If constructivist teaching is the aspirin then what exactly is the headache?

It seems that many people see constructivist teaching approaches as simply good teaching. Certainly, I used to believe that there was strong research evidence to support their use and didn’t really question this. After all, pretty much everyone seemed to agree. I had fallen for a mix of argument from popularity and argument from the authority of those who promote constructivism through education schools or as consultants.

Constructivism also possesses an element of truthiness. We all know that old-fashioned, lecture-style teaching is, well, old fashioned. And that has to be a bad thing, right?

In this post, I intend to chase the constructivist rabbit back down its hole. Brazenly mixing my metaphors, I am going to ask, “If constructivist teaching is the aspirin then what exactly is the headache?”


Headache: Poor levels of achievement

We might naively think that if constructivist teaching is simply good teaching then it should lead to better test scores. However, there is little evidence of this. Anywhere.

Instead, a whole culture of rationalisation has grown-up around constructivism and tests. To do well on tests, we merely need to regurgitate rote disconnected facts. Such regurgitation is, of course, useless and so we can dismiss evidence from test scores; evidence that tends to show the superiority of explicit instruction.

Educationalists often take their lead from America and it seems that the U.S. makes far more extensive use of multiple choice bubble tests than the rest of the world. Even so, it’s a little hard to swallow this rhetoric about tests only ever assessing rote memorisation. Even a multiple choice test, if well designed, can assess understanding; simply make one of the choices a common misconception. If the questions are then withheld until the date of the test – an advantage of much-maligned standardised tests – we can determine if students really do understand the principle rather than simply having learnt the right answer.

So I think this dismissal of tests is a convenient excuse.

Headache: Poor understanding

If we do accept the premise that most tests only assess recall then perhaps the advantage of constructivist teaching is that it enhances the non-assessed understanding.

Apparently, back in olden times, students learnt all sorts of things at school that they simply did not understand. In history, they were taught to parrot dates, oblivious to their significance. In mathematics, they were forced to memorise standard algorithms without ever understanding how these algorithms work. Presumably, the same must me true for Shanghai, Hong Kong, Korea and Singapore today whose students don’t actually understand the maths that they so skilfully apply.

I disagree that knowledge and understanding can be dichotomised in this way as qualitatively different things. Although a useful concept, understanding basically consists of more and better knowledge. To use the psychological jargon, it results from well-developed schema. If this is the case, it is unlikely that some forms of instruction will be good for improving knowledge yet quite different ones are required for developing understanding.

A key experiment by Klahr and Nigam tested this notion directly. They taught science students about the principle of controlling variables either explicitly or through the students performing their own investigations. As you might expect, far fewer in the latter group learnt the principle. But of those who did they were no better at later evaluating science fair posters. Their understanding was not superior.

Headache: Poor motivation

So perhaps constructivism does not directly lead to higher tests scores or greater understanding. On the other hand, everyone knows that traditional school is awful and boring; Ken Robinson says so. Perhaps a constructivist teaching approach is more motivating for students? Ultimately, adoption of constructivist methods will lead to a great leap forward as more students develop a passion for science, maths and history. The initial dip in test scores will be far outweighed by the greater uptake of conceptually demanding electives.

You will find studies that seem to show increased motivation from constructivist approaches. But this is usually pretty easy to explain by the fact that enthusiastic teachers implementing a new program that they believe-in will pass that enthusiasm on to their students, particularly if the approach is compared with business-as-usual.

Once you think about constructivism for a little while, the proposition that these methods are more intrinsically motivating becomes deeply implausible. Is it really motivating to be presented with novel problems that you have no idea how to solve? Indeed, the students in Project Follow Through who were exposed to a program that systematically built the skills needed for problem-solving not only outperformed other students in tests of problem-solving, but also had greater increases in self-concept when compared to more constructivist models.

I can see some logic in the motivational power of projects because students can follow their interests (if allowed). But what if a science project takes us away from learning any actual science? A few years of making posters about monster trucks might be fun but once the inevitable confrontation with reality occurs, and students realise that they don’t know any science, you are unlikely to see a huge uptake in science electives.

Constructivists also sink their own argument when they insist on designing learning experiences around the mundane and commonplace in the name of ‘relevance’. The idea that students will be more motivated by a project to discover how their local community disposes of its waste than a topic about dinosaurs is obviously absurd. There is also a whiff of chauvinism; kids like these couldn’t possibly appreciate great art or poetry or the abstract beauty of mathematics because they cannot see any farther than the end of the street.


So, what is it? What is the headache that constructivism cures? When you investigate, it seems that the idea that constructivism represents good teaching is a classic bait-and-switch.

