The most underused teaching method in physics is also one of the oldest and best documented. Predict-Observe-Explain — POE — was formalised by Australian researchers Richard White and Richard Gunstone in their 1992 book Probing Understanding. Thirty-plus years later it remains the gold standard for shifting the wrong intuitions that wreck physics learning.
Why isn't every classroom using it? Mostly because POE is hard to do in a class of thirty — each student needs to commit to their own prediction, hear back, and reconcile. In a class, half the children copy the louder neighbour and the move collapses. POE was designed for one-to-one. One-to-one is exactly what AI tutoring is. That's why Newton runs it by default.
The three stages, properly defined
Predict
A prediction is not "I think maybe…". It has two parts: a specific commitment ("the bowling ball hits first") and a reason ("because heavier things fall faster"). Newton refuses to proceed without both. If the student says "I don't know", Newton offers a binary — "same time or different? Pick one." Almost every student has a leaning; they're hiding it because they're scared of being wrong. Non-physicists most often skip this step — they think prediction is the warm-up to the lesson. The prediction is the lesson.
Observe
Whatever shows the student what actually happens. Newton uses three modes: calculation (student plugs numbers in and sees the answer), real-world reference (Apollo 15 hammer-feather drop), or household experiment ("drop a coin and a piece of paper — now scrunch the paper and drop again"). The observe step has to deliver a clear, unambiguous result. Wishy-washy observations don't break wrong intuitions.
Explain
Where the learning consolidates. The student articulates the physics — own words first, Newton sharpening. If the prediction was right: "What's the underlying principle?" — putting words to implicit intuition makes it transferable. If the prediction was wrong: "Where did your reasoning break? What did you assume that turned out untrue?" The student finds the broken assumption and fixes it. Wrong predictions are more valuable than right ones — they're where the learning lives.
A fully worked POE: dropping a tennis ball
Year 7 student. Topic: gravity and falling objects. Energy: green.
Newton: I'm going to drop a tennis ball off the edge of the kitchen table. How long do you think it takes to hit the floor? Just guess — what's your gut?
Student: Um, half a second?
Newton: That's your prediction. Now let's make it interesting — if I drop a bowling ball off the same table, longer, shorter, or the same time?
Student: Shorter. Because it's heavier — heavier things fall faster.
Newton: Good — committed prediction with reasoning. Now let's observe. Table is 0.8 m, gravity is 10 m/s². Use s = ½ × a × t². Try it.
Student: 0.8 = 0.5 × 10 × t² → t = 0.4 seconds.
Newton: Right. Now — where does mass appear in that equation?
Student: It doesn't.
Newton: So for the bowling ball: same calculation, same answer. And here's the Apollo 15 astronauts dropping a hammer and feather on the Moon — they land together. Your prediction said the bowling ball lands first. Where did the reasoning break?
Student: I thought heavier means more gravity pulling it. But it's the same gravity for both.
Newton: Almost. There is more gravity on the bowling ball — its weight is bigger. But that gravity also has to move a bigger mass. The two cancel. Heavier = more force and more mass to move. Same acceleration.
Student: Oh.
Newton: One last thing — on Earth, why do feathers really fall slower than hammers?
Student: Air resistance.
Newton: Exactly. In a vacuum they fall together. On Earth, air resistance slows the feather. Your intuition wasn't crazy — just for a different reason than you thought.
That's POE. Twelve minutes. The student now has a model of gravity that distinguishes between weight (a force) and acceleration (the same for all objects in vacuum). They also know why air resistance complicates things. They got there by being wrong, noticing they were wrong, and repairing the model themselves.
A traditional explanation-first lesson would have delivered the same information in eight minutes and left half the misconception intact.
Common POE failure modes
- Dodging the prediction. "I dunno, you tell me." Newton offers a binary: pick A or B. Almost always cracks the deflection.
- Prediction without reasoning. Newton presses for the principle, or suggests two competing reasons and asks which feels more likely.
- Observation too messy. Avoided by picking demonstrations and calculations that produce unambiguous results.
- Right prediction, no explanation. Don't let this slip — the explain step matters even when the prediction was right.
POE topics across KS3 physics
POE works best where intuition is most likely to mislead:
| Topic | Wrong intuition | POE moment |
|---|---|---|
| Falling objects | Heavier falls faster | Apollo 15 hammer-feather |
| Forces and motion | Force needed to keep things moving | Hockey puck on smooth ice |
| Current in circuits | Current "used up" by bulbs | Two-bulb series measurement |
| Heat vs temperature | Bigger = hotter | Sparkler vs warm bath |
| Floating and sinking | Heavy things sink | A 200-tonne ship doesn't |
| Pressure | Pressure = force | Snowshoes — same force, different pressure |
| Sound | Sound travels in space | "In space no one can hear you scream" |
Why this works — the underlying psychology
The cognitive science term is cognitive conflict. Until the student notices their model is broken, the model doesn't get repaired. Lectures don't generate the conflict; the student just adds the new fact and keeps the wrong bit running underneath. Predicting publicly, then being shown the prediction was wrong, forces the conflict into the open. The model fixed under that pressure is more durable than any model passively received. POE works in any subject where wrong intuitions are common — Darwin runs analogous "misconception probes" in biology.
What POE doesn't do
POE doesn't help much with pure computation, definitions and vocabulary, or drawing diagrams — those just need practice. Newton uses POE for conceptual physics and other moves (ICUCS, formula triangles, drill) for the computational bits.
FAQ
Why does the student have to predict before observing?
Because predicting surfaces what the student already (wrongly) believes. Without it, the observation just becomes a new fact added on top — the wrong intuition continues running in the background.
What if the prediction is right?
Then the explain step is about articulating why — making the implicit physics explicit. Students often get predictions right by feel without knowing why; POE forces them to put words to the reasoning.
Does this take longer than just teaching the concept?
Per concept, slightly — maybe 25 minutes instead of 15. But you teach it once. Traditional teaching has to come back to the same misconception multiple times. Net, POE is faster.
What if my child says 'I don't know' instead of predicting?
Newton offers a binary: 'OK, which is more likely — A or B?' Forcing the choice cuts through the deflection.
Related reading
- Meet Professor Newton: KS3 physics that makes you predict first
- Why Newton teaches forces, then energy, then motion (in that order)
- The Show Your Working protocol behind the faculty
Jason runs aitutors.me. He once spent two hours convinced his Year 8 was wrong about a pressure question and then realised he'd just rediscovered why POE works. Updated 21 May 2026.