Study Techniques

Two findings in this area are solid enough to build a week around: pulling information out of your head beats putting it back in, and the same total hours spread over days beat the same hours in one sitting. Almost everything else you have been told about studying is weaker than it sounds, and a few pieces of it are simply invented.

Updated 2026-08-28Source: Cognitive and educational psychology literature on the testing effect, distributed practice and interleaving; published reviews comparing the relative strength of common study strategies
The short versionBuildFigure
StrongestRetrieval practice — self-testing without the page
Also strongSpacing the same hours across days
UnderusedInterleaving mixed problem types
Weak, and popularRereading, highlighting, recopying notes
Not realLearning styles, the "10 percent of what you read" pyramid
The trapEase of reading feels like knowing. It is not.

The ranking almost nobody uses

When researchers line up common study strategies and compare them on the same material with the same students, the ordering that comes out is close to the reverse of what students report doing. Self-testing and spreading practice over time sit at the top. Highlighting, rereading and recopying notes sit at the bottom, and they are the three most commonly reported behaviors in survey after survey. That is not laziness but a specific illusion, which gets its own section below. First, the ranking.

TechniqueSupportWhere it fitsCaveat
Retrieval practice — recalling with the material closedStrong, replicated across ages and subjectsNearly everything, from vocabulary to concept-heavy scienceUseless without feedback. Recalling a wrong answer and never checking cements the wrong answer.
Spaced practice — same hours, more daysStrongAnything with more than two days leftRequires lead time you may not have. Cannot be applied the night before.
Interleaving — mixing problem types in one sessionModerate to strong, mostly in math and category learningProblem sets where recognizing which method applies is the hard partAccuracy during practice drops. It looks like it is failing while it works.
Elaborative interrogation — asking why this is trueModerateWhen you already know enough to ask a real questionWith no background knowledge you cannot generate useful questions
Concrete examples for abstract rulesModerateDefinitions, principles, legal or economic conceptsA wrong example gets memorized as reliably as a right one
Summarizing in your own wordsMixed; depends heavily on skill at summarizingReaders who have been taught to summarizeCollapses into copying phrases, at which point it does nothing
Highlighting and underliningWeakMarking a place to come back toThe act of marking gets counted as studying
Rereading the same pagesWeak past the second passFirst pass, to get the shape of a chapterThird and fourth passes return very little for the time

One honest qualification: effect sizes here vary a lot by subject, by how much the learner already knows, and by how the final test is written — a test that asks for recognition flatters rereading, one that asks for production flatters retrieval. Treat the column as a direction, not a measurement.

Retrieval practice, stated as one instruction

Close the book. Write or say everything you can about the topic. Then open it and find what you missed. That is the entire technique, and the version people get wrong keeps the book open.

The mechanism appears to be the effort of the search itself. Retrieving something on the edge of accessible changes how easily it comes back next time, in a way that reading it again does not. This produces a pattern that surprises people: a group that studies a passage twice usually feels more confident than a group that studies it once and then takes a practice test, and on a delayed test the practice-test group nearly always wins. Confidence and performance move in opposite directions here.

Practical shapes it can take: a five-minute brain dump on a blank sheet before you open the chapter; covering the right-hand column of your notes and reconstructing it; writing three exam-style questions after each lecture and answering them two days later. Cards work too, and the flashcard maker will build a deck you can print, but a deck is one delivery mechanism for retrieval, not the thing itself.

The feedback half is not optional. Getting an item wrong and then seeing the correct answer is productive. Getting it wrong and moving on is worse than not testing at all, because you have now practiced retrieving an error.

Spacing, and why the difficulty is the point

Six hours on Sunday and one hour on each of six days are the same six hours. On a test three weeks later they are not the same at all, and the split version usually wins by a wide margin. The reason is bound up with retrieval: after a gap, bringing the material back takes work, and the work is what strengthens it. Restudying something you saw eleven minutes ago takes none and does very little.

This has an uncomfortable corollary. If a review session feels easy, it was probably too soon; the value is in the strain of almost-forgetting, so the schedule you would design for comfort is the wrong schedule. Spaced repetition covers how to set the intervals and what to do with a lapsed item, and the spaced review planner will lay the dates on a calendar. For a fixed amount of material and a fixed deadline, study plan calculator divides pages or lectures across the days you have, and study hours target works backward from a total hours goal.

Interleaving, which feels like it is going badly

Blocked practice means twenty problems of type A, then twenty of type B. Interleaved means the forty shuffled. During practice, blocked wins: more right, faster, and you feel competent. On a later test, interleaved usually wins, sometimes by a lot.

The explanation that fits best is that blocking removes the hardest step. When every problem on the page uses the same method, you never practice deciding which method to use — you practice executing one you were handed. On a real exam nobody labels the problem. Interleaving forces that discrimination on every item, which is why it is slower in the moment.

It applies where categories are confusable and the choice between them matters: derivative rules, titration types, statistical tests, verb conjugation classes, differential diagnosis. It applies much less to a single set of foreign vocabulary, where there is no method to select. And it has a limit — if you have not learned method A at all, shuffling it with B is not desirable difficulty, it is noise. Block until you can execute, then mix.

