Tutors must infer from incomplete evidence. A short learner response may be a misconception, a compressed but sound intuition, unfamiliarity with technical vocabulary, or an attempt to answer a different question. Adaptive teaching cannot wait for certainty, but it also should not turn the first plausible interpretation into a fact.
Suspended judgment names the disciplined interval between noticing a possible explanation and treating it as settled. The interval is active. The tutor identifies live alternatives, asks what evidence would distinguish them, and chooses a question that makes the learner’s next contribution diagnostically useful.
Consider this exchange:
Tutor: Why can a model succeed on training examples but fail on new ones?
Learner: Because it learned too much from the training data.
A closing response says: “Correct, that is overfitting,” then supplies the definition. It may have guessed well, but it has not learned whether the student understands memorisation, model complexity, sample bias, or the gap between fitting and generalising.
An inquiry-preserving response could say:
“Learned too much” might mean that the model memorised details, or that it fitted patterns that do not represent the new cases. Which would better predict failure: giving the same model more varied examples, or merely repeating the old examples? Tell me why.
This response is not stronger because it sounds cautious. It is stronger because the question separates interpretations and makes the next teaching move depend on evidence the learner can supply.
Exercise
Choose one ambiguous learner sentence from your discipline.
- Write three materially different interpretations of it.
- State what the sentence alone cannot establish.
- Ask one question whose likely answers distinguish at least two interpretations.
- Sketch a different next move for each answer.
- Add one sentence showing how the tutor would revise an initial diagnosis.
Remove generic invitations such as “Can you say more?” unless you can explain what the requested information will discriminate. Productive hesitation has a direction: it converts uncertainty into a better inquiry.
This exercise does not establish a learner model or a validated diagnostic method. It makes one boundary inspectable: whether the tutor treats its first interpretation as a hypothesis that the learner can still change.
Continue with Probe 02 and the Dewey reading dispatch.