Learning Science · Concepts and changing explanations

Mental models and their assumptions

Make a mental model explicit so that its predictions and limits can be checked.

Use the examples to connect this lesson with the earlier topics. No specialist equipment is needed.

After reading, you can

Identify elements, relationships and assumptions in a model.
Use a failed prediction to revise a model appropriately.

An explanation you can run

You expect a room to warm when its heater is switched on. Behind that expectation is a model: the heater transfers energy to the room, while energy also passes to cooler surroundings. A mental model is an internal representation of how a situation works. It can be a spatial picture, a set of causal relationships, or both. It need not be expressed as an equation.

Models help because they simplify. A sketch of water entering and leaving a tank can focus attention on amount and rate without representing every moving molecule. But simplification requires assumptions. A prediction about the tank rising assumes the incoming rate exceeds the outgoing rate over the interval considered. Omitting this condition changes what the model can safely predict.

Distinguish the representation from the thing

An analogy borrows relationships from a familiar case. Comparing electric current with water flow may help introduce a circulating system, but the borrowed picture has limits. The learner should say which parts correspond and where the comparison stops. Unchecked analogies can import extra properties that were never intended.

Draw or describe the elements, connect them with labelled relationships, and state what is deliberately ignored. Ask what the model predicts if one input changes. Then compare the prediction with a calculation, observation, or accepted explanation. A useful model can be incomplete; an incomplete model becomes misleading when used beyond the conditions that made it useful.

Treat disagreement as information

If a prediction fails, first check the observation and the application. A thermometer may be misread or a supposedly closed tank may leak. Next inspect the model’s assumptions. Changing a model is more informative than simply attaching the new answer to an old picture that still predicts the opposite.

A learner may hold several models and use different ones in different contexts. Research on scientific intuitions suggests that an earlier explanation can remain accessible after formal learning. Therefore one correct verbal answer does not establish that an incompatible model has disappeared. Ask for predictions in varied cases, allowing enough time for reasoning rather than interpreting speed alone as understanding.

Worked example

A room that stops getting warmer

A room’s temperature stops increasing even while its heater operates. How does adding an outgoing-energy process repair an inflow-only model?

State the first model’s prediction: heater on means energy is added, so temperature rises indefinitely. It omits a relevant pathway out of the room.
Add energy loss through walls and ventilation to cooler surroundings. In a simplified well-mixed model, net accumulation depends on the input rate minus the loss rate. If input exceeds loss, the stored energy increases; if they balance, it need not increase.
Relate this to temperature under the model’s assumptions, including a stable effective heat capacity and other transfers being negligible or included. Real heater power, ventilation and outside temperature may vary, so a rapid fixed temperature is not guaranteed.

A steady temperature can coexist with an operating heater when energy input and loss balance under the stated model. The missing outgoing process gives the learner a reasoned alternative to an indefinite-warming rule.

A heater adds energy while the room also loses it

A simplified room has energy input from a heater and energy loss to cooler surroundings. If input exceeds loss, stored energy increases and temperature can rise. If input equals loss, there is no net accumulation and temperature can remain steady. Assume a well-mixed room, fixed effective heat capacity and other transfers negligible or included. Arrows show directions, not measured magnitudes.

Try it yourself

Think through each question before opening the answer.

A tank’s water level falls despite water entering. Is that logically impossible?
Show the answer and reasoning
No. If water leaves faster than it enters, the amount falls. An inflow-only model omitted a relevant relationship.
The amount depends on both rates. If outflow exceeds inflow over the interval, the tank loses water despite receiving some; the falling level is compatible with a model that includes both processes.
What should an analogy explanation include besides matching parts?
Show the answer and reasoning
It should state which relationships transfer and identify an important limit. Similarity in one respect does not license every inference from the familiar system.
An analogy borrows selected relationships. State which ones correspond and where the comparison stops, so a helpful picture does not import unsupported extra properties.
Why ask for a prediction after a learner repeats the correct explanation?
Show the answer and reasoning
Prediction checks how the explanation is being used. A learner can repeat words while applying a conflicting model to the new case.
Repeating an accepted explanation does not show which model generates a new response. A fresh prediction and its reasons reveal how the explanation is being used.

What to carry forward

Make assumptions visible before trusting predictions.
A failed prediction invites checking evidence, application and model.

References

Sources supporting the explanations and learning checks in this lesson.