Chapter 9: Interleaving: Why Mixing It Up Beats Blocking Every Time

Picture a high school classroom, the week before midterms. A math teacher assigns a review problem set covering three topics: linear equations, quadratic equations, and systems of equations.

She assigns the set in two different ways to two different classes.

Class A gets a blocked set: twenty linear equation problems, then twenty quadratic problems, then twenty systems problems. Organized neatly by topic. Easy to work through. By the time they're halfway through each section, they're in a rhythm — they know exactly what approach to use, and they just execute.

Class B gets the same sixty problems, scrambled randomly. Problem 1 is a quadratic. Problem 2 is a systems problem. Problem 3 is linear. No pattern. Each problem requires figuring out what type it is before figuring out how to solve it. Students in Class B have to work harder. Some find it frustrating.

During practice, Class A outperforms Class B. The blocking is working — the students are getting the problems right, building momentum, feeling good about their preparation.

Two weeks later, both classes take the unit exam. The problems are not labeled by type — they appear in random order, just as they would on a real test.

Class B outperforms Class A by a substantial margin.

The students in Class B — the ones who struggled with the scrambled practice set, who found it harder and less satisfying — have dramatically better retention and application of the material.

And when researchers asked both groups which method they thought helped them learn more, the majority of Class B students said the blocked method would have been better.

They were wrong. Their experience of difficulty and their learning were two different things.

Welcome to interleaving — one of the most counterintuitive and important findings in the science of learning.

What Interleaving Is (and Isn't)

Let's be precise, because "mix things up" can mean almost anything.

Interleaved practice means alternating between different topics, problem types, or skills within a single study session rather than completing all material of one type before moving to the next.

The comparison: - Blocked practice (AABBCC): Complete all of Topic A, then all of Topic B, then all of Topic C. - Interleaved practice (ABCABC): Alternate between Topics A, B, and C repeatedly throughout the session.

The key feature of interleaved practice is that you switch before you've finished one topic. This is what produces the difficulty — and the learning.

Interleaving is not: - Randomly jumping around without purpose - Switching topics because you're bored - Studying two completely unrelated things in the same session (though even this may have some benefit) - Never going deep on any single topic

Effective interleaving is deliberate mixing of related material — topics or skills within the same domain that require you to distinguish between them and apply different approaches.

The Research: What We Know and How Confident We Are

The interleaving effect was first clearly documented in academic research on motor learning in the 1970s and 1980s. Researchers studying movement skills found that blocked practice of motor sequences produced better performance during practice but worse retention and transfer than interleaved practice. This "contextual interference effect," as it was termed, was initially confined to physical skills.

Then researchers began asking whether the same effect appeared in cognitive learning — in studying academic material. It does.

Rohrer & Taylor (2007): The foundational study on math interleaving.

This study gave college students two types of practice problems involving geometry (calculating volumes of solids). Some students got blocked practice — all practice problems of the same type together. Others got interleaved practice — problem types mixed throughout the session.

The interleaved students performed slightly worse during practice. On a delayed test one week later, the interleaved students outperformed blocked students by a factor of roughly 2.4 to 1. More than double the retention, using the same total practice time.

[Evidence: Moderate-Strong]

Taylor & Rohrer (2010): Fourth-grade math.

Rohrer and colleagues extended the work to younger students, finding the same interleaving advantage in elementary school math — area and volume problems. The advantage was robust and the effect size substantial.

Kornell & Bjork (2008): Artist recognition.

In this study, participants viewed paintings and tried to learn which paintings were made by which artist. The interleaved group (seeing multiple artists' works mixed together) dramatically outperformed the blocked group (seeing all of one artist's work, then all of another's) on an artist identification test.

This result is particularly interesting because it extends interleaving beyond mathematics — a domain where "different problem types" is a clear concept — to a more naturalistic task requiring discrimination between complex, multidimensional stimuli.

What the research tells us overall:

The interleaving effect is well-established in controlled laboratory and classroom studies. Effect sizes are large — this is not a marginal improvement. The caveat: the effect has been studied most extensively in mathematics and some perceptual learning tasks. Evidence in other domains (languages, essay writing, musical performance) is growing but less extensive. We can say with reasonable confidence that interleaving helps; we can't always say exactly how much in every possible domain.

[Evidence: Moderate-Strong] for math and quantitative domains; [Evidence: Moderate] for other academic domains; [Evidence: Preliminary to Moderate] for motor skills and physical training.

Why Interleaving Works: The Discrimination Hypothesis

Here's the mechanism that makes sense of interleaving, and it's elegant.

When you do blocked practice, here's what happens: you set up to work on Topic A. You read the instructions, load the relevant approach into working memory, and then execute that approach on problem after problem. You're not choosing your approach — it's already chosen by the blocking. You're practicing execution, but you're not practicing selection.

