Paul A Kirschner is Distinguished University Professor at the Open University of the Netherlands and Visiting Professor of Education at the University of Oulu, Finland. He is an internationally recognised expert in the fields of educational psychology and instructional design. He is past President of the International Society for the Learning Sciences and former member of the Dutch Educational Council. He is also a member of the Scientific Technical Council of the Foundation for University Computing Facilities, chief editor of Journal of Computer Assisted Learning and associate editor of Computers in Human Behaviour.
His seminal paper, ‘Why minimal guidance during instruction does not work: an analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry- based teaching’, was published in 2006, co- written with John Sweller and Richard E Clark. One of the most cited papers in education, it revolutionised the attitudes of many towards the effectiveness of enquiry versus direct instruction.
researchEDitor Tom Bennett spoke to him in the British Museum, London, where they discussed that paper, what led him to write it, and the fallout afterward.
TB: What was your own education like, and how did that then lead to the career where you are?
PK: Okay, my own education. I was a top student at elementary, junior high and high school. I used to get excellent in everything except conduct. It started out with good, and then it dropped to fair and then poor. ‘If Paul could only learn to keep his mouth shut…’ – this type of thing. That was until I was 12 years old.
But I was also a very good student. I mean I never got below an A in things like mathematics and stuff like that. Same thing in junior high school. I went to the best high school in the United States: the Bronx High School of Science. 10,000 children took an entrance test and 900 were chosen to do it. You couldn’t take the test, unless they thought you were good enough to do it. I also got good grades there and then went on to university. I started as an electrical engineer at Syracuse University.
PK: Yeah, but it was a university – a semi-Ivy League university – and while I was protesting the Vietnam War, they were protesting that they wanted to have girls in the dormitory wards.
PK: My problem was ‘Can I afford to buy a Toyota?’ and their problem was ‘Should Dad buy me a Corvette?’ – something like that.
So I transferred to a different university: the State University of Stony Brook. Primarily known for being busted twice by the Suffolk County police department for marijuana, it had a very advanced engineering school. I got there after a year of getting As in Syracuse and went on to just struggling to get by at Stony Brook. So I decided there’s one of two things I could do: I could either really buckle down and really work hard – but I had no idea how to do that because I’ve never done that; or I could do something else. So I decided to just think of something else. What can I do? Psychology – that’s almost a science!
TB: [laughs] That’s quite a leap though.
PK: Yeah, so what I did was I took some more physics and chemistry courses and a few biology courses, and so when I graduated Stony Brook I ended up with a bachelor’s in psychology and a teaching certificate for chemistry, mathematics and general science in high school.
And I had no idea what I wanted to do. All I knew was that I didn’t want to go and get a master’s somewhere in the United States. So I left for a year to get my head together. I went away from my family parents and friends…and never came back. That was 1973. And after I’d worked as a carpenter and a cook, and head of a restaurant in Amsterdam and planning on emigrating to New Zealand to become a teacher there, Catherine [his partner] decided at the last moment that she didn’t want to emigrate. She said, ‘Why don’t you go back to university?’
I went back to university – the City University of Amsterdam – and got a master’s in educational psychology. After that, first at an educational publisher and then at the Open University of the Netherlands, I went to work on my specialisation: text characteristics and learning processes. That’s the study of what you can do with text to try to ensure that people study in a way that facilitates their learning. And the rest, as they say, is history.
TB: Indeed. But why not teach?
PK: I realised one thing while teaching, namely that teaching was too frustrating for me. So I wanted to learn why the normal children that I was trying to reach – independent of how I explained things to them – weren’t learning. It was very frustrating for me because I myself was a very good student; I didn’t understand why they didn’t understand and couldn’t understand.
TB: Were you teaching at this point?
PK: Middle school. Yeah, and so that was the reason I thought I didn’t want to be a teacher. I don’t want to go back to university and the United States and get my master’s whatever and get my permanent certification as a teacher – it was just too frustrating for me. So after bumming around for a few years in Europe and the East (hippie time), I went back to school in the Netherlands – Amsterdam. My driving force was to understand how people study, how people learned and how you could make effective, efficient and enjoyable learning experiences for them.
And that’s what I’ve been doing since I started my university career in the Netherlands – 1976. We’re now 43 years later and I have been doing that exact same thing with different names for different jobs for those 43 years.
TB: What was it like going from natural/physical science to something like educational psychology?
