10 Rules for Designing Effective Learning

Or How to Plan Teaching That Works for All Students, Not Just the Strongest

For a few years now, I have been reading and re-reading Theory of Instruction by Siegfried Engelmann and Douglas Carnine, which stands perhaps as education's closest approximation to a Principia Mathematica. The basic argument is that all learning follows predictable, logical patterns when instruction is properly designed and that violating these logical principles doesn't merely make teaching less effective, it makes concept formation impossible, which systematically abandons the students who most need our help whilst allowing only the strongest learners to succeed despite flawed instruction.

“Their theory is based on two assumptions: learners perceive qualities, and they generalize upon the foundation of the sameness of qualities.”¹

The book is formidable: dense, technical, and ruthlessly systematic. Yet it represents a serious attempt to decode the fundamental mechanics of reliable learning, rather than leaving success to chance or sentiment. But these aren't merely pedagogical preferences, they follow from how human concept formation actually works. The same logical processes that philosopher John Stuart Mill identified for scientific induction in 1844.

When learners encounter examples, their minds must induce general principles from specific instances. Mill showed that this inductive reasoning follows strict logical constraints: to isolate what causes what, you need systematic control of sameness and difference across your examples. The authors realised that learning is identical to this process; students are constantly making inductive inferences from the examples we show them.

The tragedy, as the book demonstrates, is that capable students often overcome our instructional failures through their own cognitive resources. They can filter irrelevant information, self-correct errors, and bridge gaps in logic. This creates the illusion that our teaching works, when in reality it only works for those who least need it. Meanwhile, students who struggle are left without the precise, systematic guidance they require to succeed.

So here are 10 things I learned from this book in the form of rules for designing effective learning.

---

1. Boundaries matter more than definitions.

Students don't just learn what something is, they learn what it is, versus what it isn’t. Without clear boundaries, concepts become fuzzy and useless. A child who's only seen red roses will call pink flowers "red." A student who's only seen mammals on land won't recognise whales as mammals. Show the boundaries explicitly, or students will tend to overfit everything.

Students can memorise that "democracy means rule by the people" and still have no idea how to identify one in practice. The definition provides no guidance for distinguishing democracies from other systems that might superficially seem to involve popular participation. But show them democracies versus dictatorships, democracies versus anarchies, democracies versus oligarchies, and the concept crystallises with remarkable clarity.

This principle extends beyond initial instruction into assessment. If you test students only on the same examples you taught, you're not measuring learning; you're measuring recognition, not understanding. A student who can identify the three triangles you showed in class but fails on a new one hasn't learned "triangle"; they've learned "those three shapes." Boundaries come alive when students can apply them confidently to novel examples they've never encountered.

Example: Teaching "triangle"? Don't merely show triangles. Show squares, circles, and other shapes labelled "not triangle." Then assess with fresh shapes they've never encountered. The boundary between triangle and not-triangle is where genuine understanding resides, not in the memorisation of particular instances.

Learning happens when students must decide what belongs and what doesn't, not when they only just repeat what belongs.

2. Not all concepts are alike, the structure determines how you teach them.

Different types of concepts demand completely different instructional approaches. You cannot teach everything identically and expect it to work. Some concepts require positive and negative examples to establish clear boundaries. Others need step-by-step transformations to show process. Still others require relational comparisons to highlight critical features. Match your method to your concept type, or you'll create confused learners who memorise surface features without grasping underlying structure.

Example: Teaching "mammal" needs boundary examples (whale versus fish, bat versus bird) to establish the essential features that define the category. Teaching long division needs step-by-step procedures that break down the algorithm into manageable components. Teaching "irony" needs contrasting examples that highlight the gap between intended and apparent meaning. Use the wrong approach and students will fail predictably, not through lack of ability but through instructional mismatch.

3. Complex skills need task analysis. Break them down first.

Before teaching any complex skill, ruthlessly analyse what students must already know. Most instructional failures occur because teachers skip this step and assume students possess prerequisite knowledge they don't actually have. This is not about lowering expectations; it's about building solid foundations. Don't guess what students know, test it systematically. Find the gaps, fill them methodically, then attempt the main skill.