Now think about what a real test looks like, or what real professional application looks like. Problems don't come labeled "this is a Topic A problem." You see a problem and you have to figure out what kind of problem it is before you can solve it. You have to discriminate.

When you do interleaved practice, you're forced to do this discrimination work on every single problem. You see a math problem. You have to ask: what type is this? Which approach applies here? Only then can you solve it.

The extra work of identifying the problem type is exactly the skill you need for real performance — and exactly what blocked practice omits.

Blocked practice gives you the feeling of competence because you're getting problems right. But the reason you're getting them right is that the relevant approach is primed and ready — you've been doing it for twenty minutes already. Remove that priming (which is what a test or real-world application does) and the competence evaporates.

This is related to the transfer-appropriate processing principle in memory research: you learn best when the conditions of learning match the conditions of use. Real tests and real work require you to identify what you're dealing with before responding. Interleaved practice trains this identification.

Think about what Keiko, the competitive swimmer, encounters in a race. She doesn't have the luxury of thinking "this is a turn, so I'll use the turn technique." She has to recognize the wall approaching, select the right action, and execute it. Variable practice — practicing different turns, different breathing patterns, different race scenarios in a mixed order — trains that full sequence: perceive, identify, respond. Block-drilling individual techniques trains only the final step.

Interleaving in Different Domains

Mathematics and Quantitative Subjects

Math is the most extensively studied domain for interleaving, and the implications are clear and practical.

The typical math textbook is organized for blocked practice: Chapter 7 covers derivatives, and the Chapter 7 problem set has forty derivative problems. Chapter 8 covers integrals, and the problem set has forty integral problems.

This structure is convenient for teachers and publishers. It is not optimal for learning.

If you want to interleave your math practice, you have to do it yourself: - Mix problems from different chapters in a single practice session - Create mixed problem sets covering material from the last two or three chapters together - When doing assigned problem sets, rotate between them rather than completing one fully before starting the next - Use practice exams and mixed review problems, which naturally provide interleaved practice

For statistics students: don't do all the hypothesis testing problems, then all the confidence interval problems, then all the regression problems. Mix them.

For physics students: don't do all the kinematics problems, then all the dynamics problems. Mix them and make yourself figure out which physical principles apply.

Sports and Physical Training

In motor learning, the equivalent of interleaving is called variable practice — practicing a skill in varying conditions and contexts rather than drilling it in a single fixed condition.

Keiko, the competitive swimmer, had been running blocked practice for months. Stroke technique drills by stroke — all freestyle technique work, then all backstroke work. Fifteen minutes of the same drill before moving to the next one.

Her coach attended a workshop on variable practice and returned with a proposal: mix the drills. Not three sets of the same drill, but one set each of three different drills, then back to the first, then to a different variant, then to a race simulation, then back to a technique drill. Variable throughout the session.

The first two weeks were harder. Keiko's practice performance dropped. The smooth competence of the blocked drills wasn't there — she felt less precise. But her body was doing more work than it had during blocked drills: not just executing a movement, but recognizing which movement to execute, then executing it. This is real racing.

By the fourth week, something shifted. Her turns were more consistently accurate under pressure. Her ability to hold technique while fatigued — a real-race demand that blocked drills didn't create — improved.

The mechanism: variable practice in sport trains the full perception-selection-execution loop. Blocked practice trains only execution.

[Evidence: Moderate] for variable practice in motor learning, though some of the best evidence comes from earlier motor learning research that hasn't always been perfectly replicated.

Language Learning

Language interleaving is somewhat less studied than math interleaving, but the structural logic is the same.

Typical language learning is heavily blocked: this week's lesson is past tense, next week's is future tense, the week after is conditional. You practice each grammar form until you feel comfortable with it, then move on.

The problem: real language use doesn't come blocked. When you're having a conversation, you're selecting tenses, moods, and structures based on what you're trying to say — you're discriminating constantly. A blocked learning approach leaves you fluent in past tense when it's the obvious thing to use, and lost when you need to choose between past and conditional.

Interleaving grammar topics — deliberately mixing exercises on different tenses and structures, rather than mastering each before introducing the next — trains the discrimination ability. It's harder. It feels less smooth. It builds better communicative competence.

For vocabulary, interleaving means mixing recently learned words with older words in practice sessions, rather than only practicing this week's vocabulary this week.

Studying Multiple Subjects

One of the more accessible applications of interleaving for students is how you structure a multi-subject study session.

Most students study blocked: two hours of chemistry, then two hours of history, then two hours of English. This is intuitively organized — you get into "chemistry mode" and stay there.

