Change the form

If you can explain an idea in one way, does that mean you understand it — or only that you can produce that one explanation?

Week 7 · A self-guided lesson

Where Week 6 left us

Week 6 changed the resources available during a performance. This week keeps attention on the idea but changes its form: audience, representation or problem. It tests a different ladder assumption: that an Explanation-rung response reflects an adaptable model rather than a rehearsed form. Our new fictional example supplies the background it needs.

How to use this lesson

Read at your own pace; no presenter is needed. Allow around 35 minutes, with paper or a text document to write short answers.

At a Pause and write slide, attempt the question before advancing. The next slide explains the reasoning so you can compare it with yours.

No finance background is required. Kavi, the colleague, and every response quoted are fictional.

The situation: an explanation that already works

Kavi read a short explanation of compound interest — how savings or debt can grow faster than a fixed yearly amount suggests — and wants to check what they actually understand, not just what they can repeat.

Kavi already has one fluent explanation ready. This week asks whether producing it again, or something like it, tells us anything new.

First, the idea itself

Compound interest is interest calculated on the original amount and on interest already added, not only on the original amount. The Federal Reserve Bank of St. Louis describes it as interest “on your original principal — plus on the interest your investment generates”: interest earning interest.

Federal Reserve Bank of St. Louis, “How Compound Interest Works” (2018). Technical grounding only.

A small worked example

Put $1,000 in an account paying 10% a year, compounded annually.

Year 1: $1,000 → $1,100. Year 2: $1,100 → $1,210. Year 3: $1,210 → $1,331.

Compare that with simple interest, which adds a fixed $100 every year regardless of the balance: $1,000 → $1,300 after three years. The extra $31 in the compound case is interest earned on interest already added — the mechanism, not a rounding error.

What we mean by “a different form”

A performance changes form when it addresses a different audience (a child, a colleague), uses a different representation (a story, a written justification, numbers), or answers a different problem posed from the same idea (explain it, apply it, compare two cases, find an error).

Last week asked what you are allowed to use. This week asks what you are asked to do with what you know.

Kavi’s first performance

Kavi’s younger sibling asks how a savings account works. Kavi explains: “Every year, the bank pays you interest. But next year, you also get paid interest on that interest — so your money is a bit like a snowball. It picks up more the further it rolls, even without you adding anything.”

The sibling nods and repeats the snowball line back. Kavi seems pleased.

Pause and write · 3 minutes

Read Kavi’s snowball explanation again. Is this strong evidence that Kavi understands compound interest, rather than a comfortable, well-rehearsed line?

Write two sentences: what the explanation shows, and one thing it does not yet show. Consider what else could produce the same fluent explanation. Advance when you have an answer.

Worked response: fluent is not the same as flexible

The explanation is accurate and clearly delivered, which is worth something — a confused explanation would be worse evidence, not better. But it is also a well-worn analogy that could have been read, memorised or reused nearly unchanged.

We have seen this limit before. Week 2’s illusion of explanatory depth showed that a fluent explanation can still conceal what a person cannot yet do with an idea.

What would tell us more

If Kavi’s explanation only ever reappears in this one shape — same audience, same story, same numbers — we cannot tell whether Kavi possesses the underlying relationship or one rehearsed description of it.

One option: ask Kavi to use the same idea somewhere the snowball line does not fit.

A name for this: transfer

Educational psychologists call this transfer: using something learned in one context to perform in a different one. Barnett and Ceci’s review distinguishes transfer to a similar context (near transfer) from transfer to a genuinely dissimilar one (far transfer) — different in domain, audience, format or the kind of problem asked.

Barnett & Ceci (2002), Psychological Bulletin 128(4), 612–637. A synthesis of the transfer literature, not a single experiment.

Why the distance matters

Barnett and Ceci argue that disagreement over whether people “really” transfer what they learn comes from treating all transfer as one thing, when studies differ enormously in how far the new context sits from the original. A performance repeated in a near-identical context is much weaker evidence of flexible understanding than the same performance produced somewhere genuinely different.

What this looked like in practice

Barnett and Ceci report a study (Schliemann & Magalhães, 1990) of unschooled Brazilian cooks solving proportional-reasoning problems — the same underlying mathematical relationship — across three domains. The cooks performed almost perfectly on pricing problems, worse on recipe problems, and worse again on medicine-dosage problems: an unfamiliar domain, even though nothing about the maths itself had changed.

What that does and does not show

The same competence produced very different performance depending on how far the new problem sat from familiar territory. That cuts both ways: a person who fails an unfamiliar version has not necessarily failed to understand the relationship — unfamiliarity with the new domain is doing some of the work too. Changing the form is informative, but it is not a clean test on its own.

Back to Kavi: an unexpected question

A colleague overhears the snowball explanation and asks something different: two loan offers quote the same 6% yearly rate — one compounds annually, the other monthly. Which one will cost more in total interest, and why?

