A Pedagogy for Powerful Knowledge

Why dialogue, enquiry and cognitive science aren’t competing answers

Here’s a guess about you, reading this. You’re probably already on a side.

There’s a talk-and-discussion camp: people who believe classroom dialogue is how thinking actually develops, and that a lesson without discussion is a lesson where nothing much happened. There’s an enquiry-and-investigation camp: people who believe pupils learn by investigating real questions, not by being told the answers in advance. And there’s a cognitive-science-and-retrieval camp: people who believe learning depends on memory and how knowledge is structured, not on how busy or engaged a room looks.

Each camp has good evidence behind it. Each camp probably has a fairly sharp critique of the other two, one you’ve heard delivered from a conference stage, or read in a Twitter thread, or felt rising in your own chest while reading this paragraph.

I’m going to argue that you’re all right about what you’ve got, and wrong about what you’re missing.

Dialogue, enquiry and cognitive science aren’t three competing answers to the question of how we should teach. They are three different things that all have to be true at the same time, in the same classroom, for any of them to work properly. Take one away, and the other two don’t just change. They get weaker. That’s the whole argument. Everything that follows is me trying to convince you it’s true, and trying to show you what it looks like in practice.

Which camp are you in?

Look at the three again, because it’s worth recognising them properly before taking any of them apart.

Talk and discussion: the conviction that pupils need to articulate, test and revise their ideas out loud, because thinking that never gets spoken rarely gets sharpened.

Enquiry and investigation: the conviction that pupils should be doing something with knowledge (comparing, weighing evidence, reaching a judgement) rather than simply receiving it.

Cognitive science and retrieval: the conviction that none of that reasoning is possible if the knowledge required for it isn’t already secure in long-term memory.

None of these convictions is wrong. Decades of research sit behind each one. What’s wrong is the idea that you have to choose between them: that picking one well makes the other two optional extras, or worse, distractions from the real work. I don’t think that’s true. I think these three are interdependent in a way the field doesn’t talk about nearly often enough, and I want to spend the rest of this piece showing you exactly how, using one example, built three different ways.

The thesis

Stated as plainly as I can put it: these aren’t three competing answers. They’re three things that all have to be true at once.

Dialogue without knowledge to reason with isn’t dialogue: it’s confident guessing in turns. Enquiry without the knowledge to interpret evidence isn’t enquiry: it’s overload dressed up as investigation. And cognitive science without a disciplinary question to serve isn’t learning: it’s fluent recall of things pupils can’t yet use.

Remove any one of the three, and the other two don’t just lose something. They stop working properly. That’s not a metaphor. I’m going to show you exactly where each one breaks down, and then show you what it looks like when all three are running together in the same sequence of lessons.

Knowledge is only potentially powerful

Before getting into the three lenses themselves, there’s one idea everything else sits on top of.

Specialised knowledge (the stuff a good curriculum entitles pupils to) is only ever potentially powerful. This is Margaret Roberts’s phrase, and it’s the right one: whatever knowledge a curriculum selects and justifies, pupils do not necessarily learn it simply because it has been prescribed (Roberts, 2014). It doesn’t do anything just by being written into a scheme of work, or sitting correctly sequenced in a knowledge organiser. A pupil can have access to brilliant, ambitious content and still come away from it unable to use any of it.

That’s because knowledge becomes usable, not by being delivered well, but by being converted. Pupils have to be able to reason with it, investigate with it, and hold onto it long enough across time to draw on it when it matters. Dialogue, enquiry and cognitive science are the three things that do that converting. They’re not three optional add-ons bolted onto good teaching once the knowledge is in place. They are the mechanism by which knowledge becomes something a pupil can think with at all.

I want to be upfront about something. This isn’t a geography piece wearing a disguise, even though I’m going to use a geography example to carry it. The argument is about how any discipline’s knowledge gets converted from something pupils have been told into something they can actually use. Whatever you teach, ask yourself as you read: what’s the version of this in my subject? Where’s the question in my discipline where the explanation genuinely depends on weighing competing causes, not just naming the right one? Hold that question. We’ll come back to it properly near the end.