I think that lots of people just like the idea of it.


33 thoughts on “If constructivist teaching is the aspirin then what exactly is the headache?

  1. Reblogged this on From experience to meaning… and commented:
    In our teaching training institute we teach our students behaviorism, cognitivism, social-constructivism and even connectivism as learning and knowledge theories (the first two say more something about how we learn, constructivism and connectivism tell more something about knowledge but are rather weak on how people learn). We don’t defend one theorie above the others. E.g. behaviorism works great for learning tables and other stuff we need to automatize, cognitivism has given us – even recently – great insights on how people learn, but also constructivism can have an impact on learning, even positive. E.g. if you look at problem based learning. For learning new stuff, a really bad idea (cfr Hattie), but when students do have the knowledge, the story changes. Also collaborative learning really can have a positive effect.
    It’s a mistake to make constructivism the one and only paradigm you believe in, but luckily fewer and fewer people do this. As Brown 2012 has shown, in daily practice most teachers are rather eclectic.

  2. Dylan Wiliam says:

    For me, to use the term “Constructivist teaching” is to commit what Gilbert Ryle (1949) called a category mistake—ascribing to an object a property it cannot possibly have. It is like describing a rock as happy. This too is a category mistake because happiness is not a property that rocks are capable of possessing.

    As Pedro points out, constructivism is a stance in psychology—it is a set of beliefs and theories about what happens when learning takes place—which can, in turn, lead people to believe that certain kinds of instructional practices will improve the learning. However, as Greg points out, the (widespread) belief that a constructivist stance on learning means that we should teach in a particular way is simply not supported by the research evidence.

    Like most theories of learning, constructivism explained pretty well the things that previous theories of learning (e.g., behaviorism) did not. But it was not very good at explaining why, for many people, lecture style presentations are highly effective, and the effectiveness of drill and practice for learning multiplication facts. Other theories, such that cognition is situated (i.e., that it is tied to the context of the learning) offer other perspectives that can explain learning and failure to learn. However, it is disastrous for students’ learning if we think that we can derive theories about how to design instruction from theories about what happens when learning takes place. Designing and delivering instruction is a creative task, for which the psychological models are useful, but not determinative.

      • Christian says:

        Something I have been wondering for some time. Pedro and Dylan give, in my view, good contributions re the label ‘constructivism’. You also acknowledge, partly through the FAQ, that the term is problematic. I even seem to recall that you felt the changing labels of what ‘constructivism’ and ‘discovery learning’ are, are problematic in discussions. True, ‘No Scotsman’ is often close by.

        Yet, when it comes to a similar misnomer (imo) of ‘Direct Instruction’ you were quite active in reframing it from DI to ‘Explicit Instruction’. In my opinion that was great because it focused attention on some of the strengths of instruction. Would you be able to make a similar reframing of constructivism?

  3. Dylan Wiliam says:

    I should indeed have read your earlier post as it lays out what I was trying to say more clearly than I did. The problem is that many try to claim that constructivism justifies what they would like to see. The result is statements that are bland platitudes (respect the student) or just factually incorrect (relevance is essential). Such muddled thinking is unlikely to help learners.

  4. Stan says:

    Here is an attempt to differentiate the words understanding and knowledge.

    First attempt. Knowledge consists of the facts I can recall such as 6×7=42. Understanding is being familiar with the ideas about multiplication such as commutivity, distribution over addition, and factorization. But being familiar is really just another way to say being able to recall so this doesn’t work.

    Second attempt. Understanding is being able to see the connections between these various bits of knowledge. Again this fails as being able to see is just another euphemism for recall.

    Third attempt. Understanding is being able to use all these things I can recall to solve problems I have no known recallable answer as a way to solve the problem.

    There are three options for this type of problem. It can be one that does have an algorithmic solution that I just don’t know yet or it might have one that is very difficult to discern or it might have no algorithmic solution.

    The first option leads to teaching while withholding the method of solution in order to let the students demonstrate this definition of understanding. It is also difficult to test by a third party as students can simply be taught the methods they are expected to figure out.

    The second option is not going to help any student.

    The third option leads to giving students problems such as those found in bonus questions at jump math or the difficult questions on the various math competitions. The issue here is that no one thinks these are good place to start learning a topic and in general exposure to these is limited to those already deemed competent in what everyone understands as knowledge of the topic.

