Analogical Problem Solving
Before you start
A reading task, then a problem to solve.
This takes about fifteen minutes. There are two parts, and they are scored separately.
Work through it in one sitting, and please don't look anything up or ask anyone — including a chatbot. The result only tells you something if you go in cold. There are no trick questions and nothing here is timed against you, though the page does quietly record how long each part takes.
Everything you type stays in your browser. Nothing is sent anywhere. At the end you'll get a summary you can copy and bring to your session.
Part one · Reading
Read the passage.
Read carefully. When you're done you'll be asked to retell it from memory — the passage will disappear at that point — so read for the shape of what happens, not the details.
A small country was ruled from a strong fortress at its center. The fortress sat in the middle of the country, surrounded by farms and villages, and many roads radiated out from it like spokes on a wheel.
A general resolved to capture the fortress. He knew that an attack by his entire army would take it easily — his force was more than strong enough. But he learned that the dictator had mined every road.
The mines were set so that small groups of men could pass over them safely, since the dictator needed to move his own troops and workers. Any large force, however, would detonate them. That would not only destroy the road, it would devastate the neighboring villages.
The general could not send his whole army down any one road. So he divided the army into small groups and sent each group to the head of a different road. When everything was ready, he gave the signal, and each group marched down its own road. Every group reached the fortress at the same moment. In this way the general captured the fortress.
Short break
Clear your head for a moment.
A quick unrelated task. Rate each word for how pleasant it feels to you. There are no right answers and this isn't scored.
Part two · Problem
Here is the problem.
Suppose you are a doctor treating a patient with a malignant tumor in his stomach. Operating is impossible, but unless the tumor is destroyed the patient will die.
There is a kind of ray that can destroy the tumor. If the rays reach the tumor all at once at a high enough intensity, the tumor is destroyed. Unfortunately, at that intensity the healthy tissue the rays pass through on the way in is destroyed as well.
At lower intensities the rays are harmless to healthy tissue — but they leave the tumor untouched.
What procedure could be used to destroy the tumor with the rays without destroying the healthy tissue around it?
One question first
While you were working on that, did anything come to mind?
Be honest either way — there's no better answer here, and this question is the whole point of the activity.
Second attempt
Try the problem once more.
In solving this problem, you may find that the passage you read earlier suggests a solution.
The tumor is in the stomach. Operating is impossible. High-intensity rays destroy the tumor but also the healthy tissue they pass through. Low-intensity rays are harmless to tissue but don't affect the tumor.
How can the tumor be destroyed without harming the healthy tissue?
The answer
Send many weak rays from many directions at once.
Each ray is too weak to harm the tissue it passes through. They all converge on the tumor, and only there do they sum to full strength.
Why the passage was relevant
| In the passage | In the problem |
|---|
Nothing on the left resembles anything on the right. Armies aren't rays and fortresses aren't tumors. What carries across is the structure: a force that must arrive whole at one point, but cannot travel whole along any one path. That's what an analogy actually transfers — not the things, the relationships between them.
Scoring
Mark your own answers honestly.
Nobody sees this but you and whoever you show it to. An inaccurate score here just makes the result useless.
Your first attempt — before the hint
Your second attempt — after the hint
Result
Your result
What this experiment found
Gick and Holyoak ran this in 1980. Given the passage and then the problem with no hint, roughly a third of people produced the converging-rays solution. Told that the passage might help, roughly three-quarters did.
The knowledge was already there in both groups. What the hint supplied wasn't information — it was retrieval. The gap between those two numbers is the size of a problem that has nothing to do with how smart anyone is.
Which is why the 1983 follow-up matters
Gick and Holyoak went looking for something that would raise unhinted transfer. What worked was giving people two stories from unrelated domains and asking them to say what the two had in common. Comparing two cases forces the shared structure into the open, where a single case leaves it buried under its own details. One example teaches the example. Two examples teach the pattern.
Take this with you
Copy this summary and bring it to your session.
Notes for whoever is administering this
Conditions. Add a query string to the link to set which version a student gets:
?c=one— single source passage (the General). This is the default and the canonical 1980 condition.?c=two— both passages plus the comparison prompt. The 1983 schema-induction condition. Expect higher unhinted transfer.?c=control— no passage at all. Baseline solution rate for the problem by itself, with the hint step skipped.?c=random— assigns one of the three at random and records which.
The measure. The number that matters is not whether they solved it. It's unhinted versus hinted. A student who converges only after the hint has the knowledge and lacks the retrieval cue — which is precisely what a calendar, a notes system, or a worked-example file is for. That reframe is the point of running this.
Verbatim answers are in the summary. Self-scoring is unreliable in both directions, so the export includes everything they typed. Score it yourself against the convergence criterion: multiple lower-intensity rays, from different directions, arriving at the tumor simultaneously. Partial credit is worth tracking — "operate through the esophagus" and "lower the intensity and go slowly" are the classic non-convergence answers and both show real problem-solving.
No data leaves the browser. This page has no backend. If you want responses collected automatically, the export block is plain text and can be posted to a form endpoint with about ten lines of JavaScript — but as written, nothing is stored or transmitted, which is also why a student can close the tab and lose everything. Tell them to copy the summary before they leave.
Don't reuse it on the same student. Like any transfer problem, it works exactly once.
Adapted from Gick, M. L., & Holyoak, K. J. (1980), Analogical problem solving, and (1983), Schema induction and analogical transfer. The tumor problem is originally Duncker's (1945). Passages here are retellings, not the original texts.