A response to Aaron Sloman

A response to: Aaron Sloman, Is education research a form of alchemy?

David Wells

Aaron Sloman after a very distinguished fifty-year career is currently the still active Honorary Professor of Artificial Intelligence and Cognitive Science at the University of Birmingham. His conclusions in this article, as he admits,

“are not the theories of a specialist researcher in education or educational technology”

but are strictly from his own AICS perspective, plus his experience of teaching programming and AI and related subjects to undergraduate students.

My perspective is essentially that of a primary and secondary mathematics teacher who has long taken an interest in educational research, so I was immediately suspicious of his title, which he attempts to justify like this:

“Alchemists did masses of data collection, seeking correlations. In the process they learnt a great many useful facts – but lacked deep explanations. Searching for correlations can produce results of limited significance when studying processes with an underlying basis of mechanisms with astronomical generative power.”

This reference to mechanisms is natural, given his background (though there are workers in AI who might object to it). Unfortunately, however, this foundational metaphor – going back to the start of the scientific revolution – the corpuscular theory of matter was extremely mechanical – has never managed to incorporate into the scientific study of man and society any remotely satisfactory explanation of the emotions.

Since emotions are an essential feature of teaching and learning – the notorious phenomenon of maths anxiety is just one disturbing example – and since emotions are especially relevant to problems of motivation [Wells 2008] which lie at one focus, as it were of education, (the other focus being cognitive,) it is not clear how Professor Sloman can believe that this mechanical science can make sense of the problems of education.

He then notes that,

“this correlation-seeking approach characterises much educational research”.

It does indeed because no one has adequate ‘deep explanations’, as Sloman puts it, of how human beings learn, especially in such complex environments as schools and colleges. It does not follow that educational researchers do nothing but collect correlations – that would be absurd: they observe apparent connections, of course, with or without the use of statistics, but then try to make sense of them, to explain them, to test their explanations, and so on, like any scientist.

Professor Sloman feels that educational researchers should be able to do better and he refers to the example of chemistry:

“Accelerated progress in chemistry came from developing a deep explanatory theory about the hidden structure of matter and the processes such structure could support (atoms, subatomic particles, valence, constraints on chemical reactions, etc.).”

Yes, it did, but this is a weak analogy. Human beings are incredibly complex and individually different, as Sloman himself notes, whereas every electron is remarkably like every other electron. He perseveres, however:

“Thus deep research requires (among other things) the ability to invent powerful explanatory mechanisms, often referring to unobservables.”

Ah! By deliberately introducing the qualification deep, which is both vague and deeply metaphorical, in this context, he effectively defines deep in terms of “the ability to invent powerful explanatory mechanisms, often referring to unobservables”. The reader, however, is under no obligation to accept this use of the term deep, which is rhetorical at best and at worst, sophistical. Prof Sloman, however, immediately uses it to damn educational researchers, and others, confirming the expectation that in his opinion the answer to the title question is, “Yes”:

“My experience of researchers in education, psychology, social science and similar fields is that the vast majority of the ones I have encountered have had no experience of building, testing, and debugging, deep explanatory models of any working system.”

The inclusion of psychology might be thought dubious, but he is correct about education, sociology and anthropology because they are studying human beings and groups of human beings interacting, for which phenomena no effective “deep explanatory models” have ever been constructed – not by me or you, and not by Sloman, who now comes to his first punchline:

“So their education does not equip them for a scientific study of education, a process that depends crucially on the operations of the most sophisticated information processing engines on the planet, many important features of which are still unknown.”

This is a remarkable claim. He is identifying the scientific study of anything at all, with “the experience of building, testing, and debugging, deep explanatory models”. [Our emphasis]

Having already defined deep to fit his own argument, he is now defining scientific study with the same goal, with the clear implication that, “researchers in education, psychology, social science and similar fields” are not and never have been genuine scientists, but something else – by implication, something less reputable.

Those researchers are, of course, likely to strongly disagree. One line of defence is that since these social sciences are extraordinarily complex and since the hard sciences took many centuries to reach a “mature” state, whereas the social sciences in their modern forms can be traced back a couple of centuries at most, it is no surprise that social scientists – struggling against this extraordinary complexity, and testing their explanations and models as best they can – have nowhere reached the success of the much older hard sciences.

To damn them, however, for that reason is to fatally confuse the definition of science in terms of method with science understood in terms of its products. In terms of method, social scientists study their experiences, create hypotheses and models in an attempt to explain them, relate these back to experience via observation in order to test them, and modify them if possible to be more effective – and so on. In other words, they behave exactly as the hardest of hard sciences – with the difference that their subject matter is incomparably more complicated – as Professor Sloman, ironically, at once highlights:

“Learning and teaching are a bit like life. As every gardener knows, almost every generalisation about life is false, because there are so many forms of life, showing enormous variation.”

Indeed they are. He now proceeds to emphasise, again, what every teacher, educator, educational researcher, and social sciences knows – that the social world of human beings is ultra-complex.

“There may not be quite so many forms of learning (and failure to learn) but the underlying information processing systems have great internal complexity and diversity (even within one individual, between infancy and grand-parenthood) and they can interact in very different ways with learning materials, the everyday environment, laboratory equipment, peers, teachers, books, computers, the internet, family, physical disability, etc..”

