Is cognitive load theory just like brain gym?

A response to Guy Claxton


I have come to the view that attacks on cognitive load theory tend to have the perverse effect of increasing the popularity of the theory. This must mean that people do not find these attacks convincing, but I think the attacks also bring the theory to people who otherwise would not have heard of it.

Recently, Professor Guy Claxton of Building Learning Power fame has launched a string of attempted refutations of cognitive load theory, spiced with his own particularly distasteful brand of superciliousness. The most recent attack is in a blog post, although you can read about previous versions of the same argument here and here. Although perhaps not strictly necessary, I think that pointing out the flaws in this latest attack is a worthwhile exercise.

Those of you who are moderately familiar with cognitive load theory can probably spot these flaws yourself. If so, stop reading this post and give it a go. However, if you are new to the theory, perhaps I can take this opportunity to highlight, in relief, some interesting aspects of it.

Claxton suggests cognitive load theory is the next Brain Gym, which is quite amusing given that Brain Gym and Building Learning Power were fads that were popular in English education at around the same time. He also labels it ‘Cognitive Load of Baloney Theory’. Perhaps his intention is ironic or comedic. It’s hard to tell.

However, at the heart of the post are some specific claims that I will address.

“1. It is based on a vastly oversimplified and antiquated notion of “working memory” (WM) that was current in psychology in the 1970s. Since then, conceptions of WM have moved on in major ways - not least in the recognition that incoming information does not have to pass through a constricted processing bottleneck on its way to Long-Term Memory (LTM). Prof Alan Baddeley FRS is the godfather of WM. Read his definitive book Memory for an up-to-date review. The computer metaphor on which the original concept of WM was based is no longer widely accepted as an accurate or adequate depiction of human cognition. Brain-based theories, in which there are no separate memory “stores” - no boxes in the head - underpin much current research, and they do not lead to or justify anything like John Sweller’s image of Cognitive Load.”

Yeah. Nah. If anything, we now think working memory is even more constrained than we used to, with Cowan (2001) suggesting we can only process about four items at a time. I’m not sure why Baddeley and computers are brought into it because the most recent formulation of cognitive load theory explicitly rejects Baddeley’s model. Rather than suggesting the mind is like a computer, it suggests the mind processes information in a manner analogous to the process of evolution and therefore does not need a central processing unit - a key feature of computers. And I am not aware of cognitive load theory asserting there are any ‘boxes in the head’.

A common motif of Claxton’s writing is to assert the authority of various experts in this way. What criticisms of cognitive load theory would a reading of Baddeley’s book, Memory, suggest? It’s not clear. If Claxton knows, he’s not telling us.

“2. CLT implied that the main reasons why students sometimes struggle or fail at learning are cognitive and “structural”. It implicitly encourages teachers to ignore all the other reasons, such as: Students are preoccupied with other worries; They don’t like the teacher and are reluctant to engage; They intuitively judge that their self-esteem will be better protected by not-trying than trying; They find the topic mystifying, boring or of no conceivable relevance to their lives; Their teacher misjudged their prior knowledge and didn’t pitch the lesson right; Their teacher has so oversimplified and watered down the content that there is nothing to hold their attention.”

Yeah. Nah. Cognitive load theory also does not have much to say about a child who needs glasses and who cannot see the board from the back of the class. That does not mean the specific claims it makes about the effect of cognitive load on learning should be rejected. In a sense, many of the factors mentioned in this list could be considered sources of ‘extrinsic load’ within the theory, but that would not be particularly helpful. Cognitive load theory makes no claims to be the best way of viewing every conceivable factor that influences learning. The clue is in the name.

It is telling that Claxton has to assert that cognitive load theory ‘implied’ certain things. I suspect this means he cannot point to a single source where a proponent of cognitive load theory has made such an expansive claim.

“3. CLT relies on a bogus distinction between the inherent difficulty or “intrinsic load” of some subject-matter and the students’ existing knowledge and ability. But there is no way of assessing the intrinsic difficulty of something independently of someone who is struggling with it. The idea of CL is circular. Why are they struggling? Because it has high intrinsic load. How do you know it has high intrinsic lead? Because some people struggle with it. (Sweller did his PhD on animal learning, a field in which BF Skinner was long preeminent. In his biography Skinner pokes fun at this kind of loose thinking. “This rope won’t break because it has ‘high tensile strength.’” someone claims. “How do you know it has high tensile strength?” “Because it won’t break.”)”

I don’t think Claxton is familiar with ‘element interactivity’. If we make assumptions about prior knowledge - such as that a child is fluent in English but has never encountered a certain science topic before - we can, in some instances, roughly calculate the number of elements inherent in different instructional procedures. These then lead to testable predictions. See, for instance, my own published study.

Element interactivity is controversial and can be critiqued, but this is not that critique.

“4. CLT is often taken to imply that teachers should facilitate learning by making is as easy as possible (so “information” can slide effortlessly through the WM bottleneck into LTM). But things that are easy-peasy are hard to stay focused on (see 2f above). And where, then, are students to develop the intellectual skills and stamina to wrestle productively with things that are genuinely difficult – as they inevitably will have to at university and beyond? If you are going to value the learning that accrues through grappling with complex and meaty challenges, CLT will offer you no practical help in deciding how much challenge a class will be able to bear.”

