Engelmann Last

Why the hardest problems in education come before instruction begins

Let me start by saying the thing that nearly everyone in education privately thinks: the Engelmann DI community has a bit of a cult problem. Not because the ideas are wrong - many of them are genuinely powerful - but because Theory of Instruction is so painful to read that the only people who finish it are the kind of people who become evangelists for it. The rest of us form our views secondhand. This distorts the conversation in both directions. The devotees over-claim, because they’ve invested so much effort that the framework feels more complete than it is. The sceptics under-claim, because they’re reacting to the evangelism rather than the actual ideas. I’ve spent a decade chatting with education friends about Engelmann and only now have I finally forced myself to read and digest every single chapter of the great book.

It wasn’t fun, but it was worth doing, and I’ve come to a view that I suspect will unsettle some of his admirers:

Engelmann is best understood as last, not first.

His methods are invaluable in certain circumstances. But they only become applicable after a series of much harder decisions have already been made. Decisions that his framework does not help you make.

Engelmann gives us a rigorous theory of instructional communication: how to present material so that attentive learners necessarily acquire it. But the hardest problems in education occur before that communication begins, and often after it succeeds too.

He solves the middle of the problem brilliantly for certain types of knowledge. But the middle may not be the hardest part.

What Engelmann actually solves

Engelmann’s core contribution is a theory of communication.

Given that you want to teach X, how do you communicate X so clearly that learners cannot reasonably misunderstand it?

His answer is meticulous. Instruction should be engineered so that ambiguity disappears. Concepts are introduced through carefully selected examples and contrasts. Critical features are highlighted through minimal differences between cases. Learners are required to respond frequently so errors appear immediately and can be corrected immediately. Practice continues until responses become fluent and automatic.

The goal is faultless communication: instruction designed so carefully that learner errors arise from inattention rather than from confusion about what was taught.

When applied to well-structured domains, this machinery is extraordinarily powerful. Early reading. Mathematics. Logical relations. Grammatical transformations. In domains where the target knowledge is well-defined and the correct response is unambiguous, Engelmann’s methods are engineered to produce reliable learning with remarkable efficiency. Few people have thought more carefully about eliminating ambiguity from instructional communication.

But notice what this machinery assumes. It assumes that someone already knows what X is. More precisely, it assumes that the knowledge to be taught has already been decomposed into discrete pieces (discriminations, rules, procedures) that can be communicated one at a time. Engelmann tells us how to communicate a defined piece of knowledge without ambiguity. He does not tell us how to decide what that piece of knowledge should be.

The harder problem upstream

Before you can apply Engelmann’s principles, someone must already have done something very difficult. They must have decided what knowledge to teach and how to carve it into teachable pieces. In some domains, this is easy or rarely done by the classroom teacher. In others, it is the central intellectual challenge of curriculum design and instruction.

I’ve written recently about why knowledge doesn’t decompose into natural units because the brain stores knowledge in distributed patterns rather than discrete, labelled pieces. This makes any attempt to carve a domain into chunks a modelling decision, rather than a discovery of truth. In Theory of Instruction, Engelmann and Carnine acknowledge Wittgenstein’s famous observation about the definition of “games”: that there is no single quality shared by all games and absent from all non-games. Their response is essentially: just pretend boundaries of definitions are crisp, even if they aren’t. Pick a working definition, select positive and negative examples that respect it, and teach to that. For younger students, this may be a pragmatic and defensible move in the sense that that simplification costs little and the pedagogical gains are real.

However, for older students, the fuzziness is not a simplification to be papered over – it is the phenomenon we study. Consider teaching the Great Depression to older students. Engelmann’s methods could ensure students reliably distinguish between stocks and bonds, understand what a bank run is, and identify the difference between a trigger and an underlying cause. Clear discriminations. Precise definitions. Faultless communication. But those are not the hardest instructional decisions (or indeed the hardest things to learn). The harder questions come first.

What is an economic depression? There is no single agreed-upon definition that would allow everyone to classify economic downturns into depressions and not-depressions. Engelmann would handle this by teaching a prototype first and expanding. But his framework treats fuzziness as a temporary state to be resolved through better example selection, when in fact the fuzziness is often the point. A student who has learned a crisp Engelmann-style definition of “economic depression” and can reliably classify positive and negative examples has learned something misleadingly tidy. The real intellectual work that constitutes studying economics, history, or philosophy is understanding why the boundaries are contested and how different theoretical perspectives lead to different conclusions about where to draw them.

