Cognitive load theory (CLT) took its time in becoming, in Professor Dylan Wiliam’s words, the “single most important thing for teachers to know”. Focused on working-memory capacity and how we learn, and originating from the work of Professor John Sweller, the theory emerged in the 1980s but tumbled quietly around academic circles for decades, with few schools beyond those involved in field trials getting even a glimpse of it.

Then, around five years ago, education began its very own reformation. No longer were the old idioms taken for granted. New authorities took centre stage to rescue the superstitious and ignorant from their archaic fumblings. An evangelistic fervour grew up around research based on randomised controlled trials (RCT) , and teacher-converts hoarded cognitive psychology papers with the zeal of the faithful.

CLT’s moment had finally arrived. Following Wiliam’s 2017 tweet, mentioned above - “I’ve come to the conclusion Sweller’s cognitive load theory is the single most important thing for teachers to know” - the rise of CLT has been rapid.

Those already conversant in Sweller’s work soon became high priests to the masses, spreading the word from the hilltops of Twitter. Commentaries on the scriptures proliferated. Policymakers, inevitably, became interested, too: in the past 12 months, CLT has received a reverential mention in the summary of the research underpinning Ofsted’s new draft inspection framework, and CLT’s fingerprints were firmly on the new Department for Education Early Career Framework.

Suddenly, teachers are being exhorted to have the “cognitive load” of their students at the forefront of their minds as they minister to their youthful congregations.

But a few vocal sceptics preach abstinence. Discomfited by this new trend - and failing to see its significance for their practice - they cast CLT as pointless, inaccurate, misleading, a fad and a dangerous subversive force.

Amid all this, I was an ill-informed CLT agnostic until Tes set me a task: read all I could read on it, speak to experts in the field, try it out. And that is exactly what I did.

So, am I a convert? Do I think that CLT is the “single most important” thing a teacher can know? Or, as one roguish commentator put it, is CLT simply “brain gym” for trads?

At the considerable risk of over-simplification, here’s an in-a-nutshell overview of CLT (don’t worry, some nuance will come later): CLT reminds us of the central significance of working memory in the learning process; it emphasises working memory’s limited capacity; it instructs us that if we overload working memory by trying to hold more in mind than we have capacity for, some information will fall out; and, if that happens, this information cannot make the crucial journey into the infinite recesses of long-term memory, where it can dwell eternally. So, as teachers, we need to bear in mind the “load” on working memory when we design our learning activities.

Sweller has a particular, and increasingly familiar, view of what constitutes learning. “If nothing has changed in long-term memory, nothing has been learned,” states a 2006 paper that he co-wrote (bit.ly/SwellerClark). In Tes, meanwhile, Sweller wrote that “the major goal of education” is the “accumulation of information in long-term memory” (bit.ly/SwellerMemory).

In CLT, learning is defined primarily in terms of information accumulation and schemata construction - that is, highly connected networks of information in long-term memory.

Assuming for a moment that we accept this definition, how can we best facilitate this?

Breaking down, building up

Various factors, we hear, exert “load” on the cognitive capabilities of pupils. The greater the level of inherent difficulty in a task, the greater the “intrinsic load”, in CLT-speak.

In addition, the load exerted by a given task is also influenced by “extraneous” factors: the way in which a task is presented, for example, which may be more or less distracting from the matter in hand, depending on clarity of presentation or the inclusion of irrelevant or tangential material. A student’s prior level of expertise will also interact with cognitive load: the greater the prior expertise on a given subject, the smaller the load exerted in addressing related tasks, since information in long-term memory can be drawn upon to support the new information in the buffer of working memory.

Great, you might think, but how does that help me in the classroom?

Sweller and others explain that to decrease intrinsic load, we should break down subject content, sequencing it in a structured way so that we build up complex ideas out of simpler ones. To decrease extraneous load, meanwhile, we should carefully plan our presentational strategies, ensuring that our instructions are clear, that irrelevant material is excluded from our task design, and that distractions are minimised. Resources should be designed with maximum simplification and integration of materials in mind.

Indeed, a range of pedagogical proposals emerge from under the broad banner of CLT. For example, CLT has become associated, among Sweller acolytes, with “spaced repetition” and “retrieval practice” - important buzzwords in Ofsted’s guidance and well-evidenced practices in the learning sciences. The notion of worked examples is prominent, and explicit and direct instruction are frequently mentioned in relation to CLT. Meanwhile, Sweller has written in Tes that “there are many other instructional procedures that flow from cognitive load theory”, while making it very clear that child-led discovery approaches were not among them.