Things that are widely taught and are not true

ClaimStatus
People are visual, auditory or kinesthetic learners, and teaching to the matched style improves resultsA myth. The specific prediction — that matching instruction to a measured style beats mismatching it — has been tested repeatedly and does not hold up. Preferences are real; the performance benefit is what fails to appear.
You remember 10 percent of what you read, 20 percent of what you hear, 90 percent of what you teachFabricated. The numbers trace to a diagram that carried no data, and the round figures with no error bars are the tell. Teaching material does help, but not because of that pyramid.
Left-brained and right-brained peopleNot a real distinction as used. Lateralization exists for some functions; personality types built on it do not.
Everyone has one optimal study block length, usually 25 minutesThe 25-minute figure came from one person's kitchen timer, not from a study. The useful part of the method is the uninterrupted single-task block, not the number. Match block length to the task: shorter for drill, longer for extended writing or dense reading.
Background music helps concentrationGenuinely mixed. Lyrics interfere with reading and writing more consistently than instrumental sound does, and the effect varies enough between people that you should test it on yourself for a week rather than trusting anyone's claim.

Fluency, which is the reason the weak methods survive

Read a paragraph three times and the third pass slides by without friction. That frictionlessness gets interpreted, automatically and below the level of decision, as knowing the material. It is not knowing; it is familiarity with the surface of the sentences, and on an exam you are asked to produce content in the absence of the cue.

That mismatch explains most of the gap between how much students study and how much they retain, and why the effective methods are unpopular. Self-testing feels bad, because it shows you what you cannot do. Rereading feels good, because it shows you what you can recognize. The countermeasure is to change what you measure: stop tracking hours, start tracking questions answered from a closed book, correct on the first attempt.

Sleep, cramming and the arithmetic of an all-nighter

Memory consolidation happens partly during sleep, so the night after learning is not neutral time. Cutting it costs you twice: some consolidation does not happen, and you sit the exam with the attention of someone awake for twenty-plus hours.

Cramming is not worthless — massed study produces a short-lived bump, and if the exam is tomorrow and you have nothing, hours beat no hours. What it does not do is survive. Material crammed on Tuesday is substantially gone the following week, which makes it a bad strategy for any subject that builds. An all-nighter trades a small immediate gain for a measurable next-day deficit, usually a losing trade. Sleep hygiene covers the rest; sleep debt calculator puts a number on a bad fortnight.

Assembling a week

A schedule built from the above looks unremarkable, which is a point in its favor. Study new material in blocks sized to the job, ending each with five minutes of closed-book recall. Revisit topics at widening gaps, not in one push. Mix problem types once you can execute each. Test yourself in the format the exam uses. Keep sleep intact through the week before, not just the night before.

Then add the part people skip: record first-attempt accuracy per topic and let it decide where tomorrow's hour goes. Time budgeting tools help with the shape of the day — the time block planner and the pomodoro timer both do it — but no timer knows which chapter you are quietly avoiding. Your error log does.

None of this makes studying pleasant. The techniques with the best evidence feel like failing while you do them, and the ones that feel like progress are mostly not producing much. One line to keep: pick the method that makes you uncomfortable, because comfort is the symptom you were trying to avoid.

Questions people ask

Is highlighting really useless?

Not useless, but much weaker than its popularity suggests. As a way of marking a place to return to, a highlighter is a reasonable tool. As a study method it barely registers in comparison studies, and it has a specific failure mode: marking a page produces the feeling of having worked on it, so the time gets counted as studying when almost no learning occurred. If you highlight, treat it as bookkeeping and follow it with something that requires you to produce the content with the book closed.

How long should a study session be?

Long enough to get into the task, short enough that you are not staring through the page, and matched to what you are doing. Drill and vocabulary tolerate 20 to 30 minute blocks; dense reading and extended writing usually want 45 to 60 because the setup cost is higher. The 25-minute figure attached to the pomodoro method was one person's choice, not a research finding. What matters more than the number is that the block is genuinely uninterrupted — returning from a notification costs several minutes of re-entry, so two hours chopped into eight pieces does less than one clean hour.

Are learning styles real?

Preferences are real. The claim that matching teaching to a measured style improves learning is the part that has been tested, repeatedly, and it fails: learners taught in their supposedly matched style do not reliably outperform learners taught in the mismatched one. The useful version of the idea is that the format should match the material rather than the person — a map is better shown than described, an argument is better read than diagrammed — and that is a different claim entirely.

Does cramming ever make sense?

Yes, in one situation: the exam is imminent, you have not covered the material, and the alternative is knowing nothing. Massed study produces a real short-term bump. The problem is that the bump decays fast, so cramming is a poor strategy for cumulative subjects, for anything you will need next term, and for professional exams that sample broadly. The version that is almost always a bad trade is the all-nighter, which buys a few extra hours of low-quality study and pays for them with degraded attention and lost consolidation on the day it counts.

If interleaving works better, why does my practice go worse when I do it?

That is the expected result and it is not a sign you are doing it wrong. Blocked practice removes the step of deciding which method applies, so you get more right, faster, and feel competent. Interleaving puts that step back in, which lowers your in-session accuracy while training the thing an exam actually tests. The one real caveat: mix only after you can execute each method on its own. Shuffling three techniques you have not learned yet is not productive difficulty, it is confusion.

Related