Back to blog

You Understand the Concept but Still Can’t Solve the Problem: You’re Missing a “Problem-Type Mapping Table”

You understand the concept but still can’t solve the problem: you’re missing a “problem-type mapping table.” “I understand this concept”—I’ve said this many times in class. Then when I do problems, I get stuck on the very first one. I check the answer: I recognize every step, but I can’t think of th

AI office assistantConversation + AgentLiterature Retrieval AI

You Understand the Concept, But Still Can't Solve the Problem: What's Missing Is a "Problem-Type Mapping Table"

"I understand this concept"—I've said this many times in class.

Then I start doing problems, and I get stuck on the first one. I look at the answer; I recognize every step, but I couldn't come up with the first step myself.

The problem isn't the concept; the problem isa missing bridge between the concept and the problem type.

"Understanding" Just Means "Being Able to Repeat It"

Understanding in class usually means "being able to nod along with the teacher's derivation." That is passive understanding.

What solving problems requires is: when you see the prompt, you can judge "what type this belongs to and what to do first." That is active recall.

Between the two, there is one missing step:translating the concept into "use it when you encounter this situation."

Three Columns of the Mapping Table

What I do now is make a table for each chapter:

Concept/MethodWhat It Looks Like in a ProblemFirst-Step Action
Equilibrium condition"At rest," "constant velocity," "just about to slide" appearDraw a free-body diagram and set up equations for two directions
Limiting state"Maximum," "minimum," "just" appearTurn the inequality into an equation
Superposition principleMultiple independent effects appearSplit into single factors, calculate each separately, then add

Of the three columns, the second is the most important:write "what it looks like in a problem," not "what the definition is."

How to Write the Second Column: Copy from Prompts, Not Textbooks

When I write the second column, I go through three problems I've done and copy down the trigger words.

For example, the phrase "just about to slide" is easier to remember than "static friction reaches its maximum," because those are exactly the words that appear in exam problems.

After one chapter, the second column usually has six to ten trigger words.These words are the key signals for that chapter's problems.

Step Three: Practice in Reverse

After finishing the table, I don't practice by chapter; I practice by "trigger word": pick a trigger word at random and think about which concept it corresponds to and what the first step is.

If I can't recall it, that row isn't memorized yet.

For this step, I have MuchSci's chat + agent generate problems: paste in the mapping table and ask it to "create one problem for each trigger word that tests only the first step."

The problems MuchSci generates usually aren't hard, but they test exactly "what to do when you see this sentence," which is more effective than my just doing a whole test paper.

Three Types of Problems, Three Kinds of First Steps

To sum up, the problems in the first chapter are basically three types:

1. Determine the type: given conditions, ask "will it happen?" The first step is to find the critical condition

2. Find numerical values: ask "how much," and the first step is to set up equations and count the unknowns

3. Explain a phenomenon: ask "why," and the first step is to find the relationship between variables

The first step is different for each type. Before solving, determine the type first; it's faster than calculating directly.

A Real Improvement

For the same chapter content, before the change, I got four out of ten problems wrong, and every mistake was "I didn't know where to start."

After the change, I got one out of ten wrong, and that one was a calculation error.

The difference isn't the amount of knowledge; it'sthat I now have the action of "judging the type first when I see a problem.".

A Little Reflection

Between "understanding" and "being able to solve," there is a layer of translation: from the language of concepts into the language of problems.

That is exactly what the mapping table does. It doesn't add knowledge; it only adds "retrieval paths."

I usually fill in this table myself, because the trigger words have to be copied from problems I've done. After filling it in, I have MuchSci's chat + agent generate problems by trigger word for practice, and if I get something wrong, I add another row. After a semester, this table is more worth reviewing before an exam than my notebook—it doesn't record knowledge; it records my own "problem sense."


Back to articles