What mental model of the stock market should students have? Not a technically accurate one since our students are not becoming economists! We need a simplified model that makes the events of the Depression intelligible. Which mechanisms must be included? Which can be left out? Which simplifications will support later learning, and which will create misconceptions that need painful unlearning? Engelmann cannot answer this question. His framework assumes the model already exists.

What narrative structure will make the events memorable? History is not just a list of facts. Students need a causal story they can reconstruct: a sequence of events that makes sense, connects to prior knowledge, and remains stable in memory. The features that make such a narrative stick, such as coherence, causation or vivid particular, is a design problem quite different from definitional clarity. Engelmann does not address narrative architecture.

Why does any of this matter? Students do not just learn facts about the Depression; they learn why the Depression was historically significant. Significance is interpretive and contextual. It connects past events to present concerns and broader themes in history. It cannot be conveyed through contrast pairs and example sequences.

Engelmann’s framework begins only after these decisions have been made.

Why this wasn’t obvious (to me)

The incompleteness of Engelmann’s framework is easy to miss because of the domains in which it was first developed. Early literacy, arithmetic, and logical reasoning are domains where knowledge really does decompose cleanly into discrete and agreed upon components. The relationship between letters and sounds is what it is. Arithmetic procedures can be specified as unambiguous rules. Logical classifications can be expressed through clear discriminations.

In these domains, the decomposition problem is almost trivial. The knowledge structure itself suggests how it should be carved into teachable pieces. The classroom teacher therefore makes almost no decisions about what to teach. When the decomposition is straightforward, communication really is the main challenge. Once you know what to teach, the remaining question is how to explain it clearly.

This makes Engelmann’s framework appear complete.

But move into domains where knowledge is more contextual and the picture changes. Concepts overlap. Understanding depends on relationships between ideas. Experts compress what novices experience as separate pieces into integrated mental models. In these domains, deciding how to carve knowledge into teachable units becomes the hardest part of the work.

In these domains, teachers have an important role in interpreting how the knowledge construct should be represented to a particular class. They decide what simplifications are acceptable and productive for a class at a particular point in time. They assess what prior knowledge students have already accumulated, often from sources that no curriculum designer had a role in planning.

All of these decisions involve compressing something continuous and contextual into something discrete and teachable. They are the intellectual work of curriculum design.

By the time those decisions are finished, Engelmann’s framework becomes applicable.

The problem downstream

I also suspect that Engelmann is incomplete in the opposite direction too, downstream of initial instruction. This is a more contentious claim, and one I want to develop properly in a separate post rather than squeeze into an argument that’s already long enough. But the outline is worth stating.

Engelmann’s methods are extremely effective at building reliable routines or procedures that learners can execute fluently and correctly. The entire apparatus of carefully sequenced practice and immediate correction is designed to move students from “can’t do it” to “can do it automatically.” His Theory of Instruction finishes with fluent execution. For many skills, that is exactly what we want.

But the strongest students don’t rely on a vast bank of memorised routines. They develop flexible understanding - mental models that allow them to navigate unfamiliar situations, see multiple routes through a problem, and adapt when the standard procedure doesn’t fit. Reliable routine execution is an essential step on this journey, but it is not the destination. Not all students make the leap from routine to flexibility, and I’m not sure anyone has a convincing universal account of how that transformation happens. But it’s unclear to me how Engelmann’s framework can produce it, because the kind of understanding involved transcends the inductivism his theory is built on. That’s for another post.

Engelmann last

None of this makes Engelmann wrong, at least for the domains where his Theory works well. It makes him late. By the time you are ready to apply his principles, the hardest design work should already be finished.

• You have worked out whether the knowledge domain can be usefully classified according to Engelmann’s principles.

• You have decided what mental model students should build.

• You have decided what simplifications are acceptable.

• You have designed a narrative or structure that makes the knowledge memorable.

• You understand where students are starting from.

Only then do you ask: How do I communicate this so learners necessarily acquire it?

Engelmann gives a precise answer to that question. If you know exactly what discrimination or procedure students must learn, his machinery can ensure they learn it reliably.

He is a novel theorist of instructional communication, and not so much a theorist of knowledge or curriculum design.

That contribution is immensely valuable. But it is not the whole of instructional design.

It is the end of it.