Quick Read
Feedback is useful, but it changes the learner’s state.
Once the answer is shown, we no longer know exactly what the learner expected beforehand.
That is why one of the cleanest Curie questions is:
“Before I tell you the answer, what do you think should happen—and why?”
The prediction makes the learner’s internal model visible before reality or the teacher corrects it.
The One-Sentence Answer
A prediction made before feedback reveals how the learner currently believes the system, text, relationship or method will behave, making hidden assumptions easier to test.
Where Prediction Sits in the Curie Loop
Observed, Inferred, Unknown teaches us to separate evidence from interpretation.
Prediction adds a forward-facing test. Before new evidence arrives, what does the learner expect?
Then reality returns. The gap between prediction and outcome becomes evidence.
Why the Timing Matters
If a learner sees the answer first, explanation can become hindsight.
“Of course the graph rises.”
“Of course that character is being ironic.”
“Of course the sign is negative.”
Prediction prevents the final result from rewriting our memory of what we expected.
A Good Prediction Has a Reason
| Prediction | What it reveals | Next evidence |
|---|---|---|
| “I think it increases because these two quantities move together.” | Relational model | Check graph/data |
| “I think the writer will reject this claim because the tone has already shifted.” | Textual model | Check later evidence |
| “I think cooling slows because the temperature difference is getting smaller.” | Causal model | Compare observed trend |
| “I expect one stationary point because the derivative changes sign once.” | A-Math structural model | Check calculation/graph |
A guess without a reason is still an observation. A justified prediction gives us a model we can test.
A Wrong Prediction Can Be Better Evidence Than a Correct Guess
A correct answer may come from luck, familiarity or elimination.
A wrong prediction with a clear reason exposes the exact model that needs repair.
Curie values a visible wrong model more than an invisible correct guess.
English: Predict Meaning Before the Text Confirms It
In English, prediction can reveal whether the learner is tracking viewpoint, tone, narrative direction and argument.
Ask what the writer is likely to do next and what textual evidence supports that expectation. Later, compare the prediction with the actual continuation.
For stage-by-stage English progression, continue through the English Tutor route.
Mathematics: Estimate Direction Before Exact Calculation
Before calculation, a learner can often predict whether an answer should be larger, smaller, positive, negative, increasing, decreasing or roughly within a sensible range.
This gives later calculation something to answer to. If the exact result contradicts the prediction, the learner has a reason to inspect both the prediction and the working.
Detailed mathematical teaching remains with the subject estate. For developmental Mathematics progression, continue through the Mathematics Tutor route.
Science: Prediction Makes the Model Testable
Science provides the most literal form of this idea: a model that predicts nothing is difficult to test.
Ask what should happen if one variable changes, then compare with observation. If reality disagrees, the explanation must remain correctable.
The existing subject page How Scientific Predictions Grow From Patterns, Evidence and Mechanisms provides the Science-specific instruction; Curie keeps the cross-subject measurement principle visible.
Additional Mathematics: Anticipate the Shape Before Finishing the Algebra
A-Math learners can predict sign, turning behaviour, plausible roots, gradient direction or the rough shape of a graph before the full calculation is complete.
Those expectations create useful checks on long symbolic work.
Repair: Compare the Model With the Outcome
- Ask for a prediction before feedback.
- Ask for the reason behind it.
- Reveal or observe the outcome.
- Compare prediction and reality.
- If they differ, locate which assumption caused the difference.
- Revise the model.
- Make a new prediction under a changed condition.
Transfer: A Better Model Predicts Beyond the Original Example
When a learner truly understands a relationship, they can often predict how it behaves when one condition changes. This links Prediction Before Feedback to the Transfer Test.
Tutorial Should Capture the Prediction Before Teaching Continues
Inside a Tutorial, the tutor can ask for a prediction, record the reasoning verbally or on paper, and only then allow the next evidence to arrive.
This preserves a clean snapshot of the learner’s current model.
What Tutor Should Say
The Tutor dashboard might say:
“Your calculation is accurate, but your prediction before calculating was in the opposite direction. We need to strengthen the relationship model, not only the arithmetic.”
Or:
“Your prediction and explanation matched the later evidence across three changed examples. The model is becoming more reliable.”
What Parents Can Ask
- Before you calculate, roughly what should the answer look like?
- Before you read on, what do you expect the writer to do next?
- If this variable increases, what should happen?
- What makes you expect that?
- What would make you change your prediction?
Voyage: Predictions Become More Conditional
Young learners begin with simple expectations. Across the Voyage, predictions should become more evidence-based, conditional and aware of uncertainty.
Darwin: A Model That Predicts Poorly Must Adapt
The Darwin Series begins where repeated prediction errors reveal that an old model no longer fits the environment. The answer is not to defend the old expectation more strongly, but to improve the model.
Boundaries
- Do not confuse prediction with guessing.
- Do not punish learners for making a testable prediction that turns out to be wrong.
- Do not force certainty where the evidence permits only a range or conditional expectation.
- Do not let the correct answer erase the learner’s earlier reasoning; compare them.
The Core Rule of Prediction Before Feedback
Capture what the learner expects before feedback changes the state, make the reason for that expectation visible, then let reality test and refine the underlying model.
Frequently Asked Questions
Why ask for predictions if the student may guess?
Ask for the reason. A prediction with a stated relationship or assumption is much more informative than an unsupported guess.
What if the prediction is wrong but the final answer is right?
That is valuable evidence. The execution may be stronger than the learner’s internal model, so both should be examined separately.
Does prediction matter outside Science?
Yes. Reading, argument, mathematical relationships, planning and adult decision-making all rely on expectations that can be compared with later evidence.
Marie Curie Series · Observed, Inferred, Unknown · Evidence Loop · Transfer Test · Tutor · Voyage · Darwin
eduKate Sengkang | Small groups of up to 3 | 1.5-hour tutorials | Properly Taught Kids Shine a Bright Light Into the Future.
eduKate Sengkang 