Research suggests that a mixed schedule — chemistry for 45 minutes, then history for 45 minutes, then chemistry for 45 minutes — may produce better retention for all subjects than the blocked approach, because you're preventing the habituation and overconsolidation that can come from too long a block on any one topic.

The evidence for cross-subject interleaving is more preliminary than the evidence for same-domain interleaving (mixing problem types within mathematics, for instance), so apply it as a hypothesis to test rather than a certainty. But many students who try it report that the variety feels energizing and that they retain material across subjects better.

[Evidence: Preliminary to Moderate] for cross-subject interleaving.

Professional and Technical Learning

David, learning machine learning, ran into this issue when working through a curriculum that was strictly structured: all the linear regression material, then all the logistic regression material, then all the neural network material.

Each section made sense in isolation. But when he tried to work on a real dataset, he found himself unable to judge which technique to apply. He knew the techniques but not the discrimination criteria.

His solution: after completing each section, he deliberately worked on mixed-problem exercises — dataset challenges that didn't tell him which technique to use, requiring him to make that judgment himself. This is interleaving at the application level, and it built the skills that blocked curriculum study had left out.

When Interleaving Doesn't Help: Important Nuances

Interleaving is not always the right approach, and knowing when to use it is as important as knowing how.

Very Early Skill Acquisition

When you're a complete beginner, you may not have enough basic competence to benefit from interleaving.

Consider someone learning to read music for the first time. They don't know what a quarter note is, what a key signature means, or how to translate notation to sound. Interleaving notation concepts with rhythm concepts with sight-reading before they understand any of them wouldn't produce the discrimination benefits of interleaving — it would just produce confusion.

At the earliest stages, some degree of blocked learning — building basic familiarity with each concept before mixing them — is appropriate. The question is when to make the switch.

The rough guideline: once you have basic functional understanding of each concept or technique (not mastery — just understanding), interleaving begins to produce better results than continued blocking.

The transition point is often earlier than you think. Many students delay interleaving because they don't feel ready — but the research suggests that starting interleaving before you feel fully comfortable with each element is usually the right move.

The Blocking → Interleaving Progression

A practical approach for most learners:

1. Introduce each concept through blocked practice. When encountering a new type of problem or skill, block it initially to build basic competence. Work through several examples of the same type so you understand the approach.

2. Once you can execute the approach reasonably well, switch to interleaving. Mix this concept with others you've already encountered. Now you're practicing selection as well as execution.

3. For review, always interleave. Blocked review — reviewing Chapter 5 problems during Chapter 5 review week — is minimal benefit. Mixed review across all covered chapters produces much better exam performance.

Dissimilar Domains

Interleaving works by building discrimination ability — the ability to tell Topic A from Topic B and apply the correct approach. If the topics you're interleaving are so different from each other that there's no discrimination required (they obviously require completely different approaches), the interleaving benefit is smaller.

Mixing calculus with macroeconomic theory in the same study session may produce some benefits, but the mechanism of interleaving — forcing discrimination between similar-looking problems that require different approaches — isn't fully activated when the topics are from entirely different domains.

Same-domain interleaving (different types of calculus problems; different types of economic reasoning) is where the effect is most reliably strong.

Practical Implementation

Here's the challenge: most textbooks, courses, and curricula are organized for blocked learning. All of Chapter 5 is about derivatives. All of Chapter 6 is about integrals. The professor's slides group topics. The problem set groups problems by type.

To benefit from interleaving, you have to impose it on yourself, often against the grain of how your course is organized.

For Students

Create your own mixed problem sets. After completing each chapter's assigned problems, build a "mixed review" problem set pulling problems from the last three or four chapters. Do it before the exam, not just during exam week.

Study multiple topics per session. Don't allocate all your study time to one subject per day. Mix two or three subjects within each session.

Use practice exams throughout the semester, not just before exams. Practice exams are naturally interleaved — questions appear in random order across topics. Using them early trains both the content and the discrimination skill.

Rotate flashcard decks. If you use Anki or physical flashcards, mix cards from different topics within a subject into a single review session.

For Skills Practice

Design variable practice sessions. Mix drills, exercises, and application tasks rather than blocking by skill or technique. Vary the conditions, the targets, the scenarios.

Practice in conditions that resemble real performance. Real performances are interleaved. Real conversations switch topics. Real races require adapting. Practice in conditions that require adaptation.

Resist the blocked drill temptation. Blocked drill feels productive and satisfying. Variable practice feels rougher and less polished. The research consistently says the rougher, less satisfying practice produces better learning.

Interleaving and the Feeling of Productivity

Let's address the psychology head-on, because it matters.

Interleaved practice reliably feels less effective than blocked practice. When you switch topics frequently, you lose the flow state of extended focus. Your performance during practice is visibly worse — you get problems wrong that you were getting right an hour earlier in the blocked section. You feel less confident.