PK: Well, there was a step in-between: I made the change at Stony Brook. You have to realise this was 1968 and the cognitive revolution had just started. And for me, the cognitive revolution began with Gagné’s third edition of The Conditions of Learning. Up to that moment, psychology was behaviouristic, and that was Stony Brook also. Dave Ausubel with meaningful verbal learning was (I think) 1966, so the seminal works in cognitive psychology hadn’t actually happened. Baddeley and Hitch was after that, so at that point in time I went over to behaviouristic psychology. And behaviouristic psychology is very, very ‘hard’ science. I mean, I even had my own lab rat.
I didn’t teach it anything, except ‘press this bar and food comes’ – kind of like how to open the refrigerator door. And I copied the lab manuals of the semester before mine because I didn’t like doing that to an animal. And when they would starve it in the holiday to see if it would learn better if it’s hungrier, I would come in every day and feed my rat and make sure that it wasn’t hungry, and do those types of things.
TB: I’m sure the rat was grateful.
PK: It was a very exact psychology at that point in time. The idea of a brain, and processing and learning like that, it was a stimulus response. It was based upon Skinner and the like. So it was a very ‘hard’ social science. When I started studying again, ten years later in the Netherlands, it had made the transition to cognitive psychology and in that point in time I was dealing with things like the use of adjunct questions from Ernie Rothkopf – as well as his work on mathemagenic activities – all of those types of things, so it was a re-introduction, a re-christening in the psychology, but then in the cognitive psychology. And I’ve been there ever since.
TB: I have to zoom in a little bit on your seminal work with John Sweller. How did that come about?
PK: Dick [Richard E] Clark’s story is different from my story. There’s an interview with him where he says how it happens. The way it happened – at least as far as I can remember – was like this. At an international conference, there were these people pontificating about constructivism and inquiry-based learning. John made a comment there. I had met him before that and had long discussions with him in the Netherlands. Afterwards I said to John something to the effect of ‘These people don’t understand what’s actually going on. It’s not that they’re unwilling, it’s just they don’t understand it.’ From that came the idea of writing this paper. The original title was ‘Inquiry learning isn’t’, which John thought was just a little too quizzical and whimsical for his taste.
TB: It’s more you?
PK: Yeah it’s definitely me. So we started working on that and talking to each other. And at a certain point, he said, ‘Well I know that Dick Clark has been doing some review work’ on things like that for Review of Educational Research and other journals. ‘He would be a good person to bounce it off.’ I also knew Dick, so I went up to him at a conference and asked: ‘You know, John and I are doing this and we would like to have you as a critical reader and bounce some things off of you.’ He said yes, and we sent him the first version of it and he came back with the question: ‘I’m a bit embarrassed to ask, but could I be the third author? This is an incredible thing that you are doing.’ For John and me the answer was a no-brainer: ‘Of course!’
So that’s how it happened. We first thought of going to Educational Researcher from the APA. We didn’t send it to them; we just asked if they wanted it. They were lukewarm. Eventually we chose Educational Psychologist, which ‘accepted’ it with major changes. Two of the reviewers gave very strong critical advice and they helped us a lot – it made the paper quite a lot better. The third reviewer was a diehard constructivist and nothing we could’ve done would’ve satisfied that person.
TB: That’s perhaps unsurprising…
PK: So I got in touch with the editor. This was around the time it went over from Lyn Corno to Gale Sinatra, so I got in touch with Gale and said, ‘This is the situation: we have two critical and constructive reviewers and we can meet their demands and it will become a great article; but if we want to meet the demands of the third, it will never happen. So if you’re going to make use of that third reviewer and treat that seriously, then tell me that now and save us the effort – we’ll go to another journal.’ It wasn’t meant as a threat; it was more a promise: ‘We’ll make use of the first two, make it a much better paper, and then go somewhere else.’ And she said, ‘No, no, do the paper.’ She really liked it and so we did that. It got accepted and the rest is history. One of the most cited papers – and when Daniel Willingham comes back on Twitter and says it’s one of the most important articles in the 21st century, it’s kind of something you’ve always dreamed about.
TB: That was my next question actually: what was your reaction to that kind of praise?
PK: There are certain papers in your life that you read, that I read, and then you say, ‘Okay, those are the papers.’ I mean, if you’re talking about levels of processing, it’s Craik and Lockhart, you know? That’s something you dream about: that you’re going to write such a paper. But you also know that it’s never going to happen in your lifetime, because there are very few that reach that status. But serendipitously, this came to be; and it became a paper that I’m incredibly proud of. And it’s just an incredible feeling.
TB: Were you surprised by its success?
PK: I knew it would…raise dust, make an impact and be controversial, because at that point in time everything you heard was inquiry, discovery and constructivist, new learning and all of those types of things. But I didn’t know it would be picked up by that many people. And I didn’t know that it would lead to debates at different conferences, and a book on constructivist versus instructivist learning by Sig Tobias and Tom Duffy. Those types of things, I had no idea at all.