One example, three lenses

Here’s the example I’m going to build three times, because abstract principles are forgettable, and one example built carefully is not.

The enquiry question is this: how and why might climate migration reshape the future of cities? It’s a Year 9 geography enquiry, and it runs through a comparison of two contrasting places: Dhaka, which is low-lying and lower-income, and a wealthier coastal city like Miami or Rotterdam, which faces broadly similar physical hazards but very different resources for responding to them.

I’m going to come back to this exact question three times. First through the lens of dialogue, to show what happens when pupils actually talk about it properly. Then through the lens of enquiry, to show what gives the whole sequence its shape and its purpose. And then through the lens of cognitive science, to show what had to be true in pupils’ memories before either of the first two could happen at all.

By the end, you’ll have seen the same lesson sequence three times, through three different lenses, and I think you’ll see it’s actually been the same lesson the whole way through, because all three were operating inside it together.

Before I start building it, here’s something worth holding in your head rather than answering out loud. If you teach history, what’s your version of this question: one where the explanation doesn’t resolve to a single tidy cause, but depends on weighing several competing ones against each other? If you teach science, where’s the question in your subject that won’t settle with one correct answer, the kind that needs pupils to interpret evidence rather than just recall it? I’m going to ask you to translate what you’re reading into your own subject a few more times before we’re done, and the work you do answering it quietly to yourself is, in a real sense, the actual point of this piece.

Part One: Dialogue

There’s a distinction the field is genuinely sloppy about, because the word “dialogue” gets used to mean almost anything that involves pupils opening their mouths. Discussion and dialogue are not the same thing, and the difference between them matters more than almost anything else here.

Discussion vs dialogue

Discussion is what happens when pupils exchange views. They take turns. They share an opinion, somebody else shares a different one, and most of them leave the room holding more or less the opinion they walked in with. It can be lively. It can feel inclusive. It is not, on its own, transformative.

Dialogue is different, and Robin Alexander’s account of it is the clearest I know. He describes dialogic teaching as collective, reciprocal, supportive, cumulative and purposeful (Alexander, 2017). The key word for me is cumulative. In dialogue, ideas get built, not just exchanged. One pupil’s contribution genuinely changes the shape of the next pupil’s thinking, rather than simply sitting alongside it.

Neil Mercer has a lovely term for this: interthinking, the idea that pupils use talk to think together, not merely to take turns speaking in the same room (Mercer, 2000). Here’s a test you can apply in your own classroom this week: did anyone’s explanation change shape because of what someone else just said? If the answer’s no, you had a discussion. If the answer’s yes, you had dialogue, and dialogue is where the disciplinary reasoning actually happens.

This matters for any subject where the questions are genuinely contested, and that’s most subjects, properly taught. If your discipline has questions worth disagreeing about, then pupils need practice testing their explanations against someone else’s reasoning, not just stating a view and moving on. That’s not an argument for more talk. It’s an argument for talk that’s built to go somewhere.

Building an explanation, turn by turn

Let’s put this against the climate migration example. By this point in the sequence, pupils have already built solid knowledge of climate hazards, migration drivers, and how cities grow (I’ll come back to exactly how shortly, because that’s not an accident, it’s the third lens of this piece). With that knowledge in place, they’re given the comparison: Dhaka against Miami or Rotterdam, and asked why a wealthier city facing similar hazards responds so differently.

Watch what a good dialogic sequence does here. One pupil proposes that money buys protection: better defences, faster recovery, that kind of thing. A second pupil complicates that: but what about the difference between being displaced and choosing to relocate? Some people in Dhaka don’t get to choose. A third pupil pushes further still: so who actually gets to choose to move, and who doesn’t, and why?

Notice what’s happened across those three contributions. Nobody just restated an opinion. Each pupil’s explanation depended on, and changed because of, what the pupil before them said. That chain of talk is the disciplinary reasoning. It isn’t decorating the geography. It is how the explanation about vulnerability and inequality actually gets constructed, in real time.

If you teach history, this is the moment two pupils are arguing over the causes of an event, and one of them genuinely revises their account halfway through because the other has just produced a piece of evidence they hadn’t weighed properly. If you teach science, it’s two pupils with competing explanations for an anomalous result, and the talk has to resolve which explanation the evidence actually supports. Different subject, same structure: talk that builds, rather than talk that just gets shared.