False. Cognitive load theory suggests that load should be optimised i.e. lower than the ~4 item working memory limit but not much lower than that. Notably, throughout his critique, Claxton makes no reference to the hundreds of replicable randomised controlled trials that cognitive load theory is based upon or the many effects these have generated, including those that intentionally raise load in certain circumstances, such as the expertise-reversal effect, the guidance fading effect or even the completion problem effect.

But, hey, it’s those pesky implications again.

“5. CLT presupposes that, in Sweller’s words, “the major goal of education is the accumulation of information in LTM.” Many, many intelligent people, by no means all of them holding “progressive” values, would beg to differ, and would see Sweller’s claim as unnecessarily narrow and contentious. The bare “accumulation of information” is no more valuable than the accumulation of mountains of food in a refrigerated larder. Information is for using just as food is for eating, and for information to be useful, it has to be digested and assimilated in a way that makes its relevance to meaningful real-life tasks and challenges apparent. Mere accumulation will not do – unless your sole goal in life is to display erudition at dinner parties or win quiz shows. The psychology behind CLT is highly naïve, limited - and limiting.”

I’m not sure where the quote comes from. Sweller may have said this, but he usually talks of the accumulation of knowledge in long-term memory. The use of the term ‘information’, and the pejorative sense Claxton ascribes to it, seems to be the main source of his criticism. Many cognitive scientists tend to view knowledge as organised in long-term memory according to meaning and in the form of interconnected webs known as ‘schemas’. Schema building is what I am trying to achieve when teaching, not least because schemas in long-term memory can overcome the limitations of working memory.

I am sure there are many, many intelligent people who disagree with this objective or think learning is about something else, but unless we know their specific objections, this is just another form of argument from authority. It’s a sophisticated version of, ‘a bloke down the pub said…’

“6. CLT was born out of experiments on mathematical problem-solving. Some studies have explored the use of CLT in other subjects, but Maths is taken to be the prototypical subject, and learning to solve well-defined problems the prototype of all learning. But this is only true if these other subjects are shrunk to fit the Maths prototype. Most teaching and learning in Drama is not like that. Nor in PE. Nor in PSHE. Nor in English Literature. Most subjects, and most of life actually, are not best addressed as if they were like mathematical problem-solving. Ask any counsellor or theatre director or sports coach.”

If this is true and a flaw in cognitive load theory then, presumably, Claxton can point to evidence of the predictions of cognitive load theory failing in these other subject areas. Perhaps there are randomised controlled trials showing inquiry-based learning is more effective for teaching the writing of English Literature essays than explicit teaching? I’ll wait.

For what it’s worth, the cognitive load theory literature does oversample subjects like mathematics and under-sample, say, history. However, every time it has been extended into these areas, the principles seem to hold. I guess our minds don’t change all that much when we go from maths to English. And to quell a common misconception, these studies have not been mainly conducted in ‘labs’ with undergraduates. Like my own study, much of the work has been done in schools.

“7. CLT presupposes and advocates a didactic, “direct instruction” approach to teaching. Time students spend exploring, arguing and grappling is adjudged wasted time if the key points to be retained could simply have been explained, practised and hammered in through testing and repetition. WM gets clogged up unnecessarily. Most teachers know that good teaching is a subtle and shifting blend of explanation and exploration. Sometimes it’s good to let kids struggle and argue; sometimes it isn’t. The craft of teaching is, in part, knowing when and how to let a discussion run and when to cut it short because unproductive. Art and History cannot be reduced to facts and explanations without having the life blood sucked out of them. CLT, in other words, seems hell-bent on making school learning as dull as possible.”

Well, I guess it depends what you mean by ‘direct instruction’. Cognitive load theory posits a gradual release of control from teacher to student. Again, the guidance fading and expertise reversal effects are relevant here. This is consistent with models of direct instruction that view it as a whole system that starts with new concepts being fully explained. At least this has the advantage of providing a rationale for when teachers should move from one kind of teaching to another. What’s Claxton’s?

In my experience, cognitive load theory does not necessarily make learning dull. That depends more on the teacher and their delivery. Students are often pretty motivated by learning new things and classes that are guided by the principles of cognitive load theory can be fast-paced and fun. I think we tend to assume alternatives like problem-based learning will always be more engaging, but they seem just as likely to lead to frustration, boredom and a lack of engagement.

What this part of the argument could really use is some evidence.

“8. And finally, CLT is catnip to fogeys (young and old) who want education to hark back to the “good old days” (ahem) of the grammar school; to middle-class parents and politicians who know that statistically their own children will inevitably rise to the top in such schools; and to simpletons who are unable or unwilling to grapple with the thorny problem of how education could function as a better preparation for life in complex and challenging times – for all children, not just those destined for the traditional professions. CLT is just a fad; it is, as someone said, like Brain Gym for Traditionalists. The sooner the fad passes the better.”

Honestly.