This is not a repeat of the sibling conversation. It changes the audience, drops the analogy, and asks for a comparison rather than a description.

Pause and write · 4 minutes

Before reading Kavi’s answer, attempt the colleague’s question yourself: will monthly or annual compounding cost more at the same 6% nominal rate, and why?

Write your prediction and your reasoning in two or three sentences. You do not need exact numbers, only the direction and the mechanism. Advance when you have an answer.

Kavi’s answer: two versions

Version A

“Monthly compounding should cost less. Each month’s interest is worked out on a smaller amount, calculated more often, in smaller pieces — so it adds up to less overall.”

Version B

“Monthly compounding should cost slightly more. Interest gets added to the balance twelve times a year instead of once, so interest starts earning interest sooner and more often — even though each individual addition is smaller.”

Worked response: check it with numbers

On a $10,000 loan at 6% for three years: compounded annually, it grows to about $11,910. Compounded monthly, it grows to about $11,967 — roughly $57 more.

Version B is correct. Version A repeats the surface feature of the sibling explanation — “interest earns interest” — without following where that mechanism actually leads: more compounding events, not smaller amounts, is what raises the total.

What version A actually tells us

Version A is not confused about the definition; it can even quote the snowball line correctly. What it has not done is use the underlying relationship to reason about an unfamiliar comparison — which is closer to Barnett and Ceci’s fishermen, who could apply an intuitive method inside their trained domain but not carry a school-taught procedure into a new one.

Fluent restatement and working transfer can come apart.

What version B’s success shows — and its limit

Version B reasons from the mechanism to a new, previously unseen comparison and gets the right answer for the right reason. That is stronger evidence of a working model than the sibling explanation alone.

But one success in one new form is still one data point. We do not yet know whether Kavi could handle a third, different reformulation — or whether this exact comparison was something Kavi had already seen.

The deeper problem with “change the form”

Reformatting and adaptation can themselves be rehearsed. If loan-compounding comparisons are a common example in whatever Kavi read, a correct answer here might reflect a second memorised template rather than a freely applied relationship — apparent transfer that actually rests on prior familiarity with this specific kind of question, not on the underlying idea.

What we still cannot tell

From the transcript alone, we cannot distinguish “Kavi reasoned from the mechanism to a genuinely new case” from “Kavi had met a version of this comparison before.” Both would produce exactly Version B. Changing the form raises the bar on what a rehearsed answer would need to cover — it does not remove the possibility of rehearsal.

Pause and write · 6 minutes

Pick one thing you already feel you understand — anything, from your own study or interests. Write a short explanation of it as you would to someone with no background in the topic.

Then write a second explanation of the same idea, aimed at someone who would ask a harder, more technical question about it — a critique, an application, or a comparison. Use a genuinely different form, not just shorter words.

Worked response: check your own pair

Compare your two explanations. Did the second one restate the first with different vocabulary, or did it use the underlying idea to do something new — justify, compare, predict, or defend against an objection?

If the second explanation could have been written by someone who only memorised the first, the change of form has not yet told you anything new about your own understanding.

Keep a short record

Write three sentences: Observation — what did the two explanations actually say? Interpretation — what does the harder-form version specifically support, that the first version alone did not? Limit — what would a third, still more different form need to show, that this pair does not yet establish?

This is the same evidence record you have used since Week 1 — audience and form are simply new conditions to record.

Changing the form is not the same as removing preparation

Every task in this lesson — Kavi’s, and your own — was still something you knew to expect, inside a lesson built to prompt exactly this kind of answer.

What Week 8 asks

If Kavi, or you, knew roughly what kind of question was coming, some of what looked like flexible reasoning could still be a rehearsed strategy for producing convincing-looking answers to unexpected-seeming questions — not evidence about the specific case asked.

Week 8 asks what changes when the test is not merely unfamiliar in form, but genuinely unannounced — and what a surprised response can show that a prepared one cannot.

What to take from Week 7

A performance repeated in the same form is weak evidence of understanding; the same idea produced for a genuinely different audience, representation or problem is stronger evidence, because it demands more of the underlying relationship.

But “different form” is not automatically “unrehearsed.” Treat a successful reformulation as evidence of something, then ask what specifically it rules out — and what it still leaves open.

Read further: transfer

Barnett & Ceci (2002). “When and Where Do We Apply What We Learn? A Taxonomy for Far Transfer,” Psychological Bulletin 128(4), 612–637. Ask: which of their nine dimensions changed between Kavi’s two performances, and which stayed the same?

This reading deepens the lesson; you can complete the activities without it.

Sources and your next step

Technical grounding: Federal Reserve Bank of St. Louis, “How Compound Interest Works”. It supports the arithmetic only, not any claim about learning or understanding.

Kavi, the colleague and every quoted response are fictional teaching examples, not real transcripts or research data.

For the full written investigation, open the Week 7 investigation. The lecture notes give a continuous reading version.