Just so, though the reference to “the underlying information processing systems” points to his own underlying AI perspective. He next suggests that most generalisations about learning require heavy qualification, a point on which all teachers can readily agree. This state of happy convergence, however, is then disturbed by his next claim:

“A deep understanding of the computational mechanisms of both reading and learning to read might provide a basis for teaching methods that are more sensitive to the diversity of child minds, home environments, brain development, and potential.” [Sloman’s emphasis]

Aha! Many people do indeed believe that human behaviour in all its vast complexity can be reduced, ultimately, to computational mechanisms – but many do not share this belief – and belief is the correct term. Moreover, as already noted, this belief in mechanism has so far totally failed to explain the affective half of human nature so anyone unmoved by Sloman’s own belief need not feel embarrassed. Sloman naturally draws his own conclusion:

“An implication is that we still need deep research into mechanisms of learning … and research into the diversity of processes that can occur when those mechanisms interact with a wide range of learning environments, tasks, external and internal motivations, and social contexts.” [Our emphasis]

No, that implication only follows, at best, if you add to Sloman’s previous point the claim I have just sketched, that such a reduction to computational mechanisms is both possible and sufficient. Sloman then once again correctly emphasises the variety of different individuals, and introduces his second punchline:

“We need a powerful generative theory to provide the context for data collection. (Compare research in chemistry.) Data collection without a powerful theory can occasionally produce gold, but mostly it’s like shooting in the dark, and any correlations found will have unknown scope. Moreover, they will tell you nothing about what might have been achieved by trying a different technique you haven’t thought of.”

This is strange: why the emphasis on “data collection”? Collecting data is easy compared to interpreting data, but the latter cannot be as difficult as Sloman claims or teachers in the classroom would never learn by experience and would never improve. Of course, “powerful theories” help, but if progress were impossible without them, no nascent science would ever get off the ground.

“When I was teaching (first at Sussex University, then at Birmingham), along with a bunch of high-powered and highly creative colleagues with diverse talents, we discovered that for a range of students, learning about programming, artificial intelligence (AI), cognitive science, logic, and linguistics, it was possible to develop interactive learning environments that allowed different students to drive the systems in different directions at different speeds with lots of readily available support (documentation, libraries, language extensions, suggested problems) for a wide range of learning trajectories.” [My emphasis]

Stranger and stranger! Suddenly he appears to be describing a classical “progressive” style of education, but in the context of older students and a particular subject area.

“So, as in kindergarten, the students learnt different things, but many of them also learnt how to learn. They learnt at very different speeds — some with an initial long plateau before taking off, others flying from the start, and a small subset for whom the approach did not work at all, even with hours of individual tuition.”

Kindergarten“! So Sloman himself does think of his experience as analogous to progressive kindergarten learning – but where did those ideas of progressive learning originate? They came from the minds and experiences of teachers who had no conception of “computational mechanisms“, and “no experience of building, testing, and debugging, deep explanatory models of any working system“, and whose, “education [did] not equip them for a scientific study of education” – so how did they successfully develop their theories of education which Professor Sloman has rediscovered today?

Sloman certainly does possess, “experience of building, testing, and debugging, deep explanatory models” – but the pioneers of progressive education entirely lacked his qualifications – so how did they do it ?

The answer is very simple. The pioneers studied their pupils, tried to explain their experiences in the classroom, observed what seemed to work and what seemed not to work, developed and modified their methods, abandoning some and introducing others, over periods of many years.

They thus discovered that they could “develop interactive learning environments that allowed different students to [move and develop in] different directions at different speeds with lots of readily available support [concrete materials, manipulatives such as Cuissenaire rods or Dienes’ apparatus, physical objects and the pupils’ environmentS – in the plural – and written materials, of course] [and] suggested problems) for a wide range of learning trajectories“, to only slightly modify Sloman’s own account.

At this point I shall leap ahead to this paragraph by Sloman:

“Our teaching/learning environments were based on a partial theory about the processes of learning — partly inspired by progress and problems in AI.
The theory presupposed that the process of learning is in some cases partly like the process of programming, including similar stages of implementing ideas, testing them, debugging them, and extending them, most of which learners do unconsciously.”

But this model of programming is simply a version of the scientific method in the context of programming as an activity – and the idea that small children when learning behave like young scientists in many respects is – I was about to say “as old as the hills.” – but I will change that to “old hat”.

“This is very different from a theory that learners develop networks of associations, for example.”

Of course it is, but progressive teachers and educationalists never supposed that that was how you learned anyway. The original associationists were, of course, from the 18th century, and their modern descendants, the behaviourists, never flourished much in the UK. (And when I was training to be a teacher 50 years ago, they only featured in passing, if at all.)

Sloman concludes that, “A much deeper theory is required“, and that,

“teachers also need help acquiring a deep understanding of the diversity of information-processing systems found in different learners and how to accommodate that diversity in their teaching.”

But competent teachers already know this – though they may well baulk at the expression “information processing systems” – and they will certainly want to add that other system which Professor Sloman omits: the affective system (however affective is interpreted) which is so intimately linked to motivation and to aesthetic phenomena.

“A major challenge is that this is all totally alien to a world with fixed syllabus structures, numerical marking schemes, and league tables.”

I could not agree more! At this point I shall break off. I entirely appreciate Professor Sloman’s claim, in effect, that students learn best in rich learning environments, a claim that progressive teachers – among whom I am glad to number myself – have always taken for granted, based not least on our own experience, whatever the age of the students and whatever the subject.

What I do not appreciate is the claim that this conclusion can only be drawn by researchers who follow Professor Sloman’s conception of science and scientific activity. The historical truth is that this conclusion was reached by a humble teachers with no expertise in scientific method and on the basis of no deep foundational theories – but simply on their own study of children. For that they deserve everlasting credit. And they were not alchemists.


Wells, D.G., (2008), What’s the Point? Motivation and the Mathematics Crisis, Rain Press.