TB: What have been the biggest criticisms of that paper?
PK: There were two: one was Deanna Kuhn, who said we didn’t understand children.
PK: The second was that we in some way, shape or form had created a straw man that was easy to knock down – although the only thing we did was cite people and what they had actually written. And you can see it now, although the criticism has become less. The diehard constructivists have died out, maybe? But what you see in their places are apologists – inquiry learning people, discovery learning people, who then add a heavy dose of directive instruction, explicit instruction, and then somehow still call it inquiry- based learning. Where’s the discovery gone in discovery learning? If you read the review articles, they say inquiry only works if there’s enough explicit instruction – to which I say, ‘Well that’s called direct instruction.’ You explicitly teach children about something, teach them how to solve problems with what they’ve learnt, and then give them problems to solve after they have the knowledge and skills to do it. So, they still call themselves constructivists and/ or adept at discovery, or inquiry, or experiential learning. But what they’re actually doing is making use of certain aspects of discovery, either after explicit instruction or with the aid of ‘just in time’ explicit instruction. It’s no longer discovery learning.
TB: Which raises another point: I’m fascinated by I guess what you might call the ecosystem educational research inhabits – why some things ‘land’ and some things don’t. You’ve written about the idea that constructivist learning goes away, comes back, goes away, comes back…
PK: Always with a different name. Actually Rich Mayer wrote a great article about this: ‘Should there be a three- strikes rule against pure discovery learning?’
I want to be very humble about it: John, Dick and I didn’t do anything earth shattering. I mean, what we did was talk about what good teaching is and put it in a theoretical framework that could be understood. And we took constructivist ways of teaching, constructivist pedagogies, and put them against the same framework and showed that it can’t work and why it can’t work. So what we are saying is that nothing more and nothing less than good instruction from good teachers works. We told them why that is the case from an information- processing and cognitive load point of view – our cognitive architecture. We said why that was the case and that’s possibly what makes it so strong, so robust, and so long- lasting, because we didn’t come up with a new fad or a new name for something – we just explained why and how good teaching works.
TB: But nobody has made a single significant or serious pushback against this paper?
PK: No! But that’s the author speaking here.
TB: So why are people still so resistant to this?
PK: Because it doesn’t fit in with their idea of explicit instruction. There are at least two or three reasons for it.
Firstly, people don’t understand what explicit instruction is. They think that you are talking about standing in front of the class and lecturing. So there are even teachers who actually do a lot of explicit teaching – and possibly do it well – who are pushing back against it because they have this strange idea of what it is. They’re creating a straw man that doesn’t exist, because nobody does that nowadays. Even in a lecture hall with 600 people, nobody does that. All they have to do is read Barak Rosenshine’s work on direct instruction and they might possibly see that they are doing direct instruction! But that’s the first reason.
The second reason is that it doesn’t fit the zeitgeist. It’s like the zeitgeist is a kind of laissez-faire approach: ‘Give that child room’, ‘The school/classroom is a prison’ – that type of thing. A romantic version of the child a la Rousseau.
In their idea, it’s kind of like we need to give our flowers room to grow and bloom. But as E D Hirsch stated, current science essentially demolishes the romantic tradition in educational thought which holds that education should develop naturally for the individual child. He states that while romanticism has produced great poetry, it has led to terrible educational ideas that have done a lot of harm to our Western nations.
And this zeitgeist problem is also seen with things like multitasking. It’s hip to think that people can do a lot of things because we see children and adolescents doing it. That’s what Marc Prensky did. He saw children multitasking but he never studied whether they were actually processing more things at once or whether they were doing it in a way that didn’t affect the outcomes. In other words: did they learn better? Did it lead to more mistakes or did it take more time to complete identical tasks?
The idea that we can multitask fits our view of the world and people believe it. But try saying the following to one of these believers: ‘Have you ever watched the news on television and had your partner walk in and ask you something? And you give them an answer, and then you’ve missed what has happened in the news. You heard it; you possibly even saw it. But you were thinking about something else.’
Or maybe explain it like this: ‘You’re having a discussion with a colleague at work and, while talking, you look at your computer screen to read an email that’s just arrived (the pop-up on the screen caught your eye). And at that point, your colleague asks you a question – and you have to excuse yourself because you were reading that email.