Failure mode: dialogue

Here’s the failure mode, and it’s worth saying plainly because it’s the thing that makes people distrust dialogic teaching in the first place.

Dialogue without secure knowledge isn’t dialogue. It’s confident guessing in turns.

Pupils cannot interthink their way to an explanation of why one coastal city is more vulnerable than another if they don’t yet know what actually makes a coastline vulnerable in the first place. You can structure the talk as carefully as Alexander (2017) would want, give pupils sentence starters, build in turn-taking, and it will still produce nothing more than well-organised speculation if the knowledge underneath it isn’t there.

Hold onto that. It’s going to matter again when we reach the third lens.

Part Two: Enquiry

This is the section where I annoy the discussion camp and the knowledge camp at the same time, which I’m told is a good sign.

Weak enquiry vs strong enquiry

Enquiry has a definition problem. The word gets attached to almost anything: a research task, a group project, a class debate, regardless of how much disciplinary thinking is actually happening underneath it. So here’s a sharper distinction.

Weak enquiry is activity-led. It asks a generic question, it accepts pupil opinion as the endpoint, and it tends to involve low guidance: pupils set loose with a topic and some source material and left to get on with it.

Strong enquiry is concept-led. It asks a genuinely disciplinary question, one that demands evidence-informed reasoning rather than opinion, and (this is the part people miss) it depends on expert guidance, gradually released as pupils gain the capability to do more on their own.

That last point is the one I most want to land, because it’s where enquiry’s own best researchers would side with the knowledge camp, not against them. Kirschner, Sweller and Clark’s critique of minimal guidance during instruction is one of the most cited pieces of evidence against discovery learning that exists (Kirschner, Sweller & Clark, 2006). Their argument is that novices learn poorly when they’re expected to discover complex knowledge largely on their own, because without a prior schema to hang it on, they’re simply overloaded.

That doesn’t invalidate enquiry. It tells you what makes enquiry strong rather than weak. Strong enquiry isn’t low-structure. It’s high-structure, aimed deliberately at a genuine disciplinary question. The phrase I’d want you to leave with is this: enquiry isn’t pupils finding things out for themselves. It’s a carefully guided induction into how people in your discipline actually ask questions, weigh evidence, and reach defensible judgements under conditions of real complexity.

The question that gives the sequence its spine

Here’s the same climate migration question, now through the lens of enquiry rather than dialogue.

The question (how and why might climate migration reshape the future of cities) is what gives the whole sequence its spine. Without it, the comparison between Dhaka and Miami or Rotterdam is just two case studies sitting next to each other, mildly interesting, going nowhere in particular.

With the question driving it, the sequence has a clear shape. Pupils build the knowledge they need. They compare evidence between the two places. They weigh competing explanations for why vulnerability differs so much between them. And they reach a judgement: a real one, defensible, that answers the original question with actual reasoning behind it, not just a description of what they’ve learned.

Notice where the dialogue from the last section fits into this. It didn’t happen in a vacuum, and it wasn’t there because discussion is generically good for pupils. It happened precisely at the point in this structure where the enquiry called for comparison and judgement. The talk wasn’t an add-on to the enquiry. It was how the enquiry’s central intellectual work actually got done.

This is the piece I’d ask you to carry into your own subject. Enquiry isn’t a different activity from explicit teaching, and it isn’t a reward you give pupils once the knowledge is safely in place. It’s the structure that gives knowledge somewhere to go: a reason for retrieval to exist, and a reason for dialogue to matter. Without it, you can have beautifully sequenced knowledge and beautifully structured talk, and still end up with pupils who can’t tell you why any of it was worth knowing.

Failure mode: enquiry

Enquiry without secure knowledge isn’t enquiry. It’s overload.

Ask a class to investigate climate migration before they know what actually makes a coastline or a community vulnerable, and you don’t get young geographers reasoning carefully about evidence. You get pupils drowning in unfamiliar information, with nothing to hold onto it with, producing answers that look like enquiry but are really just collected facts with a question mark stuck on the front.