The literature on the subject is sprawling. As I wrestled with the nuances of it, brow furrowed, reading all Sweller and others had produced, I began to think of my mental sufferings in terms of excessive intrinsic load being placed on my meagre working memory capacity, and of the inadequacy of my long-term memory bank of relevant schemata… “They’ve got me,” I exclaimed, “I’m seeing my own conscious learning experiences through the lens of CLT!”

And yet, as a broad understanding emerged from the fog, I was left cold by what I found. I couldn’t get excited about the implications of this approach being given such a central position. The most important thing I could know? I hoped not.

Serious, workmanlike and worthy, yes, but it seemed to me that something was missing from the formula. Colour? Challenge? Scope for creative digression? Purpose? Space for interpersonal interjections? Was I really supposed to stop my (entirely extraneous) attempts at humour? I baulked. Be fair to the kids. They need my dad-jokes. I’ve done an RCT.

I began to get a whiff of a DfE-endorsed set of scripted lesson plans in the offing (designed in an off-site facility for maximum schemata-delivery efficiency). After all, why risk teachers going extraneously “off piste” and damaging the prescribed learnings with “redundant” exuberances that don’t contribute directly to the programme?

I must be missing something, I thought. In a spirit of openness, I read wider, sought out new sources, spoke to academics in related fields of research.

The broad translation of the theory into instructional technique is not without criticism. Nilli Lavie, professor of brain sciences at UCL, and her fellow researcher Michael Hobbiss are sceptical about whether the variety of teaching prescriptions advocated via CLT has anything to do with the theory as broadly understood.

In fact, Lavie goes further, and suggests that CLT might not technically be a “theory” at all. For her, CLT discourse is too generalised and lacking in scientific specificity to merit the label “theory”. Not only that, she says that CLT tends to conflate cognitive processing and perception, which are related but separate.

Perception of stimuli precedes cognitive processing, she explains: her work on perceptual load theory shows that there is significantly more to supporting task-specific focus and processing than simply ensuring adequate space in working memory (see bit.ly/PerceptualLoad).

Firstly, a person’s perceptual field is filled up with sensory stimuli. Only then do the cognitive processing faculties get to work in selecting and making sense of the data received. Task-relevant stimuli are processed before task-irrelevant stimuli, though all attentional resources are being used at any one time. So, if only a small amount of task-relevant stimuli is presented to a person, the rest of their attentional resources will be spent on task-irrelevant stimuli. In other words, distractions.

Taking this into account, Lavie argues that some of the pedagogical implications frequently drawn from the broad body of CLT evidence are counterproductive. For example, the characteristic insistence that reducing load by simplification is inherently good for learning sits in tension with her research on distractibility. Lavie’s studies with Dr Sophie Forster, of the University of Sussex, show that learners will be more distracted if the presentation of relevant material is over-simplified or under-stimulating, since a learner’s perceptual field will be automatically filled with other stimuli and taken away from the intended focus (see bit.ly/ForsterStudies).

Not a scientific hypothesis?

People diagnosed with ADHD, in particular, need a high perceptual load in order to process task-relevant information. Increasing relevant perceptual load - for example, by employing dual coding - decreases distraction and supports focus. It engages more of a student’s perceptual capacity, causing a kind of blindness to possible distractions, and ensuring that a higher proportion of relevant material filters its way into working memory processes.

So, a higher perceptual processing load leads to improved attention. This nuanced but essential message tends to get lost in the simplifying trends of CLT as popularised.

But Lavie’s critique goes even further. She argues that CLT does not represent a recognisable scientific hypothesis as far as the scientific fields of cognition or psychology are concerned - “cognitive load” itself, she argues, is not an accurately measurable phenomenon. There are no units, nor is there a scientific way of telling whether a person’s load has been optimised, exceeded or under-utilised. Nor is it associated with particular, observable brain processes that can be manipulated experimentally.

As Schnotz and Kurschner have shown, studies that have attempted to measure cognitive load rely on self-reporting, physiological measures or performance measures that try to establish overall cognitive load, and do not differentiate between intrinsic, extraneous and germane load (bit.ly/Schnotz).

And yet, it is precisely the balancing of these purportedly different loads that holds the key to successful learning, according to CLT. So if there is no agreed way of knowing what level any of the loads are at, how would we know which to adjust, even if we had a clear mechanism for doing so? (It should be noted that Sweller has actually removed “germane load” as an independent source of cognitive load, but it occurs elsewhere - see “Element interactivity and intrinsic, extraneous, and germane cognitive load”, Sweller, 2010 - and, regardless, the point still remains).

Lavie wonders what the “theory” adds scientifically to Alan Baddeley and Graham Hitch’s view that working memory has limited capacity. Does CLT, in fact, simply amount to a recommendation for teachers to ensure that their teaching doesn’t “go over pupils’ heads”, and that tasks should be neither too difficult, nor too easy for the pupils, without providing a specific mechanism for knowing when “load” is in the sweet spot?

