Robots, evolution and why schools shouldn’t worry about innate skills

It should come as little surprise to hear that some of what human beings can do is innate. That is to say, we are born with various capacities and abilities which appear to be ‘hardwired’ into our brains. The evolutionary psychologist David Geary talks about such capacities as being either biologically primary or secondary adaptations. Biologically primary adaptations are those that emerge instinctively by virtue of our evolved cognitive structures, whereas biologically secondary adaptations are exclusively cultural, acquired through formal or informal instruction or training.

Evolution, through natural selection, has resulted in brains that eagerly and rapidly learn the sorts of things which allow us to survive and reproduce. Geary divides these biologically primary domains into three main areas: folk psychology (interest in people), folk biology (interest in living things), and folk physics (interest in inanimate objects), and suggests that we are naturally disposed towards learning such things as peer interaction, play hunting of other species, and exploration of the physical environment within these primary domains.

By contrast, biologically secondary knowledge is abstract, counter-intuitive and hard to learn. We may find it natural to learn to count physical objects, but understanding the abstract concept of negative numbers is hard precisely because they have no physical reality. Similarly, we appear to have evolved a capacity to learn to speak our mother tongue over millennia, but we have developed no such capacity to learn reading or writing. Writing is a comparatively recent ‘good trick’ and natural selection just hasn’t gotten round to equipping us with an in-built capacity to learn to read.

This brings us to the perennial complaint amongst many well-intentioned but sadly uninformed observers that schools don’t spend sufficient time teaching such skills as creativity, problem solving and collaboration. Here’s a recent one:

Prof Rose Luckin, an expert on AI and education at University College London who gave evidence to the committee, said that the school curriculum needs to be brought up to date to reflect that we now live in a world where problem-solving and creativity are becoming more important assets. “Regurgitating knowledge is something that you can automate very easily,” she said. “That doesn’t prepare children for the modern workforce.”

Instead, she said pupils should be spending more time working on problems collaboratively, because in future many professionals will be required to collaborate with robots.

Just as I wouldn’t dream of advising Prof Luckin on how to go about designing AI, I fervently wish she and others like her would be content to stick to their areas of expertise. To say “we now live in a world where problem-solving and creativity are becoming more important assets” is fatuous indeed. When have human beings ever lived in a world where they didn’t need to solve problems and be creative?

The answer is never. Creativity, collaboration and problem solving have always been vital for the survival of the species, so much so that we have evolved an innate capacity for developing these skills. Every child naturally learns to collaborate, solve problems and be creative without recourse to explicit instruction.

Of course, that is not to say that every child is equally creative or that we all share the same capacity for successful collaboration; as with every human characteristic there will be a normal distribution of ability. But it does mean that everyone has a natural ability to solve problems. Otherwise some people would never work out how to get out of bed and put their trousers on! Now, you could argue that because of this difference in natural ability, some children will benefit from additional instruction in these skills. Maybe they would, but not nearly as much as they’d benefit from extra instruction on culturally dependent, biologically secondary knowledge where the differences are almost purely environmental rather than heritable.

Because these skills are innate, schools don’t need to waste much time teaching them, and certainly not as generic, ‘transferable’ skills. It’s probably worth giving a few pointers on how to collaborate more effectively or offer some tips on how to solve specific problems, but these can be acquired incredibly rapidly. But, you may ask, if the ability to collaborate or be creative is innate, why do we seem to see it so rarely? Why, for instance, is it that children seem incapable of working effectively in groups?

The answer is simple: you have to be creative about something. It’s not enough to collaborate, you must have something to collaborate on. It’s all very well to have an innate ability to solve problems, but what specific problems do we want students to solve? In every case the problem is caused by the interaction between the natural and the unnatural; the primary and the secondary. If you want students to find creative ways to avoid doing any work, then clearly creativity is a doddle. If instead you want them to find creative solutions to biologically secondary algebraic equations then you’re in trouble. Asking students to collaborate in discussing last night’s telly will present no problems, but expecting them to collaborate on researching climate change is trickier because researching climate change doesn’t come naturally. The same can be said of learning. Learning is another innate capacity – we are all born with the ability to learn and find it easy for pick up stuff we find interesting. The only question really worth asking is what are students learning?