‘What you were reading wasn’t rocket science and what your colleague was speaking to you about probably wasn’t rocket science (unless you work at the European Space Agency). You definitely heard their voice (you didn’t all of a sudden become deaf) but you couldn’t process what they were saying because you were processing the text of the email. In both cases, you weren’t capable of semantically decoding one stimulus while you were at the same time semantically decoding a different one.’
At this point, the believer in multitasking will probably admit to having experienced this. But up to that point, they had the idea that they really could multitask.
The same is true for learning styles. How many times do we have to tell teachers that learning styles don’t exist? How many times to we need to present empirical research showing the contrary and they still think it exists? 93% of British teachers still think that there are learning styles and that catering to them improves learning.
TB: What’s the main takeaway for a teacher from your paper with Dick and John?
PK: Knowledge and skills are necessary to do anything further. Without those, you can’t solve problems, you can’t creatively design anything, you can’t carry out a good argument and you can’t discuss things. (Although I know a lot of people who argue without having absolutely any knowledge!) I think the main takeaway is that our brains are limited in how much they can take up at one time and how they can process that effectively, efficiently and joyfully. And if you want a learner to do that then you need to design the learning experience in a way that is synchronous with our human cognitive architecture – how our brains function. Conversely, if you do anything that contradicts how your brain functions, it won’t work. But if you do things that fit, that synchronise with human cognitive architecture, then learning will happen either more quickly with less effort – that’s my idea of efficient – more effectively, learning more deeply and learning enjoyably. Learning isn’t always fun, of course, but following these principles leads to achieving a greater feeling of achievement and success.
PK: Satisfaction, yeah. But I like ‘effective, efficient and enjoyable’. And anything that you do as a teacher – and this is possibly a second major takeaway – should be aimed at reaching at least one of those three and never to the detriment of the other two. So, if you have thought of something that makes something more effective, but it’s incredibly less efficient, then it probably won’t work. If you can make it more efficient but they learn less, you shouldn’t do it. I’m an atheist, but that’s kind of my holy trinity: effectivity, efficiency and enjoyment.
TB: Would you change anything about the paper now?
PK: If I was re-writing it now I would possibly might make more use of (or substantial use of) things like David Geary’s work on biological primary and secondary learning. We didn’t put that in because the paper was made in 1995, but his real work was in 2002, 2006.
Maybe I’d also make a slight change to talking about the cognitive load theory in it, because I’ve stopped using the three types of cognitive load – intrinsic, extraneous and germane – for a number of reasons. John is also more in that direction now at the moment. This is because there’s a certain amount of load that’s intrinsic to the task, which is based upon the complexity of the task, and a certain amount related to your own expertise – because as you become more expert, the complexity goes down.
And people have to realise that complexity is not the same thing as difficulty. You can have a very simple quantum mechanical problem, but for me it’s difficult because I don’t know quantum mechanics. It’s simple in terms of how many information elements there are and how much interaction there is between the elements. So that determines the intrinsic load. Then you have extraneous load, which is everything that deals with how you learn it: the techniques you use, the technology, all of the other things in the learning process.
And you can say, ‘Why have you got rid of germane load?’ Germane cognitive load is defined as ‘load caused by instruction that helps someone to learn’. And you can say extraneous load is ‘load that is caused by something in the environment, usually instruction, that hinders learning’. But the problem with that is that you can’t determine what is germane unless it is ex post facto or post hoc. I can only say what has helped learning if I determine the student has learned. It becomes a kind of circular way of reasoning. If someone learned from it then apparently the load was germane; but if someone didn’t learn from it, then apparently that was extraneous load.
I can measure the intrinsic load by looking at how many new information elements there are for this person and what the interaction is between them. Take playing scales on piano. I play no piano, so playing the scales is hard enough for me. A scale contains eight notes but it goes in one direction, or it goes in the other direction. The hardness/softness doesn’t change and it keeps a steady tempo, so that’s a low-complexity task. On the other hand, imagine playing a melody with fewer notes but with a greater variation in the tempo, the hardness/softness and the order of the notes. The interaction between elements is so much greater. This task is quite a lot more complex than playing a scale. So it’s always a combination of the number of elements, the number of new elements, and the amount of interaction. You can measure that beforehand; you can see it and put it into it a formula and say, ‘Okay, this is the intrinsic load of this task.’ And I can make a task more or less complex by adding or subtracting information elements or changing the level of interaction between the elements.
But I can’t do that with the other types of load – germane and extraneous – because they arise from the way the task is presented and the way you instruct in it. And if someone learns, then apparently the load that was created (and that was measured) was germane. If they didn’t learn, then apparently what you’ve caused was, in the old model, extraneous. And so it doesn’t make sense to keep talking about three types of cognitive load if I can only measure one. That was one of the major criticisms of the cognitive load model. As Slava Kalyuga noted, germane load is essentially indistinguishable from intrinsic load because it’s associated with task-related processes which are sources of intrinsic load, and therefore germane load as a concept is redundant. The dual intrinsic/extraneous framework is sufficient and non-redundant.
