An economics student reads, “When interest rates rise, the price of existing bonds falls,” nods, underlines it and moves on. Two weeks later the exam asks: “Why would a pension fund holding long-dated bonds be worried about a rate rise?” He remembers the underlined sentence and has no idea where to start. The sentence was correct. What he lacked was the mechanism behind it.
Self-explanation is the habit of pausing after each important idea and answering for yourself: why is this true, what does it connect to, and what would change if one condition changed? It sounds almost too simple, yet it is one of the better-evidenced techniques in learning research, and one of the most often done badly.
Memorising a sentence versus understanding it
Memorising means keeping the shape of the sentence. Understanding means being able to rebuild it from other pieces. A student who understands bonds can reason: old bonds pay a fixed coupon; new bonds now pay more; nobody will buy the old bond at its old price; so its price has to fall until its yield matches. From that chain, the pension-fund question answers itself, even though the textbook never mentioned pension funds.
Self-explanation forces you to build that chain while you read, instead of hoping it appears in the exam hall.
Why it works
- It exposes gaps. You cannot explain a step you do not understand without stumbling, and that stumble is valuable information that rereading never produces.
- It links new knowledge to old. Memories with many connections are easier to retrieve than isolated ones.
- It gives every step a reason. When the exam changes the context, you lean on the reason rather than the wording.
Prompt sets by type of material
| Material | Questions to ask yourself |
|---|---|
| Concepts and theory | Why is it so? How does it differ from its nearest neighbour concept? What looks like an example but is not one? |
| Processes and mechanisms | What does this step need from the one before? What breaks if it is skipped? Could two steps swap order? |
| Worked solutions | Why did the solver choose this move here? What in the question signalled it? Which number, if changed, would make this approach fail? |
| Code | Which variable does this line change? What happens on empty input? Why does the loop stop here? |
| Cases and scenarios | Which fact decided the outcome? Remove it: does the conclusion change? |
Five shallow habits that neutralise the technique
- Paraphrasing in the book’s own words. If your explanation is the textbook sentence with a few words shuffled, you are repeating, not explaining.
- Explaining only what you already get. It feels good to explain the easy parts. Linger longest where you most want to skip.
- Never checking. After explaining, reopen the source. A wrong explanation repeated a few times hardens into a confident misconception.
- Stopping after every sentence. That is slow and exhausting. Stop at the end of paragraphs that introduce something new, at each move in a solution, at each figure.
- Keeping it in your head. Silent thinking lets vagueness slide. A line in the margin or a sentence said aloud does not.
The beginner test: a stretched version
What many people call the Feynman technique is self-explanation with an imagined listener. You explain the idea to someone who has never taken the course, in plain language, using no jargon you have not first explained. Wherever you find yourself saying “it just is”, go back to the book.
A useful variant for study groups: each person explains one section to the others, and listeners may only ask “why?”. After three whys in a row, most shallow explanations come apart.
In practice: one textbook page in 25 minutes
- Minutes 0 to 3: skim headings, figures and first sentences to see which question the page answers.
- Minutes 3 to 15: read paragraph by paragraph. After each, write one line in the margin answering “why is this true?” or “what does this connect to?”.
- Minutes 15 to 20: close the book and say or write the main line of reasoning for the whole page in your own words.
- Minutes 20 to 25: reopen, compare, and correct errors or gaps in a different colour. Those corrections are your review list for next time.
For worked examples, replace step two with covering the lines below and predicting the next step before you look, along with the reason you expect it.
When not to use it
Self-explanation costs more effort than ordinary reading. For information with no internal logic, such as a date, the name of a constant or isolated foreign vocabulary, asking “why?” leads nowhere. Those items suit self-testing and spaced review better. Save self-explanation for material with mechanisms, arguments and cause and effect.
This week, pick a chapter you “understood” but still got questions wrong on. Work through a single page with the 25-minute routine above, count the lines you had to correct in the second colour, and you will see how much your previous reading really left behind.
Câu hỏi thường gặp
How is self-explanation different from summarising?
A summary restates what the text says; a self-explanation says why it is true and how it connects to what you already know. You can summarise by shortening the text, but you cannot self-explain that way.
Do I need to self-explain out loud?
Not necessarily, but speaking or writing beats silent thinking because you cannot skate over vague spots. Some students record themselves on their phone and listen back for hesitations.
What is the Feynman technique?
It is the popular name for explaining a concept in plain words as if to a beginner, noticing where you stumble, going back to the source and explaining again. In essence it is self-explanation with a built-in check.
Does self-explanation take a lot of extra time?
Reading slows down by roughly a third to a half at first. In exchange you reread far less, and what you understand stays with you much longer than after several fast passes.
What if my explanation is wrong?
That is why you check against the text or solution after each section. An unchecked wrong explanation, repeated often, becomes a very confident misconception, so the checking step is not optional.
