As people grow up, they gain knowledge of their surroundings. For example, if a child goes outside on a cold day, they will no doubt feel the sensation of cold all around them. However, when they touch something metal, it will feel more cold than something which is not metal. The child could quite reasonably conclude that metal is therefore naturally, or essentially, cold.

On an evolutionary scale, it does not matter whether or not this conclusion is correct. All that would matter is whether or not the conclusion is useful in terms of survival. The mind has evolved to generate beliefs which are evolutionarily beneficial.

In a previous post, I summarised David Geary’s theory of Educational Evolutionary Psychology. The fundamental principle is that our brains have evolved to easily acquire some knowledge, like verbal language, but not to acquire other knowledge, like written language. The former is called primary knowledge and the latter is called secondary knowledge.

John Sweller described this theory as a “rare advance” in the study of cognition. I personally have found it to be incredibly interesting and have been trying to think about scientific misconceptions in its light. Unfortunately, I could not find much, in fact anything, by way of reference to it in academic sources dealing with misconceptions.

Conceptual change

The phrase “conceptual change” is used to describe the process by which a person moves from their misconception (or their folk knowledge) to the correct conception. So when a student enters my lab believing that metals are naturally cold and leaves believing that metals are good conductors of heat and not naturally any particular temperature, they have achieved conceptual change.

The problem is how to get from A to B. All the literature I saw acknowledged that this is incredibly tricky. The dominant approach seems to be to cause cognitive conflict (CC). This is ideally where a student’s conceptions are first elicited and brought to the surface. Contradictory evidence, in the form of data, a practical or anecdote, is then presented. This contradiction causes cognitive conflict, a form of dissonance whereby the student is torn between two conceptions. Given the right evidence, the new material is accepted into the student’s cognitive architecture (their schemas) and the old, folk, conception is discarded. This philosophy is based in Piagetian constructivism.

I don’t particularly want to get bogged down in the arguments for and against constructivism as a whole. I’m more interested in what works: does this approach secure better understanding of correct conceptions?

A lack of evidence

Before I started reading into this area, it seemed to me to be obvious that this approach would not be effective. This wasn’t based on my own classroom experience, but a basic idea of the history of science. Whatever particular philosophy of science a person adopts, it’s fairly clear that throughout history theoretical conceptions of the world have been held to despite masses of evidence. I wrote about one particularly illustrative – and fatal! – example here, but the scientific revolution is replete with old theories being overturned only once the weight of contradicting evidence became overwhelming. Until then, scientists just found a way to incorporate the new evidence into old theories (as well as my example and many hundreds of others, the use of epicycles to explain planetary retrograde motion is a case in point). So I felt that the burden of evidence would certainly be on the positive hypothesis that “cognitive conflict is an effective way to achieve conceptual change.”


When I first started reading into this, I expected to be drowned in a sea of high quality, robust studies with quantitative findings. Presumably it isn’t that difficult to set up an RCT looking at different instructional techniques. Unfortunately I could only actually find a couple. I saw a lot of references to other, older studies that I couldn’t always find or didn’t really meet my expectations. There are a lot of studies without control groups, with multiple confounding variables and with qualitative analysis based on conversations with students.

Limon (2001) brings a lot of evidence that CC has been shown not to work. A number of experiments are then referenced which showed positive results but again, a number of these studies have no control group to compare CC against. The paper concludes that CC is a good route, but no fewer than fifteen variables need to be considered and optimised.

In fact I only found a couple of sources that actually conducted an RCT comparing cognitive conflict to a more explicit instructional technique. One of them (Potvin, 2015) compared three theoretical models for conceptual change; two focused on CC and one a “traditional” method of teaching which focussed on repetition. The results showed that one of the CC models is superior, but its effect sizes, despite being significant, are weak. Furthermore the experimental method used was to subject participants to videos, with the “traditional” approach having the same video twice. I’m not sure how well this maps at all onto classroom practice.

Another study (Zohar, 2003) found that cognitive conflict was more effective for high ability students but less effective for lower ability students. The authors argue that this confounding variable can explain inconsistencies in other research. Difficult to extrapolate much more from that.

There might be more stuff out there, but I didn’t find it.


Duit (2009) has a bank of 8400 articles and books on the topic of misconceptions and conceptual change. 8400 of them. 8400. That’s a lot of articles. I had a quick look but the sheer volume was overwhelming. Paying weak deference to the gods of thoroughness I searched the document for “randomised” or for “RCT” and found no references in the titles. The same happened for “explicit.” When I searched “Direct Instruction” there were two contenders; one dealing with DI vs. discovery learning, and the other (Hänze, 2007), for which I could only access the abstract, didn’t show much of a difference between DI and a “cooperative” model of learning. So no luck there either.

Suppressed or overwritten?

Nick Rose directed me to a paper (Shtulman, 2012) whose authors designed an experiment to figure out if, even in adults with a good grounding in science, original misconceptions are ever actually totally erased from memory, or if they are just suppressed. The paper is fascinating and worth a read but essentially concludes that our misconceptions are never erased. We carry them with us always, despite having formal knowledge which directly contradicts them. So I might be able to tell you that “air is indeed made of matter” but deep down my original conception that “air is not made of anything” still exists. That conception has been suppressed over many years, but it is still there. Perhaps over time, if I ceased to think and to teach about such things, it would come once again to be dominant.