The temptation is to dive straight into the complex skill, assuming that students will somehow pick up the prerequisites along the way. This approach virtually guarantees that struggling students will be left behind, whilst stronger students who already possess the prerequisites will appear to validate the approach. The result is a misleading sense that the instruction works, when in fact it only works for those who least need it.

Example: Before teaching essay writing, test whether students can write clear sentences, identify main ideas, and organise thoughts into coherent paragraphs. If they cannot, teach those component skills first rather than attempting to teach essay structure to students who lack the building blocks. The essay becomes possible once the foundations are secure, but not before.

4. Don't reinvent examples. Recycle them to teach similar concepts

When you need to teach related concepts, don't start from scratch. Use the exact same example sequence you already designed, but change how you question students about those examples. Many concepts are linked by convention rather than logic: synonyms, related terms, multiple labels for the same phenomenon. This recycling approach prevents confusion and accelerates learning by building on established foundations.

The efficiency gains here are remarkable. Rather than designing entirely new example sets for each related concept, you can leverage the cognitive work students have already done. They've already learned to attend to the relevant features; now you're simply teaching them different ways to label or think about those same features.

Example: Teaching “photosynthesis”? You’ve already used a diagram of a plant to show how it produces food using sunlight, water, and carbon dioxide. When moving on to “cellular respiration,” don’t invent a brand-new diagram. Reuse the same plant diagram, but this time highlight the flow of oxygen and glucose instead of sunlight and carbon dioxide. The recycled example helps students see the processes as complementary, not isolated.

5. Every example needs a signal

Examples without labels are merely noise. You must explicitly tell students what to pay attention to in each example. Don't assume they'll notice the right feature; direct their attention deliberately. This isn't about spoon-feeding; it's about ensuring that the cognitive work students do is focused on the right elements.

The assumption that students will naturally attend to the relevant features is one of the most persistent errors in instruction. Students are constantly bombarded with sensory information, and without explicit guidance, they have no way of knowing which features matter and which are incidental. The signal acts as a spotlight, illuminating what deserves attention.

Example:

Engelmann, S., & Carnine, D. (2016). Theory of instruction: Principles and applications. National Institute for Direct Instruction.

6. Variety is your friend, but don’t randomise.

Show students the full range of a concept so they don't learn narrow prototypes that fail to generalise. But that variety must be systematically planned, not randomly shuffled. Random examples create random learning; students will form whatever concept the accidental sequence happens to suggest. Systematic variety, by contrast, reveals the underlying structure by carefully controlling which features vary and which remain constant.

The goal is to show students the boundaries of the concept whilst maintaining logical coherence in the sequence. This requires considerable forethought about which examples to include and in what order. Each example should serve a specific purpose in building or refining the student's understanding.

Example: Teaching "bird"? Show robins, eagles, penguins, ostriches, but in an order that systematically reveals that flight is variable whilst feathers are constant. Begin with typical flying birds, then introduce flightless species to show that wings don't define the category. Random order will teach random concepts, leaving students confused about what actually makes something a bird.

7. If your examples allow two rules, some students will learn the wrong one

Here lies the heart of faultless communication: keep everything constant except the one thing you want students to notice. If your examples vary in multiple ways, students will form competing hypotheses about what matters, and those who struggle will inevitably latch onto the wrong pattern. This isn't a failure of intelligence; it's a predictable consequence of ambiguous instruction.

Every unnecessary feature in your examples represents a potential trap. Show three red circles to teach "red," and some students will learn "circular" instead. This happens because both features are present in all your examples, making both equally plausible as the defining characteristic. The strongest learners can filter out irrelevant information, but those who need our help most cannot manage this cognitive load.

Example: Teaching "bigger"? Use the same two balls in the same position; just change which one is larger. Don't mix in different objects, different locations, or different orientations. Control everything except size. This way, students cannot form incorrect rules about colour, shape, or position because these variables remain constant across examples.

8. Teach the minimum set that makes the maximum rule

More examples aren't better; the right examples are better. Find the smallest set that creates the biggest, most accurate generalisation. This is efficiency at its purest: maximum learning from minimum input. Each example should earn its place by revealing something essential about the concept's structure.

