience Before Buying It
Direct experience can be costly.
A learner can reduce search cost by observing worked cases first, especially when the problem space is large and early errors teach little.
The investment logic is:
borrow experience cheaply → extract structure → pay for direct practice where direct practice adds unique information.
Observation and action should therefore complement each other rather than compete as ideologies.
20. The Learning Route: Observe → Predict → Perform
A simple vicarious-learning protocol:
- Watch the start of a model performance.
- Pause before the next important decision.
- Predict the move and reason.
- Reveal the model’s choice.
- Compare outcomes.
- Summarise the governing rule.
- Close the model.
- Perform a new case independently.
The independent case is the receipt that observation became capability rather than recognition.
21. The Education Route: Do Not Confuse Objective Accuracy and Subjective Strength
The 2026 study is valuable because direct experience increased confidence without producing a significant recognition-accuracy advantage on the headline comparison.
Education repeatedly meets this problem.
A method can feel richer, more memorable or more engaging without producing an equally large difference on the criterion that ultimately matters.
Measure both experience and performance. Do not let one stand in for the other.
22. The Training Route: Model Comparison
- Show two solutions to the same class of problem.
- Hide the final answer.
- Ask which solution is likely to work better and why.
- Reveal the outcomes.
- Identify the decisive feature.
- Give a third problem where the preferred strategy changes.
This makes observation conditional. The learner must discover when a model is useful, not merely copy what an expert did once.
23. The Improvement Route: Measure the Observation-to-Action Gap
After a demonstration, test three levels:
- recognition: can the learner identify the correct move?
- explanation: can the learner explain why?
- generation: can the learner produce the move without the model present?
The difference between recognition and generation is the observation-to-action gap.
24. The World Route: Civilization Learns Vicariously
Case law, engineering incident reports, clinical case conferences, aviation investigations, scientific papers and military histories all allow people to learn from events they did not personally experience.
That is vicarious learning at civilisational scale.
The value lies not in storytelling alone but in preserving enough causal structure for another person to make a better future decision.
25. Parent and Tutor Guide: Ask the Observer to Commit Before the Model Continues
When showing a worked answer or demonstration, pause at meaningful decisions.
- What would you do next?
- Why?
- What would make another method better?
- What outcome do you expect?
Afterwards, close the model and give a fresh case.
Help should turn observation into eventual independence, not into permanent spectator status.
26. What Not to Do
- Do not claim observation is generally equal to direct practice.
- Do not infer skill transfer from incidental recognition memory.
- Do not treat passive viewing and active prediction as the same observational state.
- Do not let model answers remove all decision-making from the learner.
- Do not assume higher confidence means higher objective accuracy.
- Do not extend reward-learning findings directly into every classroom without testing the educational task.
- Do not keep the learner observing after independent performance becomes the real learning job.
27. Evidence Boundary
The Woitow and colleagues study involved 79 young adults in an experimental reinforcement-learning task with incidental image memory. Recognition accuracy did not significantly differ between experiential and observational learning, while memory confidence was higher after direct experience. Observed and enacted play choices were both associated with better memory than pass choices, and some reward-prediction-error relationships were similar across conditions.
The study does not establish that observation and direct practice are equivalent for academic skill acquisition. The educational protocols in this article are applications built around a plausible mechanism: observation can carry meaningful decision and outcome information, but independent performance must still be measured.
28. Delayed and Independent Performance Check
- Observe a model today.
- Explain the key decisions without looking.
- Wait.
- Perform a new case alone.
- Change one important feature so copying the surface no longer works.
- Ask why the chosen action remains valid.
If performance survives the missing model and changed surface, observation has contributed to a transferable representation.
29. Return: Watching Can Be Learning, but Spectating Is Not Mastery
Humans are built to learn from other humans.
We can encode choices, outcomes and consequences that happened to somebody else. Under some laboratory conditions, the objective memory traces can look more similar to direct experience than intuition predicts.
But education needs the next step.
Watch actively. Predict the move. Explain the outcome. Then close the model and act. The observer becomes a learner when another person’s experience changes what the learner can independently do.
Continue through The Enactment Effect, The Locus of Generation, Prediction Error and One-Shot Memory, the How Studying Works Numbered Series Reading Index and the How X Works Hub.