  5. To make things clearer it may have been useful to note that the Klahr and Negam study set up up two polar opposites, chosen to make the results of the experimental work clearer. Having stated that there is no accepted definition of constructivist teaching the researchers chose to test the effect of the teacher being fully present in the room against the teacher being fully absent. This defines constructivist teaching in a particular way. What you are describing with a single term covers a range of pedagogical approaches, with the teacher being less or more directive, from hands-off free for all where there is no direction given (as in the study quoted) to structured inquiry, as in Mantle of the Expert work or my own approach to behaviour support, where the teacher has a clear thought-out directive role in the inquiry. Again referring to the study quoted, as a science teacher I know that students should not be required to discover for themselves the rules that govern natural science, the meaning of control and experimental variables in experimentation in this case, any more than I would expect them to come up with the periodic table, given a few samples of metals and non-metals, a bunsen burner and a Geiger counter. I know the periodic table because I rote-learned it. Know-what knowledge is best taught by an expert teacher as a firm base for the development of know-how in performance, doing experimental work according to the rules, with know-why understanding emerging out of the combination of the two. Your interesting piece is a good illustration of why it’s hard to transpose empirical findings produced under controlled conditions into the messiness of the classroom community.

    • Geoffrey:

      “To make things clearer it may have been useful to note that the Klahr and Negam study set up up two polar opposites, chosen to make the results of the experimental work clearer.”

      I think this is correct for the “constructivist” condition, in which students are literally left to their own devices. But the direct instruction condition seems like a fairly weak contrast:

      “For each experiment, the instructor asked the children whether or not they thought the design would allow them to ‘tell for sure’ whether a variable had an effect on the outcome. At this point, the students haven’t received any instruction in how to ‘tell for sure,’ making this condition more akin to problem-based learning than explicit instruction. Then the instructor explained why each of the unconfounded experiments uniquely identified the factor that affected the outcome, and why each confounded experiment did not.”

      Students are trying to answer questions they haven’t yet been taught how to answer! Have these researchers not heard of the worked example effect? I hope Greg can explain how these principles derive from explicit instruction. Otherwise this seems like a test of two derivations of constructivism.

      • Stan says:

        I hope I am not spoiling Greg’s rather pointed question here. The first sentence in the second quote appears in the third paragraph on page 663 of the paper Greg linked to. The next sentence in the quote starting with “At this point” does not appear in the paper.

        My reading of the full paragraph from the text gives me the following understanding of what was done.
        1. Teacher describes an experiment.
        2. Teacher asks the children whether or not they thought the design would allow them to tell for sure if a variable had an effect on the outcome.
        3. Teacher explains why the experiment did or did not using the principles involved.
        4. Repeat 1 to 3 several times.

        So for the first experiment the children were asked to provide an answer without a worked example. Then given the worked example for that problem and then the next experiment.

        The next paragraph reinforces this interpretation. It also explains that reasoning for asking questions of the students before providing each worked example of the answers. This was to avoid the children being completely passive and confounding the experiment.

        Let’s assume Dan’s misquote is an honest mistake and that he sees this as a description of constructivism at work. Perhaps he can clear up the obvious questions this begs. Would the outcome and conclusion have been significantly different if the experiments had done the first example without step 2? Is there any point to be made by looking at the continuum of pure instruction and pure discovery and is this study making any distinction between two points along this continuum?

        We will know if Dan’s misquote was an honest mistake as he want to quickly explain it.

      • Good catch, Stan. An errant copy/paste job.

        From pg. 663 in Klahr & Nigam:

        “For each experiment, the instructor asked the children whether or not they thought the design would allow them to ‘tell for sure’ whether a variable had an effect on the outcome. Then the instructor explained why each of the unconfounded experiments uniquely identified the factor that affected the outcome, and why each confounded experiment did not.””

        My question of Greg still stands.

      • Stan says:

        My questions also still stand.

        The authors are explicit that they ask the question to engage their audience and they are aiming to avoid discovery learning so we can expect they don’t wait for answers while the students discover the solution even for the first case before any of the worked examples.

        So the problem is asking a question to engage them with the problem can lead to some people deciding this is clearly constructivist teaching.

        So the word loses all meaning. Constructivist teaching means anything except assigning reading from a dictionary.

        There are lots of folks out there that think it means something much more specific and the people using the term owe it to their readers to say what they mean and what they don’t mean by it.

      • Your question is based upon a false premise. Students are *not* trying to answer a question that they have not been taught how to answer. The procedures are demonstrated and fully explained. The clue is that the experimenters describe this as direct instruction. The question here operates in a similar way to stating a learning intention.

        But since you think it is important to answer questions then perhaps you could have a go at this one?


      • “For each experiment, the instructor asked the children whether or not they thought the design would allow them to ‘tell for sure’ whether a variable had an effect on the outcome.”