This is perhaps the most important metacognitive challenge in all of learning science: your judgment of how well you're learning during a session is not reliable. Smooth, fluent practice feels like good learning. It is often a signal of over-learning in a narrow context that won't transfer.

Effortful, struggle-filled practice feels like poor learning. It is often a signal of deeper encoding and better transfer.

This doesn't mean discomfort is always productive. It means you should be skeptical of the feeling of flowing competence during blocked practice, and more trusting of the outcome data (which consistently favors interleaving).

The research on student preferences is humbling here: students consistently prefer blocked practice, rate it as more effective, and predict it will produce better exam scores. The evidence consistently contradicts these preferences.

This isn't a reason to distrust your own experience entirely — your experience contains real information. But it's a reason to run your own experiment (like Exercise 3 in this chapter), and to trust the evidence when your intuition conflicts with it.

The Science Beneath the Effect: What's Actually Happening in Your Brain

The discrimination hypothesis explains the behavioral pattern, but let's get a bit deeper into why the forced identification of problem type produces better learning.

When you work a blocked problem set — twenty quadratic equations in a row — something neurologically interesting happens. Your brain loads the "quadratic framework" into working memory early in the session and keeps it loaded. You're in quadratic mode. The cognitive signature of each problem is "same as the last one; execute the same procedure." No comparison, no categorization, no loading of alternative frameworks.

When you work an interleaved problem set, every problem requires something different. First, your brain must categorize: what type is this? This requires pulling up multiple frameworks — linear, quadratic, systems — and comparing them to the features of the problem. This comparison process exercises each framework more thoroughly than simple repeated execution does.

Think about how you recognize faces. You don't learn what your friend Alex looks like by staring at photos of Alex for an hour. You learn what Alex looks like by seeing Alex next to many different people — friends, strangers, colleagues — and noticing what's distinctive about Alex compared to everyone else. The contrast and comparison builds the recognizable schema.

Interleaving mathematics works the same way. You recognize a quadratic problem by comparing it to linear problems (single-variable, first degree), systems problems (multiple equations), and quadratic-lookalike problems that are actually something else. The comparison is how the schema gets built precisely enough to generalize.

This is also why interleaving in artist recognition works: seeing Monet next to Renoir next to Cézanne builds richer, more specific schemas for each artist's style than seeing all Monet, then all Renoir, then all Cézanne. The similarities and differences between artists teach you what to look for in a way that studying each in isolation doesn't.

Interleaving in Professional Development: The Difficult Case

Professional learning is often organized in exactly the wrong way for interleaving. You attend a three-day workshop on a specific topic. You read a book on a specific technique. You do a deep dive into one tool. Everything is blocked by design.

David's machine learning education ran into this. Each module of his online course was a deep dive into one algorithm family: first linear models, then tree-based models, then neural networks. Appropriate for introducing each approach, but terrible for building the discrimination skill that says: "I have this specific dataset and this specific problem — which family of approaches is right for it?"

The fix he implemented: after completing each module, rather than moving directly to the next, he spent several sessions working on problems that didn't tell him which approach to use. Mixed problems. He had to read the problem, consider the data structure, and decide.

This is the professional equivalent of building your own interleaved practice sets — it requires initiative, because professional learning programs rarely build this in for you. But the payoff is the same as for students: better discrimination ability, better transfer, better performance when the situation doesn't come labeled.

A second approach David used: comparative study. Rather than studying neural networks in isolation, he studied "when do neural networks outperform gradient boosted trees, and why?" This comparative framing forced him to hold both frameworks in mind simultaneously, comparing and contrasting — which is what interleaved practice simulates across many problems.

The Contextual Interference Effect: Why It Has This Strange Name

The research literature calls the interleaving advantage the "contextual interference effect," and the name deserves explanation because it's counterintuitive.

"Interference" here means that the mixing of multiple tasks or topics interferes with each other during practice — you can't fully establish a smooth flow with any one topic because the others keep interrupting. This interference is what produces the worse performance during practice.

"Contextual" means the interference comes from the varying contexts — the changing task demands, the shifting frames of reference, the different cognitive schemas required.

Why would interference help? Because the interference is what forces your brain to do the harder cognitive work — identifying, comparing, selecting, and applying frameworks — rather than just executing a pre-loaded one. The interference is not a bug; it's the feature.

The original contextual interference research was done in motor learning by Battig (1979) and replicated by Shea & Morgan (1979). They found that when people practiced three different movement sequences in a random order (rather than completing all of sequence A before starting B), their practice performance was worse but their retention and transfer were dramatically better. This finding has been replicated in dozens of subsequent motor learning studies.