TB: What are your thoughts on David Geary’s biologically primary learning?
PK: It was incredibly insightful; but if you read it, it was so incredibly basic – you could kick yourself and say, ‘Why didn’t I write that?’
TB: [laughs] Yeah.
PK: There are certain things that are evolutionarily determined because if they weren’t there, the species would’ve died. For example: recognising someone’s face, communicating with a parent, having a sense of community and wanting to be with others, etc. Without these, a baby is doomed to die. So a child that doesn’t recognise its parent’s face won’t reach adulthood and won’t procreate. It’s incredibly basic that such a thing exists. This leads to things Geary discusses like folk psychology, folk biology and folk physics that are there because we need them to survive.
For example, in the wild, if a bush moved unexpectedly then we needed a flight reflex to get away. Because while it might have been a rabbit, it might actually have been a tiger. And if you didn’t have that reflex, you were probably consumed by the tiger. That’s biologically primary learning: you don’t have to teach a child that. Some people say that because we learn a first language that way, we can learn a second language like that. But they don’t understand – the second language is different: writing, reading, those things aren’t necessary for your primary survival. Those are secondary knowledge bases, and we need to teach those things more explicitly. And that’s something that’s usually the result of schools. It’s such an incredibly simple theory in the most positive sense of the word, as in, ‘Why didn’t I think of that and write it down?’ That’s how good it is.
TB: You’re only allowed one major breakthrough! [laughs]
PK: That’s the type of paper it is. It’s an incredible eye-opener, and it gets to the core of something – like John’s cognitive load theory. It brings together things like information processing from Baddeley and Hitch. You have sensory information, and you have long-term and short- term memory. And information held in the short-term memory is lost if you don’t repeat it after a certain period of time. And what happens if you read your slides to your audience while they also read them silently to themselves? You’re asking them to semantically process what they are reading and hearing at the same time. They just can’t do that! They’ll learn less, but you think they’re learning more because you’re saying it twice, in two different ways.
TB: A lot of teachers think they’re teaching with greater impact that way.
PK: But you’re not! If you had a picture that they were iconically interpreting in what is known as a visuospatial sketchpad, while on the other hand the words that they’re hearing is being semantically decoded in what’s called the phonological loop, then that’s dual coding from Allan Paivio. This, along with cognitive load theory, is one of the foundations of the multimedia principle in Rich Mayer’s cognitive theory of multimedia learning. Once again it’s one of those theories that is so robust, it can explain almost anything. It explains why certain things work for experts and not for novices: the expertise reversal effect. It’s at the roots, it’s foundational. It’s about Geary’s work and John’s work, and it’s incredibly foundational because they both deal with the essence of learning. Firstly: how our brain evolved and what that means for learning and education, and secondly, how does our cognitive architecture function and what does that mean for learning and education? What do you need that’s more fundamental than that?
TB: One last question: what are you working on now?
PK: Two things. I just published an article, which I hope will have an impact and that’s called ‘From cognitive load theory to collaborative cognitive load theory’. It expands cognitive load theory to collaborative learning situations.
TB: That will grab a lot of people.
PK: I hope so. It came about from the fact that I used have a chair in computer-supportive collaborative learning and one of the things that intrigued me was what I call transaction costs and transactive activities. That’s when you’re working with someone else on something, and you have to expend time, effort, energy on communicating about and coordinating what you’re doing with others. Those are intrinsic costs to the task of learning collaboratively. If the task itself isn’t complex enough that the benefits of working together with other don’t exceed the transaction costs caused by working together, then people won’t work together.
That’s one of the many things that’s problematic for teachers using collaborative learning. They really don’t either think about or are capable of designing tasks that are complex enough to require collaboration. They can’t or don’ make tasks where the benefits of working together are greater than the costs caused by transactive activities. And you’ll see that they’ll say things like ‘ you have to contribute five things to the discussion group’ because people don’t communicate and contribute enough.
The second is my little secret. All I’ll say is that it deals with modern assessment. I’m trying to find funding to do the research, but don’t want to alert any hijackers on the horizon.
1. Kirschner, P. A., Sweller, J., Kirschner, F. and Zambrano, J. (2018) ‘From cognitive load theory to collaborative cognitive load theory’, International Journal of Computer-Supported Collaborative Learning 13 (2) pp. 213–233.