How People Actually Learn: The Frameworks Behind Physica

A physics lesson often starts with a definition, followed by equations, a worked example, and some problems to solve. The order is familiar. I wanted to understand what a learner needs before the definition will make sense.

Building Physica made me ask a question I should have spent more time on during my own education: how do people actually learn something?

Instructional design and inquiry-based science education gave me ways to think about that question. They deal with practical decisions: when to explain a concept, when to let someone experiment, and how to find out whether they’ve understood it.

The frameworks below inform how I want to teach physics here. Each draws attention to a part of learning that’s easy to overlook when you’re focused on getting the explanation right.

Start with what the learner already believes

Constructivism describes learning as building on and revising existing understanding. A learner arrives with ideas about how the world works. Some are useful; others will get in the way of a new explanation.

That matters in physics. Someone who believes heavier objects fall faster may be able to repeat the correct equation while still relying on their earlier belief when making a prediction. The lesson needs to give them a chance to examine that belief.

Jean Piaget and Lev Vygotsky are associated with different strands of constructivist thought. For Physica, the practical point is to ask what someone expects before showing them what happens. Their answer gives the explanation somewhere to begin.

The 5E model

Developed by Rodger Bybee and colleagues at BSCS in 1987, the 5E model has five phases: Engage, Explore, Explain, Elaborate, and Evaluate.

The order of Explore and Explain is especially useful to me. Learners first investigate a phenomenon. They then have observations to discuss when the lesson introduces an explanation and the language needed to describe it.

For a physics lesson, that might mean letting someone change a simulation’s inputs and watch the result before presenting the equation. When the equation appears, they can connect its variables to something they’ve already seen.

Elaborate gives them a chance to use the idea in another situation. Evaluate checks their understanding. Those phases also remind me that an explanation needs to lead to something the learner can try.

The 7E model

Arthur Eisenkraft’s 2003 extension gives more attention to prior knowledge and transfer. Its phases are Elicit, Engage, Explore, Explain, Elaborate, Evaluate, and Extend.

Elicit asks learners to make their current thinking explicit. Before a falling-object demonstration, for example, ask which object they expect to land first and why. Afterwards, they have a specific prediction to compare with what they observed. If the result surprises them, the lesson can address the reasoning behind their answer.

Extend asks whether they can use the idea beyond the setting in which they learned it. Near transfer involves a problem similar to the original. Far transfer involves recognising the same underlying idea in a less familiar context.

A pendulum and an LC circuit offer an example. In the small-angle approximation, an ideal pendulum’s motion and the oscillations of an ideal LC circuit follow the same mathematical form. Working through that connection asks more of a learner than changing the numbers in another pendulum problem.

Of these frameworks, 7E comes closest to the sequence I want Physica’s lessons to follow. It gives me a reason to include both an opening prediction and a later problem that looks different from the first one.

Predict, Observe, Explain

White and Gunstone described Predict, Observe, Explain (POE) in 1992. It’s a short activity that can sit within a larger lesson.

First, the learner predicts what will happen and explains their reasoning. Next, they observe the result. Finally, they try to explain what happened, including any difference between the result and their prediction.

Writing down the prediction matters. Without it, it’s easy to see the result and feel as though you expected it all along. A recorded answer gives you something specific to reconsider. Even a correct prediction is worth discussing: the reasoning behind it may still need work.

This is the idea behind what I’m calling the Calibration Block in Physica: an input where you estimate a numerical value before seeing the result. The useful part comes afterwards, when you work out why your estimate was close or far off.

A lesson can use this activity more than once. After an explanation, a new prediction gives the learner a chance to see whether their reasoning has changed.

Kolb’s experiential learning cycle

David Kolb’s 1984 model describes four stages: Concrete Experience, Reflective Observation, Abstract Conceptualization, and Active Experimentation.

You have an experience, think about it, develop an explanation, and test that explanation. The test gives you another experience to reflect on. This cycle makes room for understanding to develop over several attempts.

In a physics lesson, a learner might observe a simulation, describe a pattern, propose a rule, and then change an input to test it. If the rule fails, the next attempt starts with a more specific question.