        If this were intended as the recitation of a learning standard, why wouldn’t they /say/ they were reciting a learning standard. You’re reading the text as you wish it to be, Greg, not as it is. I can’t help here.

      • Stan says:

        Let’s try another quote
        “The main distinction is that in direct instruction, the instructor provided good
        and bad examples of CVS, explained what the differences were between
        them, and told the students how and why CVS worked, whereas
        in the discovery condition, there were no examples and no explanations,
        even though there was an equivalent amount of design and
        manipulation of materials.” from page 663 paragraph 4.

        So any child that didn’t discover the general principle after the first question was given a full explanation. Any that didn’t get it after the second question had two full explanations and so on. And they were not given time to experiment between getting a question and answer the question was only posed to get student engagement. The question was also a closed yes or no question not a request for them to explain their answer.

        Dan is this really what you call a constructivist teaching approach?

      • “And they were not given time to experiment between getting a question and answer the question was only posed to get student engagement.”

        I don’t think either part of that assumption is correct, Stan.

        I wish the researchers were more specific about how long students spent on the prediction question, but your idea that they spent zero seconds in thinking about it seems farfetched to me.

        Also, it is only your assumption that this question resulted in engagement alone. Lisa Kasmer’s research has demonstrated cognitive benefits to prediction, not just affective benefits.

        My opinion is that a better test of direct instruction would have had the instructor explicitly instruct students in those four example experiments. Nothing more.

        My other opinion is that the discovery condition represents a model of instruction that very few people of any influence seriously consider.

  6. One problem we would all agree with I think, is that ‘constructivism’ means radically different things to different people, and its meaning has changed since the 1960s. Something similar has happened with Direct Instruction, Explicit Teaching, and my own favourite version of this idea ‘Whole Class Interactive Teaching’. Meanings for these are often what you wish they meant, as Dylan says above.

    I would prefer we either drop these terms and start talking about the nature of quality learning and teaching:

    Or we accept that the following notion, once called ‘constructivism’, has been blown out of the water:
    1 Don’t teach them anything, let them work it out for themselves
    2 Make everything relevant to student experience
    3 When you use discovery learning don’t give any guidance
    4 Get students using high order thinking, but don’t tell them how to do it.
    5. Praise everything in sight

    ( I exaggerate for effect, lets call this ‘old constructivism’, or perhaps ‘Greg’s constructivism’!

    Does anyone believe the above nostrums now? But a baby has been thrown out with this bathwater if we don’t salvage from old constructivism a most important set of ideas, which I believe there is very strong evidence for, and which can guide instruction (along with other principles). Good explicit teaching or good whole class interactive teaching would conform to these principles as well as good GUIDED discovery learning. The principles might be something like:

    1. Set tasks that require each student to create their own meaning of what you are teaching (their ‘construct’)

    2. Set tasks in a way that gives each student feedback on any errors or omissions in their construct, allowing them to correct it. (For example require peer to peer discussion on tasks and questions, and have genuine class discussion, led in a way that requires students to express their understandings and examine them openly. This allows constructs to be checked and corrected by the learner, by peers and by teacher.). This is ‘Interactive teaching’. http://geoffpetty.com/training-materials/questioning/

    3. Include challenging tasks towards the end of the teaching of each topic, as then, students’ constructs are tested, checked and corrected more severely than with easy tasks.

    4. When using Discovery Learning ensure that there is sufficient prior learning, guidance, and class discussion for students to make the discovery and understand it. (This approach has a good effect size, see Cognitive Acceleration for example or Furtak, E. (2012) ‘Experimental and quasi-experimental studies of inquiry-based science teaching: a meta-analysis’. I agree students might not be able to devise the periodic table single handed, but they might be able to predict the properties of an noble gas from its position on the table and from the properties of the other noble gases.

    5. At the beginning of a topic check and correct any prior learning necessary to form the constructs you are trying to get students to create.

    6. etc

    I am seriously considering dropping the term ‘constructivism’ and going for ‘co-constructivism’ as this captures the idea that peers, teacher, and learner should be working together to ensure students create and steadily improve deep meanings for what they are learning. An alternative would be to use the term ‘Quality Learning’. Any thoughts?

  7. Pingback: David Klahr writes | Filling the pail

  8. Pingback: The difference between what teachers think and do (on constructivism) | From experience to meaning...

  9. Pingback: On “powerful” progressive teaching: using ed-tech and storyboarding to secure key concepts | Education: the sacred and the profane

  10. Pingback: Top Blogs of the Week : Schools Week (September 2015) | Scenes From The Battleground

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.