Where some children appear to struggle to be creative or solve problems the issue is in fact caused by a lack of biologically secondary knowledge. Contrary to Luckin’s assertions, regurgitating knowledge is not something that you can “automate very easily”. In order to ‘regurgitate’ knowledge you’ve first got to know it, and knowing abstractions is not nearly as effortless as many otherwise very smart people seem to believe. It takes a lot of time to teach children all the cultural knowledge they need to make sense of the modern world. So much time in fact that there’s precious little of it to spare for fripperies like contentless creativity or hollow collaboration. After all, curriculum time is strictly finite and there is always an opportunity cost to every decision we make about what to teach.

Rather than investing time on working on problems collaboratively just in case we have to one day collaborate with robots, we’d be much better off developing students’ capacity to collaborate by giving them lots of powerful knowledge which they can then use to solve problems with and collaborate on solutions to problems we don’t yet know exist. Understanding the evolution of how we learn asks some difficult questions about the curriculum we put in front of children. Whilst they will be highly motivated to engage when they have opportunities to socially interact and develop other biologically primary abilities, these are things they would probably learn independently without giving up curriculum time.

31 Responses to Robots, evolution and why schools shouldn’t worry about innate skills

  1. Problem solving and creativity are part of the front end of our (New Zealand) curriculum. I was a mathematics teacher. I think the link that was missing to some extent in the past in mathematics both here and in the UK was that we focused too much on skills practice and not enough on problem solving in mathematics. To solve a problem students need to identify what is required and then select the best tools to solve the problem.

    In my opinion the message should be that once the tools are there it is vital we spend sometime using them to solve problems. But we cannot solve problems until we have the tools in the toolbox.

    • David Didau says:

      Solving mathematical problems is different to generic problem solving. What makes solving maths problems hard is the maths, not the problem solving. If we want students to solve maths problems we’re best off teaching them maths.

      • Pat Jennings says:

        I have taught Maths for decades and completely agree with David. You cannot solve a Mathematics problem without knowing the Mathematics required – and that means having practised routine techniques so that you recognise when and how to apply them.

        • 100% agreed. And I’m a professional mathematician whose life is largely research in the subject (an intrinsically demanding creative-problem-solving activity that emphasizes those skills at a higher level than most advocates of this stuff even imagine!). But you can’t do it – period — without sufficient appropriate domain knowledge. I am no newb at this. I was a champion in math competitions in high school — the book prizes I won from regional competitions and the pins and stickers from national ones are still in my collection 40 years later.

          Today I train Mathletes (competitive mathematics students who must solve problems whose difficulty level would make 99.9% of the population’s minds reel) who take place in inter-university competitions, at the University of Manitoba. I have done this for many years, and trained many champions, many of whom go on to do very interesting things in life. One of the, Craig Kasper, likes to create puzzles. many kinds, but these include crossword puzzles, cryptic crossword puzzles and the highly challenging types that only the true officionados can crack. If you think DOING a cryptic crossword is a challenge, try writing one. Craig submits some of his puzzles to the NYT and on the occasional Saturday you’ll see his byline there. He has been commissioned to write challenging puzzles for the world championship (yeah … there’s actually one of those). And also for the U.S. crossword puzzle team (and … one of those too!). He was also a champion for our teams during his time as a student here – one of many.

          I am the Director for the Manitoba High School Mathematics competition. I interact directly with the top students in this activity and help our National trainers identify the most promising candidates for our International Mathematics Olympiad team. I build this contest from scratch each year with a crew of masterful problem solvers and experienced competitions trainers.

          I know what creativity and problem solving look like. I know what it looks like for a student to develop these skills.

          And it never happens unless they have sufficient domain knowledge. The more they have, and the more deeply ingrained it is — so they hardly even think about the details when they invoke it — the more potential they have. As one of my colleagues here — possibly the most storied high school mathletics trainer in Canadian history — likes to put it, to do well in mathematical problem solving, you need a “Well-Stocked Toolbox”.

          Domain knowledge is probably 70% of making a champion problem solvier. And while there is such a thing as developing these skills through training, I believe David is completely correct here that the fundamental skill of solving problems is innate — you can’t really teach it. What one can do is assist a student in the difficult process of applying that innate skill in the abstract, secondary arena of mathematical problems. It is the matter of transference, or rather focussing, of that skill on something that nature has not trained us for. Like learning to ride a bike: Nature has not taught you how to do that. But you can adjust some of the innate skills (balance, hand-foot-eye coordination, “walking” etc) to this end. That part is learned.