If you teach maths, this is the difference between telling pupils to explore patterns in a data set with no prior framework, and giving them a problem structured so they have to choose a method and justify why it’s the right one. If you teach English, it’s the difference between discussing a poem in the open and asking a question that forces pupils to build a defensible interpretation against the actual text in front of them. Same failure mode, different subject: enquiry without the knowledge to do it with is just overload wearing enquiry’s clothes.

Part Three: Cognitive Science

This is where I puncture whichever camp has been feeling reasonably smug for the last few sections, on the grounds that everything I’ve said so far has secretly depended on them being right. Let’s see if that holds up.

Cognitive science isn’t a checklist

Here’s the problem worth naming directly. In a lot of classrooms, cognitive science has been reduced to a small number of routines. A retrieval starter at the beginning of the lesson. A slide with a picture next to some text, because someone once mentioned dual coding. Content broken into smaller chunks, because cognitive load theory apparently means simplify everything. Tick, tick, tick: cognitive science, job done.

It isn’t job done, and here’s why each of those three has a problem underneath it. A retrieval starter that’s disconnected from the wider curriculum (testing isolated facts that don’t connect to what comes next) does very little to build the kind of structured knowledge pupils actually need. Dual coding misapplied becomes decoration rather than design: an image added because it looks nice, not because it’s doing any cognitive work. And cognitive load theory, badly applied, gets used as an excuse to strip complexity out of a subject altogether, when the actual point of the theory is to help you sequence complexity, not avoid it.

There’s a deeper distinction underneath all three of these, and it’s the one I most want you to leave with. Performance is not the same as learning. This is Robert Bjork’s phrase, drawn from Soderstrom and Bjork’s integrative review of the evidence (Soderstrom & Bjork, 2015), and it should worry all of us slightly more than it usually does. A class can describe a process fluently in the lesson it was taught and have learned almost nothing durable from it. Fluency in the room, in the moment, is not evidence of retention. It’s often the opposite: the conditions that make learning feel easy and immediate are frequently the conditions that make it forgettable.

Cognitive science doesn’t tell you what’s worth retaining, or why it matters in your discipline. Only your subject can tell you that. What it tells you is how to make sure that once you’ve decided what matters, it actually stays.

What made the dialogue possible

So, the third pass through the climate migration sequence, and this is where the first two lenses get explained properly.

The knowledge pupils needed for that dialogue about Dhaka and Miami (what makes a coastline vulnerable, how migration decisions get made, how cities grow and where people end up living within them) didn’t appear from nowhere in the lesson where the discussion happened. It was built across several earlier lessons, through clear explicit explanation and through spaced retrieval, deliberately revisited before the enquiry ever asked pupils to compare anything.

Think about what would have happened if that retrieval had instead been a disconnected five-minute quiz on isolated facts, the kind that gets used as a settling activity at the start of a lesson with no real connection to where the lesson is going. The dialogue in part one would have had almost nothing to build with. Pupils would have had facts available, but not the kind of structured, connected knowledge that lets one idea complicate another in real time. And the enquiry in part two would have collapsed into exactly the overload described earlier: too much unfamiliar information, no schema to hang it on.

I want to say this as plainly as I can. Cognitive science is what made the other two lessons possible. It isn’t competing with dialogue and enquiry for space in the curriculum, or for your attention as you read this. It’s sitting underneath them, doing the quiet work that makes the more visible, more exciting parts of teaching actually function.

Whatever you teach, this is the question worth asking before any lesson that depends on rich discussion or genuine investigation: what does a pupil need already secure in memory before today’s talk, or today’s enquiry, can work at all? If the honest answer is nothing, the lesson is about to fall over, however well you’ve planned the discussion.

Failure mode: cognitive science

Cognitive science without purpose isn’t learning. It’s fluent but inert.

You can build a beautifully sequenced programme of retrieval, spacing, and dual coding, and still end up with pupils who can recall an impressive quantity of disconnected facts and cannot use a single one of them to reason about anything. Knowledge retained for its own sake, without a disciplinary question it’s ever asked to serve, is knowledge that sits in memory doing nothing. It’s the opposite failure to the first two, but it’s just as damaging, and it’s the one cognitive science’s own advocates are least likely to notice in their own classrooms, because the test results usually look perfectly good.