The extension to cognitive learning — the finding that interleaved math practice produces the same pattern — was an important discovery because it suggested the mechanism wasn't specific to motor learning but reflected something fundamental about how memory and schema acquisition work.

Why Textbooks Are Organized for Blocked Learning (and What You Can Do About It)

Here's a structural problem worth naming explicitly: virtually all textbooks are organized for blocked learning.

Chapter 1: Linear functions. Chapter 2: Quadratic functions. Chapter 3: Polynomial functions. Chapter 4: Exponential functions.

This organization makes sense for introducing concepts — you need to understand each type before you can compare them. But once you've been introduced to all four types, the textbook's organization becomes an obstacle to the interleaved practice that would best consolidate your ability to distinguish and apply them.

The same pattern appears in foreign language textbooks (one chapter per grammar topic), chemistry textbooks (one chapter per reaction type), law courses (one module per area of doctrine), and virtually every other structured curriculum.

Teachers organize instruction this way because it's administratively clean, conceptually sequential, and pedagogically easier — students find it easier to perform well during lessons when each lesson is focused. This creates a feedback loop: students feel good during blocked lessons, teachers see good in-class performance, everyone concludes that blocked organization is working.

It's working for in-class performance. It's not working for transfer and long-term retention.

Your job as a learner is to recognize this structural limitation and compensate for it:

During a chapter: Pay attention to where the current topic is similar to and different from previous chapters' topics. Make these comparisons explicit in your notes.

After a chapter: Create a brief comparison table: how is this topic similar to what I've learned before? How is it different? When would I use this approach vs. the previous ones?

When building practice problem sets: Deliberately pull problems from multiple chapters. If your textbook has a cumulative review section, prioritize it over the chapter-specific sections.

When studying with others: Give each other problems from mixed chapters without labeling the chapter. This forces the discrimination practice that blocked homework sets avoid.

Interleaving and the Tests That Matter

The irony of interleaving's benefits is most acute on standardized tests.

The SAT math section doesn't organize questions by topic. The MCAT doesn't label which biochemistry topic each question addresses. The bar exam doesn't organize questions by course. The GRE quantitative section doesn't tell you whether the problem requires algebra, geometry, or statistics.

All of these high-stakes exams present problems in exactly the interleaved format that interleaved practice trains. And most students prepare for these exams with blocked practice — drilling practice tests section by section, topic by topic.

The students who do best on these exams are often those who have, intentionally or through experience, developed strong problem-identification skills. They see a problem, rapidly categorize it, and apply the appropriate approach. This skill — which interleaved practice specifically trains — is often described as "test-taking ability" or "test sophistication," but it's really domain-specific discrimination ability that results from varied, interleaved practice.

If you're preparing for a standardized exam, this is one of the most practical implications of the interleaving research: design your practice to be interleaved across topics, not blocked by topic. Use mixed practice sets from the beginning, not just in the final weeks.

A Note on Interleaving Difficulty

The interleaving effect is real and powerful, but it's worth being honest about how difficult it can be to implement consistently.

Interleaved practice is genuinely harder — not just subjectively, but in the sense that you're doing more cognitive work per unit of time. You can't get into a flow state on any one topic. You have to keep context-switching. This is mentally taxing.

The practical implication: interleaved sessions may be shorter in absolute time than blocked sessions while accomplishing similar learning. If you normally study one subject for two hours in a blocked session, an interleaved session covering three subjects might feel about as cognitively demanding as the two-hour blocked session, even if it's only ninety minutes.

This is fine — you're not trying to maximize time spent; you're trying to maximize learning per time spent. But it means you should expect interleaved sessions to feel more tiring, and you should plan accordingly.

One strategy: use blocking for the first exposure to new material (you need to get oriented before you can mix), and use interleaving for review and consolidation. This preserves the initial scaffolding benefit of blocking while capturing the retention and transfer benefits of interleaving.

Another strategy: interleave within a broader block. Study chemistry for an hour, but within that hour, alternate between different types of chemistry problems. This is more interleaved than completely blocked (all Chapter 5 problems, then all Chapter 6 problems) without requiring you to jump between subjects.

The Transfer Problem: What Interleaving Is Really Solving

Let's zoom out for a moment and think about what the interleaving research is really telling us about how learning works.

The central problem interleaving addresses is called the transfer problem: the gap between what students can do in the context where they learned something and what they can do when they encounter that knowledge in a new, different context.

Transfer failure is one of the most consistent findings in educational psychology. Students who mastered a physics problem-type in Chapter 5 often fail to apply the same principle when it appears in Chapter 8 embedded in a different context. Medical students who learned to diagnose a condition in a textbook often struggle to recognize it in a real patient with a different presenting story. Law students who memorized a doctrinal rule may fail to apply it when the facts pattern is novel.