          I think the resources expended on trying to teach these innate skills is probably the most harmful general problem in education today in North America.

          • Grumpywearymathsteacher says:

            Domain knowledge is necessary but not sufficient for mathematical problem solving. People definitely improve at solving maths problems when they have experience of solving lots of maths problems. Conversely, in the UK we often meet students who get the top grades at GCSE or A-level in maths but are poor at problem solving because they have not been exposed to the right kind of task. You could argue that what good problem solvers have is knowledge of strategies as well as content…

          • Grumpywearymathsteacher says:

            No reply button under your comment, DD, so this is in the wrong place, sorry.

            I followed the link and read your older post…brilliant! Thank you.
            (But some people will interpret this post as saying you shouldn’t teach problem solving at all, rather than that you shouldn’t teach it as a separate thing?)

            I guess we don’t think of the problem solving knowledge as domain knowledge because (as far as I know) nobody has explicitly written it in to the maths curriculum. I reckon the majority of secondary maths teachers, and almost all school leaders, are happily unaware of its existence…and/or consider it completely unnecessary (“just teach them the maths”).
            I was unaware of its existence when I left school; at university I discovered I didn’t have it; now, after 14 years in the classroom, I still only have incomplete knowledge and regularly get into trouble for bothering to attempt to teach some small parts of it. This makes me grumpy and weary…
            Thank you again for the link. That’s my CPD for this week 🙂

  2. Thanks for what I think is a response to our conversation yesterday on Twitter.

    I agree with 99% of what you say here – certainly with the need not to waste time addressing decontextualised skills, which, as learning objectives, are empty. If you want to teach people problem-solving skills, give them really tough problems to practice solving. The same goes for collaboration & creativity. I also agree with the criticism of Rose Luckin’s position on knowledge. What is wrong is not that prominent academics continue to state positions with which you or I might disagree, but that they do so while appearing to ignore well rehearsed arguments against these positions. It is the job of the academic to argue their case against all serious challengers – yet too many simply continue to expound their views while ignoring rebuttals. Much of this is connected to the sort of soft relativism which I have been addressing recently on my blog (, in connection to Gert Biesta’s views on educational purpose.

    On the other side, I think the loose group of educational bloggers to which we are both associated in different ways (the neo-traditionalists, perhaps?) could be criticised for being too quickly satisfied by their attacks on easy targets. The ideas of child-centred constructivists may not be easy to dislodge, maybe, but they are relatively easy to attack. I believe that the underlying problem – the difficult problem – faced by education is one of scale. It is not that we do not know what good education looks like – Socrates showed us that 2,000 years ago. But good education that depends on the wise teacher has always been done at small scale, which is one good reason why education has tended to be elitist. I am not sure that the neo-traditionalists’ call for a return to direct instruction, with its focus on the subject knowledge of the individual classroom teacher, has an answer that is sufficient to the challenge of education at scale.

    The point at which we may disagree with regard to transferable skills is more nuanced. I agree with you that abstract skills need concrete context through which to be taught and I also agree with you that skills cannot be disentangled from knowledge: they are two sides of the same coin. But, paradoxically, I think it is your position that is tending to delaminate the two, with domain-specific knowledge being the hard bit and the skills being the biologically primary easy bit. For example, when you say (above) “What makes solving maths problems hard is the maths, not the problem solving” or (on Twitter) “collaboration is the easy bit”. Socrates said that philosophy was an activity not a body of knowledge: maths *is* the problem solving, it is not just the inert knowledge. It is doubtful whether information gets converted into long-term memory at all unless it is made meaningful to us by being applied in some way, e.g. to problem-solving. Collaborating can be very hard, depending on the context/content – and there is always a context.