This principle requires disciplined thinking about what each example contributes. If an example doesn't add new information or refine an existing boundary, it's cognitive clutter. Students have limited attention and working memory; every unnecessary example reduces the clarity of the essential pattern.

Example: Teaching "democracy"? You don't need every democratic country. You need examples that systematically show: people vote (versus dictatorships), leaders can be removed (versus autocracies), multiple parties compete (versus one-party states). Three well-chosen contrasts teach more than dozens of similar cases.

9. Relying on corrective feedback to bail you out means your initial design was flawed

Correction is a plaster for broken instruction. If you're constantly fixing student errors after the fact, your examples were poorly designed from the start. Good instruction prevents errors instead of correcting them. This doesn't mean errors never occur, but they shouldn't be the primary mechanism through which students learn what you meant to teach.

When errors are frequent and predictable, they signal that the instructional sequence itself is creating confusion. Rather than treating symptoms through correction, address the cause through better design. This shift in perspective moves responsibility from the student (who must recover from confusion) to the instructor (who must prevent it).

Example: Teaching multiplication versus addition? If you introduce both with word problems like “Sam has 3 bags with 4 apples in each” but don’t contrast it with an addition case (“Sam has 3 apples and then gets 4 more”), many students will default to adding. If half the class keeps answering 3 + 4 instead of 3 × 4, the issue isn’t their inattention, it’s your design. Build the contrast explicitly from the start, so they see why multiplication is groups of equal size and addition is combining totals.

10. Most “personalised” learning is backwards

When students struggle, the solution isn't simplified work; it's stronger foundations. Reducing the challenge of the current task often obscures rather than addresses the real problem. Instead, diagnose missing prerequisites and teach those systematically. True adaptation goes backward to fill gaps, not forward to circumvent them.

This approach requires diagnostic thinking about why students are struggling. Surface-level difficulties often mask deeper gaps that must be addressed before progress becomes possible. The goal is not to make tasks easier but to make students more capable of handling appropriate challenges.

Example: Student failing at algebraic equations? Don't provide easier algebra problems; investigate whether they can solve arithmetic equations first. Missing that foundation? Teach it explicitly, then return to algebra with confidence. The gap was never in algebra itself; it was three conceptual steps earlier. Address the real problem, and the apparent one dissolves.

---

Theory of Instruction reminded me of The Brothers Karamazov in the sense that when you first read it, you don’t understand everything going on but at the same time, you have this vivid sense that something really important is going on. Perhaps the link is that Karamazov argues that there’s a moral law beneath human chaos, and Theory of Instruction argues there’s a logical law beneath the apparent chaos of learning.

Perhaps most importantly, both works suggest that understanding these underlying laws carries profound moral weight. Once we know how learning really works, we become morally obligated to design instruction properly. Once we understand human nature, we're responsible for creating systems that honour rather than violate it.

Faultless communication removes guesswork from the learner's side and places full responsibility on the instructional design. If students fail to learn, it is not because they are inattentive, lazy, or incapable. As Engelmann and Carnine put it: "If kids mislearn, the fault is in the design, not the learner."

This, to me, is a profoundly hopeful message, because it suggests that educational failure is not inevitable but engineered, and what can be engineered can be re-engineered. It liberates us from fatalistic thinking about ability and aptitude, moving us instead toward a world where systematic design can create systematic success.

Their insight is a profoundly equitable one. They demonstrate that what we attribute to individual differences in ability often reflects differences in instructional quality. The child who "just doesn't get maths" may simply have encountered instruction that violated the logical principles of concept formation as opposed to not being clever enough.

This doesn't deny that children bring different strengths and interests to school. But it suggests that the basic capacity to form concepts, to reason, to learn; these are universal human capabilities that can be systematically developed through proper design. The real scandal of education is not that some children cannot learn, but that our instruction too often makes it impossible for them.

1. https://www.nifdi.org/85-research/article-abstracts/411-engelmann-s-carnine-d-1982-theory-of-instruction-principles-and-applications-new-york-irvington-publishing-inc