Cognitive Load Theory and its implications for instructionPosted: June 1, 2015
I was a little perturbed to see the following tweet from celebrity maths teacher Dan Meyer in a thread that I was following:
John Sweller is one of my PhD supervisors. Although I would never attempt to put words into his mouth, I would like to point out that the devil resides in two significant details here; details around which there are some troubling misconceptions.
What is fully guided instruction?
It may seem obvious, but many people don’t actually get what fully guided instruction is. They conflate it with lectures. Once this is done, it becomes possible to draw on studies that show that lectures where students can interact via “clickers” are more effective than ones that are purely one-way. This may then be used as evidence against fully guided instruction.
Instead, fully guided instruction simply means explaining concepts and ideas and modelling processes prior to asking students to do this themselves. It places problem solving after explicit instruction but it does not require there to be no problem solving at all. In my view, good practice at fully guided instruction involves lots of interactions between teacher and students in order to ensure attention – this is why the clickers work. In some cases, as is common in East Asian countries, the practice phase is completed as a whole class.
I set out my summary of a cycle of explicit instruction here. Rosenshine made much of the early running on this when analysing the process-product studies of the 1960s. You can find his description of explicit instruction here. And the Direct Instruction that was so successful in Project Follow Through is also known for being highly interactive. All of these methods aim to move students from dependence upon explicit instruction to independence. Otherwise, what would be the point? Fully guided instruction is not just a one-way lecture.
Should we always provide fully guided instruction?
I responded to Dan Meyer and mentioned the expertise reversal effect. Slava Kalyuga, my other supervisor, has been a leading researcher in this area and he, Sweller and others have written extensively about it.
Essentially, it shows that fully guided instruction, although the most effective approach for novice learners, is not as effective for those who already have considerable expertise within the given domain. We reach a point where the effectiveness reverses and solving problems becomes a more effective way to learn. This is for a number of reasons: Fully guided instruction provides information that experts already have stored in their long term memory. If these elements cannot be easily ignored then the requirement to pay attention to them leaves less attention available to process the problem.
Christian Bokhove suggested on Twitter that Sweller’s acceptance of the expertise reversal effect indicates that his position is perhaps not as ‘extreme’ as it once was. Again, I do not wish to write on Sweller’s behalf. However, it is decidedly odd to suggest that the idea that experts are different to novices is some kind of softening of Cognitive Load Theory. There must be some point to becoming an expert. What is it, if not to be able to solve problems independently? For instance, experienced readers do not require continual phonics instruction. Expertise must gradually develop over the process of learning something. What Kalyuga and others have done is uncover something of the mechanism.
Indeed, the central problem with constructivist teaching strategies is that they don’t properly acknowledge the differences between novices and experts. “Experts do X,” say constructivists, “and so we must make novices also do X”. It is the fallacy that by mimicking what experts do we can become more expert. Hence we see science students running open-ended investigations and Jo Boaler noting that maths PhD students do maths differently to how it happens in your average maths classroom.
The key point is when and how we release students from fully guided instruction to more open-ended problem solving. I would suggest that some students studying coding, for example, might be such enthusiastic hobby coders that their level of expertise is such that they don’t need much explicit instruction. However, most students studying a new mathematics concept will need the careful guidance of an expert before they can start solving problems on their own.
You certainly don’t start the process by removing instructional supports or withholding guidance.