That’s the trap. Everything looks like it’s working, because recall is exactly what you measured for, and recall is exactly what you got.

Pull one out, and the other two fail

So let’s put the three passes back into one, because that’s really the whole point of this piece.

Pupils retrieve and consolidate knowledge about hazards, migration and urban growth. That’s cognitive science doing its job. That knowledge is what lets them have a genuine dialogue about Dhaka and Miami, not a guessing match dressed up as discussion, but pupils building an explanation turn by turn, each contribution changing the shape of the next. That’s dialogue doing its job. And that dialogue happens at exactly the point it does because an enquiry question demanded a comparison and a judgement, not because discussion is generically good for pupils on a Tuesday afternoon. That’s enquiry doing its job.

Pull any one of those three out, and watch what happens to the other two. No retrieval, and the dialogue has nothing to reason with: it becomes confident guessing in turns. No dialogue, and the enquiry becomes a worksheet, pupils completing tasks without ever testing an explanation against another person’s reasoning. No enquiry question, and the retrieval has no purpose to serve: it becomes facts retained for their own sake, fluent and inert.

This is the same lesson sequence described three separate times, through three different lenses. It only worked, at any point, because all three were operating together, the whole way through. Not in sequence, taking turns. Together, continuously, each one depending on the other two to do its own job properly.

Whatever you teach, the question isn’t which of these three to prioritise this term. It’s whether all three are actually present in the same sequence of lessons, reinforcing each other, rather than living in separate parts of your curriculum where they never have to depend on one another at all.

The oppositions dissolve

Once you see it this way, I think most of the oppositions this field argues about start to look less like real choices and more like a failure to notice how dependent each side actually is on the other.

Knowledge versus skills. There’s no meaningful skill in any discipline that doesn’t depend on secure knowledge to exercise it on, and there’s no point to secure knowledge that isn’t eventually put to use doing something.

Instruction versus discussion. Meaningful dialogue depends on pupils having something worth discussing, which only careful instruction reliably gives them. And instruction without any opportunity to discuss what’s been taught rarely produces pupils who can reason with it afterwards.

Traditional versus enquiry. Strong enquiry, as I’ve already shown you, requires exactly the kind of careful, structured, expert-guided teaching that the word “traditional” is usually used to describe, the same guidance Kirschner, Sweller and Clark (2006) argue novices need throughout. The opposition was never really there.

Cognitive science versus creativity, or cognitive science versus rich disciplinary thinking. Secure memory is what makes complex, creative reasoning possible in the first place. You cannot think flexibly with knowledge you don’t securely have.

None of these are real choices once you see how dependent each side is on the one it’s supposedly opposed to. The actual challenge isn’t choosing a camp. It’s the much harder, much less glamorous work of integrating all three properly, in the same classroom, in the same sequence of lessons, week after week.

Closing

So here’s what I’d like you to take away, more than any single piece of research mentioned above.

The question worth taking away isn’t which camp you’re in. It’s this: which of the three is missing in your classroom this week? Is there knowledge sitting inert because pupils never get to discuss it or investigate it? Is there discussion happening that’s drifted loose from the knowledge it needs to be productive? Is there an enquiry running in your department right now that’s quietly overwhelming pupils because the knowledge underneath it isn’t secure yet?

Whichever one it is, that’s where the next piece of your own development sits, not in becoming more committed to the camp you’re already in, but in noticing what’s quietly failing in your classroom because one of the other two isn’t there yet.

References

Alexander, R. (2017). Towards Dialogic Teaching: Rethinking Classroom Talk (5th ed.). York: Dialogos.

Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist, 41(2), 75–86.

Mercer, N. (2000). Words and Minds: How We Use Language to Think Together. London: Routledge.

Roberts, M. (2014). Powerful knowledge and geographical education. The Curriculum Journal, 25(2), 187–209.

Soderstrom, N. C., & Bjork, R. A. (2015). Learning versus performance: An integrative review. Perspectives on Psychological Science, 10(2), 176–199.

If you want to keep arguing about this, I’m @EnserMark on X.