The problem isn't lack of knowledge — it's knowledge that's too context-bound. The student knows "Chapter 5 physics problems involve conservation of momentum." They don't know "problems where two objects interact and we want to know final velocities involve conservation of momentum." The knowledge is stored with the context of learning, not with the principle that should guide its application.

Interleaved practice is specifically targeted at this context-dependency. By forcing you to encounter the same type of problem in varied contexts — without the contextual cue "you're currently doing Chapter 5 problems" — it requires you to attach your knowledge to the principle rather than to the context. The principle becomes retrievable directly, not only through the original learning context.

This is why interleaved practice on standardized tests matters so much. These tests specifically strip away contextual cues. There's no chapter number, no topic header, no "we've been studying momentum for the last hour." There's just the problem. Students who've only studied in blocked contexts are disadvantaged not because they don't know the material but because they don't know how to access it without contextual priming.

Interleaving in Creative and Qualitative Domains

Most of the research on interleaving has been conducted in mathematics and perceptual learning, where "problem type" is a fairly clear concept. But the principle extends, in somewhat different form, to creative and qualitative domains.

Writing and rhetoric: Different types of writing tasks — argumentative essays, narrative description, analytical summaries, technical explanations — each have different conventions, structures, and rhetorical moves. Practicing only one type at a time (argumentative essays all semester, then narratives all semester) is blocked practice. Interleaving them — writing different genres across sessions — builds the discrimination ability to recognize what each task requires and shift approach accordingly.

Visual art and design: Design students who study different artistic styles or design approaches in a mixed order show better ability to identify and apply each style than students who study them sequentially in blocks — which is essentially the Kornell & Bjork artist recognition finding, applied to production rather than identification.

Reading complex texts: If you're building expertise in a subject through reading, interleaving texts with different perspectives, methodologies, or arguments within the same subject forces you to track the differences between approaches. Reading all of Marx, then all of Weber, then all of Durkheim in sequential blocks gives you three separate mental models. Reading them interleaved — a chapter of Marx, then a chapter of Weber, then back to Marx — forces comparison and contrast that builds a richer, more integrated understanding of each.

Studying multiple languages simultaneously: Controversial but supported by some evidence: studying two related languages simultaneously (French and Spanish) in an interleaved fashion, rather than fully mastering French before beginning Spanish, may build stronger discrimination between the languages. Students learning both together are constantly comparing, which clarifies what's distinctive about each.

Measuring Whether Interleaving Is Working for You

Unlike spaced repetition, where Anki gives you statistics, and unlike retrieval practice, where blank-page recall percentages are a direct measure, interleaving is harder to quantify in the moment. How do you know if your interleaved practice is having the intended effect?

Delayed test performance. The definitive measure is performance on a delayed test — a test taken a week or more after your practice, with problems in a random, interleaved order. If your interleaved practice is working, you should see better performance on such tests compared to what you'd expect from blocked practice.

The identification speed test. After a period of interleaved practice, try this: look at a set of mixed problems from all the topics you've been studying. Before solving each one, write down what type of problem it is. Can you identify the type quickly and accurately? If yes, your discrimination skill is developing. If you frequently misidentify problem types, you have more interleaved practice to do on distinguishing those types.

The "novel context" test. Take a problem you know well and embed it in an unfamiliar narrative context. If your knowledge transfers cleanly to the novel context, interleaving is working. If you need the familiar context to access what you know, it's a sign that your knowledge is still too context-bound.

Classroom performance over time. The clearest evidence, for students in courses: do your unit exam scores drop when the exam is cumulative (covering all previous material in mixed order) compared to chapter tests? A large drop suggests your knowledge is context-dependent — good for blocked tests, weaker for mixed tests. Interleaved practice should narrow this gap.

Building an Interleaved Study Plan From Scratch

Let's get concrete about what an interleaved study plan actually looks like for a student taking four courses simultaneously.

Suppose you're taking organic chemistry, English literature, calculus, and macroeconomics. The naive approach:

• Monday: 2 hours chemistry
• Tuesday: 2 hours calculus
• Wednesday: 2 hours literature
• Thursday: 2 hours economics
• Friday: Review whichever feels weakest

This is blocked at the daily level. A step toward interleaving:

• Each day: 45 min chemistry, 45 min calculus, 30 min literature, 30 min economics
• Within each subject block, rotate between topic types (for chemistry: 20 min mechanisms, 20 min stereochemistry, 20 min spectroscopy)

This is better. A full commitment to interleaving:

• Each day: Alternate between subjects every 25-30 minutes
• Within each subject, mix topics from different chapters and problem types
• Reserve one 90-minute block per week per subject for forward progress on new material; the rest of the time is interleaved review

The full interleaved approach requires more active management — you have to track what you've covered, remember where you were across multiple topics, and resist the temptation to finish a blocked section before switching. But the cognitive overhead is also part of the benefit: the management itself is a form of interleaving, and the switching costs build scheduling and self-regulation skills.