    @LeoToAquarius argues not only that you need context to teach these things, but that there is no such thing as common, abstract skills, as I would argue there to be, even while accepting that they need to be learnt in concrete contexts. Problem solving in Maths does nothing to help problem solving in Design; criticism in History does nothing to help criticism in English. There is no synergy achieved by the uomo universale, but just a serendipitous aggregation of discrete bits of isolated knowledge. Though I have yet to review @LeoToAquarius’ evidence, this strikes me as wrong. The chief point being made by Bill Schmidt when he talked about the “coherent curriculum” (a phrase recently much mis-quoted in the UK, notably by Tim Oates) was that knowledge is interlinked and that it is the networking of discreet items of information that turns information into knowledge. A famous anecdote illustrating scientific creativity is the discovery of the circular structure of the benzene molecule by Friedrich Kekulé, who saw in a dream the image of a snake eating its own tail. Would he have made this creative leap if he had not been familiar with Celtic art, from which this image is drawn? I remember a talk by an admissions tutor at Cambridge advising us to advise any student who wanted to study law to avoid at all costs opting for law at A level, but instead to take Latin, on the grounds of the mental training that such a highly structured language offered. Such an attitude is common among employers – they don’t want people with content knowledge but with thinking skills. How could this advice be right unless higher-order thinking skills are transferable?

    How does this connect to my comments about scale in paragraph 2, and my suggestion that the neo-traditionalist bloggers are ducking the hard questions? It is because I think that the reliance on the common sense of the individual classroom teacher, well trained in the methodology of direct instruction and the importance of knowledge, will not solve the problem of scale in our education system. For that, we need a more systematic approach to education, the centralization of the production of instructional resources (a point on which I *do* agree with Tim Oates), a greater willingness to view education as an explicitly describable technological process and not just a matter for the private intuition of the individual teacher. If that is even to be conceivable, we need to start with a clear description of our instructional objectives. Everyone at the moment has jumped to the conclusion that this cannot be done because our 25 year experiment with criterion referencing failed so badly. But the fact that something was done badly is not a reason why it cannot be done well.

    One of the many problems with criterion referencing was the way that we percieved the criteria to be binary metrics. So the “collaboration” skill gets ticked off after the students show that they can collaborate over which TV channel to watch, rather than on organising a complex and difficult project. A second problem is that we did not look carefully at the empirical evidence to check that the criteria were being applied consistently. A third problem is that we did not start to model how different criteria related to one another. One of the ways that such relationships need to be modelled is to understand that some objectives are more concrete and others are more abstract. When (if?) you say that there is no such thing as transferable skills, then from my point of view, you are challenging what I see as the urgent need for more explicit models of our various curriculum objectives and how they relate to each other.

    This is a theme which I shall be developing on my blog over the coming weeks. In the meantime, thanks for hosting such a long comment on your blog. I hope I have been able to clarify the points, if any, on which we disagree.


    • David Didau says:

      The substantive point seems to be that you think I’m “delaminating” skills from knowledge, and the accusation that I have said there is no such thing as transferable skills. As I have neither said nor believe either of these things it’s not easy to see what your disagreement is.

      • No, I have not assumed that you do not think there is any such thing as transferable skills. My disagreement is with your assertion that you cannot teach transferable skills, if that statement is made in any sense that implies that such transferable skills (like creativity or collaboration) are therefore innate. I am sorry if this was not clear last time – let me try one more time.

        1. Transferable skills are defined as skills that can be transferred between different subject domains, and are therefore, in themselves, not dependent on domain knowledge (hence the “delaminating” analogy). I hope we can agree on that definition.

        2. Skill and knowledge are inextricably intertwined and the progressive position in the past has mistakenly tried to separate them. Again, I think we agree.

        3. You deduce from (1) and (2) that we cannot teach transferable skills, which must therefore be innate.

        4. I deduce from (1) and (2) that transferable skills cannot be taught *directly*, without domain knowledge. But this does not mean that they cannot be taught at all and that they are therefore innate. I propose that by teaching “Maths creativity” in the context of mathematical knowledge, and “English creativity” in the context of knowledge about English etc., the student starts to develop a “general creativity” capability, which makes it easier to be creative in any knowledge domain. Although this is not taught directly in “creativity lessons”, it is nevertheless the consequence of teaching.

        @LeoToAcquarius understands this point and disagrees with me. He says that creativity in maths is incommensurate with creativity in English and never the twain shall meet. I think this is wrong. In response to @LeoToAcquarius’ justification of his position, I agree that if you learn to be creative in English, that will not enable you to be straight away creative in Maths because you will lack the necessary domain knowledge and skills are not applicable to concrete problems without contextual knowledge. But the student’s understanding of creative processes will enable him or her to learn that new domain knowledge more quickly, make sense of it more quickly, and use that knowledge more productively.

        I have put forward three further arguments in this respect, none of which, in my view, you have responded to.

        1. It is not enough to say that people are creative or people collaborate naturally. They may do it, like dogs may walk on their hind legs, but that does not mean that they do it well.