Kolb also developed a learning-style account associated with preferences within this cycle. I don’t use those preferences to assign learners a fixed type or limit how they’re taught. The cycle is the part I’m drawing on here.

For Physica, it means a lesson should be easy to revisit. Someone who returns after working through a problem may notice something they missed the first time. I want the structure to support that return, with the explanation and opportunities to experiment still easy to find.

Gagné’s nine events of instruction

Robert Gagné’s work, beginning in 1965 and developed in later editions, including 1988, describes nine instructional events:

1. Gain attention.
2. Inform learners of the objectives.
3. Stimulate recall of prior learning.
4. Present the content.
5. Provide learning guidance.
6. Elicit performance.
7. Provide feedback.
8. Assess performance.
9. Enhance retention and transfer.

I find this useful when checking whether a lesson asks enough of the learner. Presenting the content is only one event. The lesson also needs to help someone attempt the work and respond to their attempt.

Informing learners of the objective can be straightforward: state the question they’ll be able to answer by the end. That gives them a reason to pay attention to the details that follow.

Eliciting performance and providing feedback belong together. Asking a question without helping the learner understand their answer leaves part of the teaching unfinished. Feedback should explain the reasoning, especially when an incorrect answer came from a plausible assumption.

The final event, retention and transfer, also pushes me to look beyond the immediate exercise. Can the learner recall the idea later? Can they recognise when it applies to a different problem?

Teaching through problem solving

Teaching Through Problem Solving (TTP), associated with Japanese mathematics teaching, puts a problem before the formal introduction of its solution method.

Students attempt the problem using what they already know. The teacher then selects approaches for the class to compare and discuss, including attempts that didn’t work. This discussion is often called neriage. The lesson develops the mathematical idea through those comparisons and gives students time to reflect on what they’ve learned.

The initial attempt gives the later explanation a purpose. A student who has tried an approach can examine where it worked, where it failed, and what a different method makes possible.

Manu Kapur’s research on productive failure explores a related sequence of problem solving followed by instruction. In his 2014 studies on learning standard deviation, students who attempted problems before instruction performed better on measures of conceptual understanding and transfer than students taught first. Both groups showed comparable procedural knowledge.

That finding gives me a reason to include carefully chosen problems before an explanation. The follow-up teaching matters: an unsuccessful attempt needs to be revisited so the learner can understand what went wrong. I don’t take it as a reason to leave someone stuck indefinitely.

Merrill’s first principles of instruction

M. David Merrill’s 2002 paper looked for principles shared by several instructional design theories. He described five: learning centred on real-world problems, activation of prior knowledge, demonstration, application, and integration.

The last four form phases around the problem. Learners recall relevant experience, see how a task is done, attempt it themselves, and connect what they’ve learned to their own use of it.

I use this as a check on the whole lesson. What problem gives the material a purpose? What does the learner already know? Where do they get to practise, and how will they use the result?

Merrill’s paper is a synthesis of design theories, rather than an empirical test of all five principles. It gives me a way to review a lesson; the lesson still needs to work for the person using it.

What this means for Physica

These frameworks help me make different decisions. POE shapes an individual activity. Gagné helps me check the support around it. The learning cycles help me think about the order of the lesson and what happens when someone returns to it.

I’m using a nine-stage sequence to guide Physica’s development, drawing especially on 7E, POE, and productive failure:

1. Activation: ask what the learner already believes about the topic.
2. Provocation: introduce a phenomenon that gives them a reason to question that belief.
3. Exploration: let them change a simulation’s inputs and observe what happens.
4. Prediction: ask for a specific estimate, then compare it with the result.
5. Intuitive explanation: explain the behaviour in plain language and address the misconceptions that arose.
6. Formalization: introduce the mathematics and explain the reasoning behind each step of the derivation.
7. Near application: practise with problems similar to the example.
8. Far application: work through a problem where the same idea appears in a different setting.
9. Metacognitive reflection: revisit the initial prediction and identify what remains unclear.

This is my adaptation for a site someone can work through on their own. It gives me concrete questions to ask while building a lesson: did I give the learner enough to attempt the problem? Did the explanation address what they observed? Can they use the idea without following the worked example line by line?

Those are the checks I want to make before calling a Physica lesson finished.