Start where you are. Even blocking at the topic level (within a subject, alternate chapter types) is a significant improvement over straight sequential blocking. Work toward more interleaving over time as the habit builds.

What Teachers and Coaches Misunderstand About Performance During Practice

One of the reasons interleaving hasn't been more widely adopted in schools and coaching programs is a systemic misalignment between what teachers observe and what actually produces learning.

Teachers naturally evaluate students based on performance they can see: how well are students doing during practice? How accurately are they answering questions in class? How cleanly are they executing drills?

Blocked practice produces excellent observable in-class performance. Students get problems right. They feel confident. The teacher observes competence.

Interleaved practice produces worse observable in-class performance. Students get problems wrong more often. They look uncertain. They struggle to identify problem types. The teacher may feel that the students aren't learning — or worse, that the teaching approach is failing.

This creates a powerful feedback loop pushing toward blocked practice: teachers who adopt interleaving may face student complaints, lower in-class performance metrics, and the uncomfortable feeling that their class isn't going well — all while the research says their students are building more durable learning.

The only way out of this misalignment is a different relationship with the performance-learning distinction. Teachers who understand desirable difficulties can frame interleaving for students explicitly: "I'm going to mix up these problems on purpose. It's going to feel harder. You're going to get some wrong that you'd get right if I gave them to you in blocks. That difficulty is the point. The research says it helps you learn better, even though it doesn't feel that way."

That framing — honest, explicit, trusting students with the mechanism — helps students tolerate the discomfort rather than interpreting it as failure.

For self-directed learners (which you probably are, if you're reading this book), the equivalent is understanding the mechanism well enough to trust it when your experience says it isn't working. Your experience says: I'm getting problems wrong, I feel less confident, this session felt rough. The research says: you're building discrimination ability, your transfer will be better, your delayed test will reflect this. Trust the research.

The Interleaving Effect and Long-Term Expertise

The interleaving research is mostly measured over weeks and months. But there's a longer-range dimension worth considering.

Expertise in any complex domain is characterized by rich, flexible schemas — mental models that can recognize problem types quickly, select appropriate approaches efficiently, and adapt when standard approaches don't fit. Novices have narrow, inflexible schemas: they can solve problems in the exact form they practiced, but struggle to generalize.

The development of expert-level schemas requires repeated experience with varied problems that require the same underlying principles — which is what interleaved practice provides. Each interleaved problem session is one increment of schema refinement. Over hundreds of sessions across months and years, this refinement builds the kind of automatic, flexible pattern recognition that characterizes genuine expertise.

This is consistent with Ericsson's deliberate practice research (Chapter 4): the distinguishing feature of expert practice is not quantity but quality — specifically, the degree to which practice pushes into the zone of difficulty and demands active engagement with what isn't yet automatic. Blocked practice, by definition, automates things that then get practiced once automated. Interleaved practice keeps pushing into the discrimination and application work that builds expertise.

Keiko's experience with variable swim training is a long-term version of this: the initial performance dip, the gradual improvement, the eventual robust, adaptive performance under varied race conditions. The performance dip was the expertise being built. The adaptive performance was the expertise expressing itself.

Common Questions About Implementing Interleaving

"Should I start a new topic using interleaving from day one?"

No. When you first encounter a completely new topic, you need some blocked exposure to build initial orientation. What is this? How does it work? What does a problem of this type look like? These foundational questions need to be answered before you can mix this topic with others and practice discrimination.

The practical guideline: spend two to three sessions (or enough time to complete a first pass through the key concepts) in a roughly blocked fashion, then introduce interleaving with previously studied topics. The transition point is when you can at least attempt problems of the new type, even if your accuracy is low.

"How do I interleave if I only have one subject to study?"

Interleave within the subject. Most subjects have multiple subtopics, problem types, or skill components. In calculus: differentiation, integration, limits, optimization. In French: present tense, passé composé, imparfait, future. In anatomy: regions of the body, layers of tissue, blood supply, innervation. Alternate between these within your study session.

"What if the topics I'm interleaving are too similar?"

"Too similar" is rarely a problem — if topics seem so similar that you're not sure how to discriminate between them, that's precisely where interleaving will help most. The confusion between very similar topics is resolved by explicit comparison practice, which is what interleaving forces.

"What if the topics I'm interleaving are too different?"

If the topics are completely unrelated — organic chemistry and macroeconomics, say — interleaving them won't provide the discrimination training that makes interleaving effective for same-domain topics. You'll still get some benefit (context-switching keeps you cognitively engaged, prevents over-consolidation in one track), but the main mechanism of interleaving — building discrimination within a domain — won't be as active.