The Veritasium videos are great at this. In them, the presenter will often ask questions of the public and then discuss their misconceptions with them. Presumably these people have been educated and, at one point, knew what the true conception was. Yet, over time, their original misconceptions return.

This smacks a big old hole in the constructivist approach and still just leaves us with the same challenge as before: what is the best way to tackle student misconceptions?

Primary and Secondary Misconceptions

Some of the literature vaguely flirts with the difference between misconceptions that people naturally acquire and ones that they acquire after learning something about the world. For example, the false conception that “heavy things fall faster than light things” would be one which people acquire naturally as part of their folk physics. It makes sense; when something heavy falls on you it hurts more than something light. This must be because it moves faster. I would term this a primary misconception. However, the misconception “when a substance is heated it expands because the particles expand” cannot be primary because no one naturally comes to the idea that substances are made of particles! I would therefore term this a secondary misconception.


I don’t know of any research on this but one could hypothesise that secondary misconceptions would be easier to overwrite/suppress than primary ones. Primary ones would be, well, primary; more embedded in our foundational cognitive architecture. However, the fact that none of the literature I saw notices a stark difference between misconceptions leads me to question that hypothesis (I know I know absence of evidence etc.).

Based on Geary himself I think that there is another good reason to doubt that secondary misconceptions would be easier to tackle. He argues that when we acquire secondary knowledge it is through our primary pathways; a process which he calls co-optation. As I mentioned in my summary, primary knowledge leads to secondary knowledge.

So taking the example above, why would someone think that the particles increase in size rather than think that the gaps between them increase in size? No doubt part of it is because our diagrams tend to deemphasise the space between particles, but I think there’s a better reason than that.

Driver notes that a common misconception on the origin of life is preformationism. In essence this assumes that organisms have grown from miniature versions of themselves. In a sense, we all started as tiny humans in our mothers’ wombs and, over time, our parts have grown in proportion to make us bigger humans. Vox’s scholarly article shows how depictions of children in Renaissance art often has them in the same proportions as adults which makes them look like tiny scary old men. It makes sense to assume that when things grow they grow in proportion and to not realise that babies’ heads are actually out of proportion with the rest of their bodies and that when they grow their parts grow at different rates.


If this bit of folk biology is true, then it makes sense to co-opt that theory when learning about how things expand when heated. The whole thing grows in size together, in proportion, just like expanding an image in word or powerpoint.

The same is presumably true of other secondary misconceptions. Take for example the widespread misconception that humans are descended from monkeys as opposed to sharing a common, sea dwelling ancestor. Folk biology tells us that children look like their parents, so it makes sense that we descend from animals which look like us. The secondary knowledge that we acquire in error is based on other areas of our primary knowledge.

If my hypothesis here is true, then, in terms of tackling misconceptions, there would be no difference between primary and secondary. This is a testable hypothesis and, whilst trying to avoid being that guy, I think we need more, and better, research here.

So where does this leave us?

Unfortunately, nowhere. I started with the question what is the best way to tackle student misconceptions and I don’t feel like I have a good, research based answer. The problem is that we already have over 8,000 articles on the topic. Are we any closer to the truth? Doesn’t look like it to me.

List of Things I read (I know not properly referenced and I don’t really care):

Note 1: my computer crashed with all my tabs about halfway through (no search history due to school account) so there are a few things which should be here but aren’t

Note 2: a lot of the stuff I read I only got through the Chartered College

Note 3: this was a really long post. If you made it this far, sorry about that.

Note 4: there are some people out there who know shedloads more about this than I do. This was the dipping of toes into water.

Two great twitter chats here: and here:

How do I get my students over their alternative conceptions (misconceptions) for learning?

Applying the “cognitive conflict” strategy for conceptual change—some implications, difficulties, and problems (1990)

Educating the Evolved Mind (Geary, 2007)

On the cognitive conflict as an instructional strategy for conceptual change: a critical appraisal (Limon, 2001)

Experimental Evidence of the Superiority of the Prevalence Model of Conceptual Change Over the Classical Models and Repetition (Potvin, 2015)

Cooperative learning, motivational effects, and student characteristics: An experimental study comparing cooperative learning and direct instruction in 12th grade physics classes (Hanze, 2007)

Cognitive conflict, direct teaching and student’s academic level (Zohar, 2003)

(Duit, 2009)

Veritaseum on trees

Making Sense of Secondary Science (Driver, 1993)

Teaching Secondary Science, (Ross, 2004)

Misconception bank:

Scientific knowledge suppresses but does not supplant earlier intuitions (Shtulman, 2012)

e-mc2 on diSessa

Vox on Renaissance art:

Greg Ashman:

Nick Rose:

UPDATE 28/07/17: just saw this very interesting piece from the learning scientists with more research for me to follow up on