Duncan Astle, programme leader at the Cognition and Brain Sciences Unit and a fellow of Robinson College at the University of Cambridge, prefers to think of CLT as, at best, an important and necessary “health warning” about working memory capacity prior to further considerations about what constitutes good learning.

Guiding star

It’s certainly the case that the limited capacity of working memory is important to bear in mind when contemplating instructional design (or “teaching”, as some of us call it) - and this is especially important when dealing with students with weak working memories. But ensuring that working memory is not overloaded does not ensure that good learning will take place. It is not, on its own, a philosophy of, or a handbook for, effective learning. All sorts of other factors come into play.

Astle cites a couple of examples. Research shows that pupils who are in a heightened state of curiosity are more likely to retain mundane information, while “prediction error” studies have shown that learning can be dramatically improved by delivering information that proceeds to confound the learner’s expectations.

For example, you would be very much more likely to remember that Charles Darwin failed to complete his medical degree than that his middle name was Robert. The first piece of information is memorable because it jars with the representation of Darwin in our “semantic memory”. The latter is not memorable because it is unremarkable. Not all information is equal.

So, as a teacher and as someone in a role leading teaching and learning in a school, what is my conclusion on CLT?

No doubt supporters of CLT will protest that they are not advocating it as the only relevant consideration in learning. Sweller himself has written that “it is not a theory of everything” (bit.ly/SwellerCase).

I do think Sweller et al’s findings have a place in the wider debate about the science of cognition. A range of lines of research enquiry are loosely bunched under the banner of CLT.

They constitute an important part of research in cognitive science, even if the field’s boundaries are somewhat blurred.

Extrapolating teaching practices from CLT is a murky business, though. Teachers are not in controllable experimental situations; they cannot see, nor measure the load currently being exerted on any given pupil any more that they can see who is genuinely listening and who is pretending to.

If a student fails to complete a task successfully, it may well be that their working memory was over-stretched at the point of instruction or it may be that their perceptual load was under-filled. Alternatively, they may have decided, consciously or unconsciously, not to engage with the “instruction”. This may be because they weren’t interested in the subject matter or because they didn’t like the teacher; it may be because their academic self-esteem was too low to make a serious attempt to pay attention, or they may have a naturally high level of distractibility. It may be that the format of the specification thrust upon them was just so deadeningly exam-focused and irrelevant to their lives that it was impossible for them to muster an ounce of engagement.

Teachers tend to be rather good at seeing the pupils in front of them as individuals with rich complexes of motivations, capabilities, hang-ups, interests and prejudices. Through relationships, we can begin to make “best guesses” at what might be driving a particular behaviour. Sure, CLT may provide one arm of what we should consider, but it is one of many - and one that teachers, somewhat intuitively, were considering already. In short, I am not convinced CLT adds much.

What I fear, though, is that it may take away so much that is important. Regardless of how robust you believe Sweller’s science to be, there does seem to be a case for warning against a kind of CLT hegemony in edu-discourse. Intentionally or not, the practical implications of CLT for the impartation of information can be used to fuel a particular, rather narrow, narrative about the purposes of education. The comments of CLT advocates and of Sweller more often than not reinforce this and the elevation of CLT in policy narratives give weight to a particular educational perspective. If we first accept that learning is best defined as a change in long-term memory, it is a short hop to concluding that teaching is aimed at filling up empty heads with content. And a short hop from there to designing a content-heavy curriculum, which is bolstered all around with retrieval practice or, as pupils know them, tests and exams.

Now, even “progressives” are in favour of some testing. It is clearly important. Knowledge, too. And we should be wary of false dichotomies. But there is reason to fear that those who emphasise knowledge strongly, as though the skills brigade weren’t interested in it, will seize upon CLT as further evidence that children should sit down, be quiet and listen to the script. The more it becomes a guiding star for Ofsted inspectors - who, let’s face it, will only have a limited grasp of its nuances - the more likely it is that curricula and pedagogy will tilt towards its perceived edicts to please the judges. Schemata Max. The EBacc. The almost ubiquitous “knowledge-rich” agenda. CLT is used to prop it all up.

From there, teacher agency is eroded, the life of lessons is depleted. There are those - myself included - who think that curricula should be not only about knowledge-retention but also about socio-emotional development; not only about grades but also about service learning; not only about efficiency of depositing information but also about character development and values. None of these are mutually exclusive; it is all a matter of balance. But CLT is being used to tip the scales one way with an influence it simply has no right to exert. And education looks all the worse for it.

Alistair McConville is director of learning and innovation at Bedales School in Hampshire

This article originally appeared in the 26 April 2019 issue under the headline “The cult of CLT”?