For very different subjects, you're mostly relying on the benefit of distributing time across multiple topics (which has its own benefits) rather than the discrimination-building benefit of interleaving within a domain.

"Can I interleave excessively?"

Yes. If you switch too frequently — every five minutes, for instance — you may not have enough time within each switch to actually process and attempt the material. You want to switch before you've finished, but not so early that you haven't engaged at all. A reasonable unit is 15-25 minutes per topic before switching, or 3-5 problems of one type before switching to another type.

The Honest Verdict: How Strong Is the Evidence?

Let's be precise about the evidence, because the research on interleaving is more nuanced than the research on retrieval practice and spaced repetition.

Very strong evidence for: Interleaving different problem types within a single domain (e.g., different types of math problems, different artists in a recognition task). Effect sizes are large, findings have replicated, and classroom studies confirm the lab findings.

Moderate evidence for: Interleaving different subjects within a study session. The benefits exist but are somewhat more context-dependent and the research base is smaller.

Preliminary evidence for: Interleaving in motor skill learning (variable practice). The contextual interference effect in motor learning is well-established, but the translation to specific training protocols is less precisely studied.

Less studied: Interleaving in creative domains, writing, and qualitative subjects. The theory predicts benefits, and the limited available evidence is consistent, but the research base is thinner.

[Evidence: Moderate-Strong] is the overall rating, reflecting strong evidence in the core domains (math, perceptual learning) and moderate evidence in broader applications.

The practical takeaway: the evidence is solid enough to act on, especially in quantitative and technical subjects. In qualitative and creative domains, treat it as a hypothesis to test in your own practice rather than a certainty. You have little to lose by trying interleaving and a lot to gain if it works as well as the research suggests.

All three techniques — retrieval practice, spaced repetition, and interleaving — are in the same family: desirable difficulties. They all work by making practice harder in specific, productive ways.

Interleaving + retrieval practice: When you practice retrieval from mixed topics, you're doing both simultaneously. The identification work (which concept does this question require?) is interleaving; the production work (retrieving the answer from memory) is retrieval practice. Mixed flashcard decks, mixed practice test questions, and mixed problem sets combine both effects.

Interleaving + spaced repetition: Anki naturally interleaves topics because your due cards come from many different decks and topics simultaneously. Your anatomy card might be followed by a pharmacology card, then a biochemistry card, then another anatomy card. This is interleaving by another name — you're constantly identifying which domain you're in before retrieving the answer.

All three together: A well-designed review session might look like: twenty Anki cards from three different subjects (interleaved retrieval practice), with the cards scheduled to appear at optimal forgetting-curve intervals (spaced repetition). Each individual card requires effortful recall (retrieval practice). The switching between subjects forces discrimination (interleaving). This is close to optimal learning in a small, daily dose.

Try This Right Now

Think about a subject or skill you're currently learning that has multiple distinct components, types, or techniques.

Design an interleaved study session for it: identify three to five topic areas or skill types, and plan a session where you rotate through them, spending no more than fifteen to twenty minutes on any one before switching.

For a math course: pick five different problem types and create a ten-problem mixed set. For a language: pick three grammar topics you've covered and mix practice exercises from all three. For a sport or physical skill: write a variable practice sequence for your next training session, mixing at least three different drills or exercises.

Actually do this session within the next 48 hours. Notice how it feels compared to your normal blocked practice. Notice whether it feels more or less productive. Then hold judgment and wait to see how your retention looks when you test yourself a week later.

The Progressive Project

What is the natural blocking structure in your chosen learning goal, and how can you introduce interleaving?

Most structured learning paths are organized for blocked delivery: topics in sequence, skills in order, concepts grouped by type. Map out the blocked structure of what you're learning. Identify at least three topic areas or skill types that you could begin mixing.

Then answer: - What would a mixed practice session look like for your domain? - What's the earliest point at which you have enough basic competence in multiple topics to start interleaving? - What resources do you have for mixed-problem practice (practice exams, mixed exercise sets, varied training conditions)?

Write a one-paragraph plan for how you'll implement interleaving in your specific learning goal, starting this week.

What Comes Next

Chapters 7, 8, and 9 have all involved making your practice harder in specific ways: effortful retrieval, effortful spacing, effortful discrimination. All are desirable difficulties — the struggle is the mechanism.

Chapter 10 takes a different angle. Elaboration is about going deeper into what you're learning — connecting new material to what you already know, asking why, building rich networks of understanding rather than isolated facts.

You can retrieve and space facts without understanding them deeply. Elaboration is what turns retrieval practice from memory maintenance into genuine comprehension.

Chapter 10: Elaboration and Elaborative Interrogation: Connecting New Knowledge to What You Already Know →