I am going to describe some ways that knowledge of cognitive load theory (CLT) has changed how I teach. Prior to learning about CLT, I used what I have previously termed ‘default teacher led instruction’. I stood at the front and explained things but it was not an optimal form of explicit instruction. On the positive side, I had developed good classroom management over time and, since the mid 2000s, I had implemented some of Dylan Wiliam’s formative assessment techniques. This was, on reflection, an excellent basis for learning about CLT.
I am going to write in the context of my maths teaching because this provides the clearest examples. Maths is also an area that has been heavily targeted in CLT research. However, I have been involved in transferring these ideas to other subjects and have always found such transfer to be complicated but possible – again this mirrors the research. But that discussion is for another day.
1. I don’t tend to read out my slides
I have a PowerPoint for every lesson that acts as my lesson plan. It has a starter activity, lists the homework, states the learning intention, reminds me to take the register, displays any notes and contains all class examples and questions, along with the solutions on subsequent slides.
I print a selection of these slides and hand them to my students. They glue the slides in to their exercise books, annotate them and then answer any questions in their books.
When we get to a slide that presents some theory such as the origin of Euler’s number, I ask the students to read it and I give them ample time to do this before I start talking about the content of the slide. I often use this separation-in-time strategy to work around cognitive load issues: Instead of doing two things at the same time, I do them one after the other.
I avoid reading out my slides because of the ‘redundancy effect’ found in CLT: If we provide additional, unnecessary information for students to process then is far from harmless. If all the information is available for students in the text, then the addition of a narration, particularly if it is not faithful to exactly what is written on the slide, will cause unnecessary cognitive load.
This issue is also related to dual channel processing and is, as such, is a feature of Mayer’s cognitive theory of multimedia learning which is closely related to CLT: Our working memories effectively have two ‘channels’. One processes visual information and the other processes spoken language. Reading starts off in the first channel with the letters being observed visually. These are then decoded into virtual sounds that are processed in the spoken language channel. If a student is required to process spoken language at the same time as reading then this jams up the spoken language channel. Some students may be able to select either the spoken or the written and pay attention to just one of the sources but it likely benefits nobody.
The redundancy effect does depend on the information actually being redundant. If you cannot read a word or it is not in your vocabulary then information about the word and what it means is not redundant.
2. Break it down, further
New learning can be fairly transient. Information will quickly disappear from students’ minds without the opportunity to process it.
I used to present a series of examples before I asked students to complete a task. For instance, I might show them how to sketch an exponential curve before reminding them of how to apply index laws. They would then complete an exercise on these problems.
These days, I pause for practice between each individual problem type. So after the graph example, students will complete a graph question and after an index law example, students complete an index law question and so on. Which leads to the next technique.
3. Example-problem pairs
In research into the use of worked examples, the use of example-problem pairs was found to be optimal for learning. A worked example is written on one side of a page with an almost identical question posed on the other side. This way, students can apply the method of the example directly to the question. This minimises extraneous cognitive load and focuses attention on the key features.
I replicate this with my slides: one side has an example and the other has a question.
If we step back a little from CLT then I think example-problem pairs serve a wider purpose. In his review of process-product research, Barak Rosenshine suggests that effective teachers aim for an 80% success rate. This is one way to achieve success and I suspect it is motivating for students.
In the past I think I have gone for transfer too early. After demonstrating a worked example, I have tended to ask slightly different questions, thinking that this will lead to more flexible learning but I probably just caused frustration in students who were struggling.
4. Stop after five minutes
And I’ve come to a much more nuanced understanding of cognitive struggle. If you know little else about CLT then you probably are aware of the idea that too much struggle can overload working memory. However, it’s not just the total amount of cognitive load that matters but the type of load.
Dan Willingham tells an anecdote in ‘Why don’t students like school‘ that has been referenced on many blogs. He writes of a teacher who wants students to learn about the Underground Railroad, the historical system of safe-houses and routes that were used to smuggle slaves out of the south of the U.S. The teacher asks the students to bake biscuits of the kind that the freed slaves would have eaten. Willingham points out this would have caused the student to think about measuring butter and flour rather than the Underground Railroad. Teaching should cause students to think about the right things because ‘memory is the residue of thought’.
Now let me outline a much more subtle example. Students given a maths problem to solve and with no immediate solution strategies to use will engage in means-end analysis. This is a generic problem-solving strategy that we all possess and it involves measuring your current state, evaluating how far you are from the solution state and then deciding which moves may get you closer. This is cognitively demanding to the extent that even if you manage to solve the problem, you might not recall the solution method. You might not actually learn anything from the process; one reason why worked examples beat solving problems in classic worked example studies.
So my advice to students is, “Never spend more than five minutes trying to solve a homework problem.” Five minutes is actually quite a long time. If students don’t recognise a solution pathway quickly then they are likely to start applying means-end analysis. Rather than engaging in worthwhile practice they will learn little and would be better served by moving on to other questions.
“Put a circle around the question,” I explain, “and raise it with me in class or come and see me. I am your teacher and it’s my job to explain how to do it.” I always ask for homework issues at the start of each lesson and spend some time going over them.
Yes, students will have to persist with some non-obvious exam questions but the best way to prepare them for this is to practise lots of different strategies and not to ask them to practise persisting.
No doubt some of you will be thinking that this is all just obvious; that you don’t need cognitive load theory to work this out; that good teachers have always known about these strategies. Maybe they have. I did not.
Edited 19th April 2018