How to practice maths and physics when re-reading feels like studying

The gap between following a worked example and solving a new problem is normal, not a character flaw, and it has a mechanical cause. A worked example does the hard part for you before you look — the method is chosen before you see the question — and the exam hands you back exactly that part. Re-reading has never closed the gap, because it re-reads the one thing you were never missing.

So the question is not how to study harder. It is what practice has to look like so that the gap actually closes.

Why does re-reading a worked example feel like studying?

Because following a solution is recognition, and the hardest step in the whole problem — choosing which method applies — was already done for you.

Read a solution and you can see exactly where every line comes from. The method was announced by whoever wrote the worked example, so your brain was doing the easy half of the job: following. In the exam nobody announces the method. You have to pick it from a question you have never seen, and that is a different skill, one that only attempts build.

This is not a new observation. A survey of study techniques by Dunlosky and colleagues in 2013 ranked re-reading and highlighting low on what actually works, while practice testing came out among the techniques with strong utility. In plain words: re-reading feels fluent and delivers little, and the technique that feels worse is the one that pays. Psychologists call conditions like this desirable difficulties (Bjork, 2011) — learning feels harder in the moment and lands better afterwards. The fluent feeling that re-reading gives you is an illusion of fluency: material that looks familiar is not material you can use.

The mechanics of why testing yourself works at all are in this post on active recall. What follows is what the loop looks like when the material is problems, not definitions.

What does effective maths and physics practice actually look like?

It is a loop built from problems, not notes: attempt, check one step, log the error, redo it a few days later.

Here is the loop, in the order it should run:

1. Attempt the problem cold, with a time limit. A fixed time keeps the attempt honest; open-ended "I'll just think about it" quietly becomes re-reading the question for the third time.
2. When you get stuck — and you will — open the solution and read only the first step. Then close it. The first step is usually where the method choice lives, and that is the part you are actually training.
3. Finish the problem, then compare your version with the solution. The comparison is where a lot of the learning happens, because it is where you see the difference between your path and the shorter one.
4. Write one line in an error log: why you went wrong, not what the answer was.
5. Redo the problem from a blank page a few days later. Spacing the redo matters: the review works because it happens after the memory has started to fade, not because of the repetition itself. That is the spacing effect, reviewed by Cepeda and colleagues in 2006, in its plainest form.

Here is what one pass of the loop looks like. The problem: a block sliding down a ramp with friction, given the mass, the angle and the coefficient of friction — find the acceleration. Your cold attempt: you stare at the diagram, write "second law" at the top of the page, and reach for the equations of motion, unsure whether they apply here. Five minutes later you open the solution and the first line is "resolve the weight along the plane" — one line that unblocks everything after it. You finish the rest yourself, and the error log takes one line:

Reached for the equations of motion before checking what the question actually asked — and never wrote down what was given first.

A few days later a ramp-and-friction problem shows up in a mixed set. The log line does its job: you write down what is given before touching any equation, and the resolve-the-weight move is the one you find yourself.

The error log

The log is the cheapest part of the system and the easiest to skip. One line per mistake, and tag the type of error rather than the content of it:

• Method choice — "used the wrong equation; should have resolved the forces first."
• Concept — "assumed energy is conserved when friction was doing work."
• Algebra or arithmetic — "sign error in the final line."
• Units — "mixed joules and kilojoules in the energy balance."
• Misread question — "found the velocity when it asked for the displacement."

Tag the type, not the topic. After a couple of weeks of logging, the pattern is usually the same two or three types recurring, and those are what your next practice session is spent on. A log full of topic names tells you what you have studied; a log full of error types tells you where the marks are actually going.

How do you get better at knowing which method to use?

By mixing problem types within one session, so that choosing the method is a task instead of a default.

Do ten problems of the same type in a row and your brain locks into the method after the first two. The remaining eight train nothing new — you are recognising, not choosing. That is why ten identical problems feel easy and still leave you stuck on the exam question, where the wrapping is unfamiliar and the method has to be chosen again from scratch. Interleaving — mixing problem types so each one forces a fresh choice — is the fix, and it is a desirable difficulty in action: it feels slower and harder, and the slowness is the training, not a sign it is not working.

A before-and-after makes the difference concrete:

• Blocked set: ten projectile problems in a row. By the third, the method is on autopilot; the last seven are mostly bookkeeping.
• Mixed set: one problem from projectile motion, one from circular motion, one from energy — then the next round again. Every single one starts with the choice, and the choice is exactly what the exam tests.

Pair it with a two-second habit: say or write the governing principle before you touch any numbers. "Energy is conserved here, so…", "friction is doing work, so energy is not." The sentence forces the method choice to happen before the arithmetic buries it. It is also the answer to the students who post that exam problems feel unhinged when the textbook examples felt fine: an unfamiliar wrapper plus a required choice of method is the whole design of a hard question, and it is exactly the combination that re-reading never exposes you to.

Is there any place for flashcards in maths and physics?

Yes, but only for the lookup layer — the small slice of the subject that is actually memorisation.

What needs remembering is smaller than it looks: the formulas and, more importantly, the conditions under which they apply; units and dimensions; definitions; and prompts like "which method for this question type." That layer is not where the marks are lost, but an unreliable lookup layer makes every problem slower and more stressful than it needs to be. Write the question on one side and the rule on the other — "conditions for this equation to apply", not just the equation itself.

A pen and index cards do this job completely, and any free flashcard app does it too. If you photograph your formula sheet or the chapter summary page, a photo-to-cards tool like SnapDeck AI saves the typing, which is the whole of its value in this subject. How to make the cards themselves work is not a maths question — it is the retrieval question, and this post covers it.

What if you are out of practice material?

Three sources, in order of usefulness: past papers or practice exams from your own exam system — the closest thing to the real thing, because they test the method choice under unfamiliar wrapping; the end-of-chapter problems in your textbook, the standard supply; and problems you make yourself, by changing the numbers or the setting of any worked example you already have. If that is all you have, the last option is a real one, not a consolation: a worked example with the angle changed is a different problem to your brain, because the choice has to be made again.

If you are short on time, the trade runs the other way: fewer problems, done properly. A session of three problems through the full loop beats an hour of problem-staring that skips the log and the redo. And if the pile of problems you still need to do feels like too much to start, the first-hour plan in this post applies to problem sets exactly as it does to notes.

The short version

• Attempt problems cold, with a time limit. Re-reading solutions trains recognition, not choice.
• Check only the first step of a solution when you are stuck; the method choice is the part you are training.
• Log the type of every error in one line. After a couple of weeks, the pattern shows you where the marks are going.
• Mix problem types within a session, and name the principle before you touch the numbers.
• Use cards for the lookup layer only — formulas, conditions, units, which-method-when.

The uncomfortable part of all this is that it works by making studying feel less like studying. Fewer hours of fluent re-reading, more hours of slow attempts and logged mistakes, and the feeling that you are getting nowhere is the fee, not the verdict. The loop is small enough to run on your worst day: one problem, one log line, one redo. Start with the problem type you find hardest, not the one that feels safest, and let the log tell you where the next session goes.