On the clever but troublesome discourse of Cognitive Load Theory


I’ve previously written about clever discourse. It is a type of discourse that generates plausible ideas purportedly rooted in evidence. Such ideas seem unassailable,  common-sense facts from reputable sources; however, further scrutiny suggests otherwise.

A classic example of clever discourse is the current government’s claim to have increased the number of children in good and outstanding schools. It is an unassailable fact, there are more children in those schools. Equally true, there are more children in the school system and the government made it harder for Ofsted to downgrade outstanding schools.  The government has not improved the education system; rather, it has engaged in the artful use of clever discourse.

I tend to think clever discourse also occurs, because of unconscious conceptual drift,  as ideas are transposed from one field to another. I was minded of this, whilst reading Greg Ashman‘s blog on the issue of executive function.  Greg writes powerfully and informatively about such issues, but I think in this case he is wrong to dismiss executive function.

I worry about the consequence of speculative theories in one field being used to make strong inferences in another.

Baddeley and Hitch’s working memory model

Influenced by academics, working in the field of instructional design on Cognitive Load Theory (CLT), Ashman takes issue with Baddeley and Hitch’s working memory model. In simple terms, the model seeks to explain how a multi-component system processes perceptual and conceptual information and then stores it (see below).

Plancher and Barrouillet sum up Baddeley and Hitch’s contribution to cognitive psychology:

Thus, the way that Baddeley and Hitch interpreted their results has had a robust and durable impact on cognitive psychology, by introducing a separation between the WM functions of processing and storage.

Baddeley suggests the model is a useful metaphor and, on that basis, it is widely accepted.

baddeley working memory model.jpg

Figure 1: Baddeley and Hitch’s Working Memory Model

Ashman disagrees, citing Cognitive Load Theory:

“In Cognitive Load Theory, John Sweller, Paul Ayres and Slava Kalyuga argue that the popular model of working memory developed by Baddeley and Hitch has a fundamental flaw. This model proposes a ‘central executive’ that directs individual attention – a kind of decision-making module. And yet the entire mind is a kind of decision making module and so Sweller et al. argue that this is a bit like suggesting the mind is controlled by a mini-mind. What controls the mini-mind? Is there an even mini-er mind inside the mini-mind? And so on.

It is a common enough paradox, familiar to anyone who has ever contemplated the question of who created God. We have introduced, oddly enough, a deus ex machina to miraculously right the narrative at the eleventh hour.”

According to Greg, the CE represents a flaw in the model. I disagree,  the model needs a central executive, which is why it was posited by Baddeley and Hitch. Remember, Baddeley and Hitch, are concerned with both perceptual and conceptual information not just the latter. If working memory is component-based, and works as Baddeley and Hitch suggest, then something needs to be able to distinguish between different types of perceptual information, how that information is processed and finally stored. In effect, no central executive: no model.

It is important to note I do not argue for a discrete central executive; rather, that a central executive is a metaphor to think about executive processes that are likely distributed throughout cognition.

An alternative to Baddeley and Hitch: Sweller et al

Ashman offers an alternative to Baddeley and Hitch arguing that “knowledge is what we think with”:

“Instead, Sweller et al. argue that it is schema held in long-term memory that act to direct and control the action of working memory. We are what we know. Knowledge is what we think with.”

Surely, the executive function issue centres on how we think; not what we think? What are the processes that lead us to consider different propositions, challenge our own assumptions and make new connections?

As previously observed, the desire to make memorised knowledge the central issue of cognition leads to the constant conflation of how and what we think.

More problematically, how do we process new information? Sweller et al suggest that long-term memory “directs and controls”, but by definition, new information does not have associated schema in long term memory (LTM).

Sweller acknowledges the problem:

“Not only do schemas act as a central executive, they are the only conceivable central executive. If schemas are not available, as occurs when dealing with new information there is no alternative central executive to call upon. As previously indicated when relevant information in long term memory is not available random generation along with effectiveness testing is the only remaining alternative. Contrary to theories such as Baddeley (1992) there is no logical manner in which a central executive other than a learned central executive can function.”

The suggested alternative to Baddeley and Hitch begs lots of questions not least the reference to a learned central executive and assertions like these: 

“Knowledge in long-term memory provides a central executive that indicates (1) what should be attended to, (2) how information should be processed and (3) what actions should be taken. Failing knowledge, a viable central executive is not available.”

“Knowledge does more than provide a central executive to organize information. Because information is organized by knowledge, there is no need for working memory limits and the associated problem of randomly organizing information and testing a particular organization for effectiveness.”

I have no idea how “knowledge in long-term memory provides a central executive” or “how information is organised by knowledge”. Like Ashman, Sweller conflates how and what we think. 

Delivering the “coup de gras” of clever discourse

Ashman then sets about constructing the coup de gras of clever discourse,  by doing what Sweller at al don’t; namely, killing off executive function all together by suggesting you “cannot train something that does not exist”:

“This resolves the question as to why we cannot directly improve executive function in any general way, either by working memory training or by attempts to teach thinking skills. We cannot train something that does not exist.”

Executive function, like the DODO, is now dead, but Greg has a new target in his sights: thinking skills. His train of thought is clear: no central executive, no executive function. therefore, no thinking skills.

Ashman drives his point home describing the CE as inert and paradoxical:

“Much of this is speculation, but until there is evidence from randomised controlled trials that types of thinking such as decision-making or problem-solving can be improved in a general and transferable way, models of the mind probably do not need to posit an inert and paradoxical central executive. The machine does not need its ghost.”

The Dodo has returned as a ghost and found a place haunting Baddeley and Hitch’s working memory model. And yet, whilst he has managed to do away with a discrete CE, Ashman hasn’t cited any evidence to substantiate the assertion that executive function does not exist. 


I agree with Greg when he says this is all very speculative.  The cleverness of his argument is in constructing a logical statement that is inherently true. You could walk into any bar in the country and declare “we cannot train something that does not exist”, and even the most argumentative of post-modernist drunks, propping up the bar, would have to agree.  Except, “the something” does exist.

And the point is this,  just like a faulty “sat nav” if something starts off in the wrong direction there is no telling where it is going to end up. Look at Ofsted’s definition of learning: (…) an alteration in long term memory. If nothing has altered in long-term memory, nothing has been learned and ask yourself; where did it come from? And the answer? A clever discourse that over time disappears the CE, then executive function and finally thinking skills; leaving only long-term memories.

6 thoughts on “On the clever but troublesome discourse of Cognitive Load Theory

  1. This article is an example of the thing it condemns. You have misrepresented Greg (Swellers quotes are real though) and defeated a phantom argument. Greg’s point (and Swellers) is that there is no need for a central executive as schemas fulfil the role. This may or may not be true. Finally I am confused by your use of metaphor and why it is relavent.

  2. I was an undergrad reading psychology when Baddeley & Hitch proposed their model of working memory. It caused a stir at the time because it was a significant step forward from Atkinson & Shiffrin’s.

    I can’t recall anyone assuming that, to paraphrase Charles Spearman, “there really exists a something that we may provisionally term a central executive”. It was clear from Baddeley & Hitch’s model that the central executive was a necessary *function* of the information processing system; whether there was a dedicated biological mechanism for it was unknown.

    For Baddeley & Hitch and most of their readers, the central executive was as Peter points out, a metaphor. Psychologists were well aware of the infinite regress problem for central executives, consciousness etc. I’ve seen no evidence of Baddeley & Hitch proposing that the central executive is any sort of ‘module’ as Greg claims. In their 1974 paper they avoid speculating beyond what their data told them. It’s a good example to follow.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s