        2. I asked whether you believe that Gradgrind had a point. Is it possible to have a lot of knowledge yet *not* to have the associated skill to apply that knowledge well? If so, then although knowledge and skill may be entwined, they are not entirely the same thing. I disagreed with @LeoToAcquarius about onus. If knowledge and skill are not the same thing, then the onus falls on your side of the argument to show that there is a 1:1 relationship between them, not on me to show that there isn’t.

        3. I raised the question of eugenics, which you seemed to find absurd. But if you accept that there is a variation in ability in respect of abstract skills and you really believe that this is more than 60% accounted for by our inherited natures, then selective breeding is the only way in which we can improve the fundamental abilities of our students. If that is politically unacceptable, then anyone who wants our young people to be more creative, to show more initiative, work better together etc, must accept that there is nothing we can do about it – a very pessimistic position. The PISA reports have highlighted the belief in innate (in)abiilty as one key reason why Asian jurisdictions outperform Western ones, and Brian Simon makes the case that the whole history of education theory in the twentieth century has been blighted by such attitudes (see “Why no pedagogy in England? at

        I hope that makes my disagreement with your position more clear.


        • David Didau says:

          OK, let’s do this. Broadly speaking I accept your definition of ‘transferable skills’ and agree that such skills are are inextricably intertwined with domain knowledge. My deduction is that such a transferable skill is therefore oxymoronic.

          You, on the other hand say, “I propose that by teaching “Maths creativity” in the context of mathematical knowledge, and “English creativity” in the context of knowledge about English etc., the student starts to develop a “general creativity” capability, which makes it easier to be creative in any knowledge domain. ” Well, that seems to contradict what evidence we have on subject. I’m going to require rather more more than your say so to accept that such a transference might ever result in ‘general creativity’.

          Your argument in support of this position seems to boil down as follows:

          1. If you’re taught to think creativity about domain knowledge then you will begin to understand the “creative process” and therefore “learn that new domain knowledge more quickly, make sense of it more quickly, and use that knowledge more productively.” That seems to be an unreasonably strong assertion and I think the burden of proof is with you.

          2. Whether or not people are innately creative or collaborative they can learn to do it better. Yes, of course they can. I maintain that this depends on the quality and quantity of what we know.

          3. You ask “Is it possible to have a lot of knowledge yet *not* to have the associated skill to apply that knowledge well?” I’d suggest that ‘inflexible knowledge’ might have this characteristic but the more we know, the more flexible our knowledge becomes. I would further argue that what you call skill is better thought of as procedural knowledge, i.e. more knowledge, not something qualitatively different. If you want to argue that ‘skill’ is different to ‘knowledge’ then you’d have to show how you can be skilled in something you don’t know. This is a logical absurdity. Maybe we’d be better off talking about expertise instead?

          4. You seem to suggest that my ‘belief’ in the heritability of human traits is somehow open to question. This is pretty much a scientific consensus. The evidence from twin and adoption studies is really very robust and clearly shows that every trait is between 50-80% heritable. But to then argue that the only solution is eugenics is a bit odd. On the contrary, instead of selective breeding wouldn’t it make a great deal more sense to work on the 20-50% that *isn’t* heritable? There’s absolutely nothing pessimistic about that! The whole point of education is to provide the knowledge to enable the less advantaged to fulfil their potential. The alternative is somehow deny individual differences and to believe the only difference between us is the quality of our education. I think *that* would be cause for pessimism! You might find ‘G is for Genes’ an enlightening starting point on this subject:

          • Thanks for the reply.

            The possibility that skills might be inseparable from the domain knowledge in the context of which they are first acquired (or, in your terms, that “transferable skills” might be an oxymoron) is precisely the point that I address with the Gradgrind example; and that Grumpywearymathsteacher also raises. It seems to be common that people often have plenty of propositional knowledge that they cannot apply effectively. This strongly suggests that the two are not the same.

            It seems to me strange that you take such a radical position against transferability of skills in this post, yet in your more recent post, “The trouble with transfer” ( you “define learning as the long-term retention of knowledge and skills and the ability to transfer between contexts”, recognising how difficult is this process. Are you not being inconsistent to argue at one moment that transfer is the holy grail of education and the next moment that it is an oxymoron? Is there a qualitative difference in the mental ability required to transfer between contexts and the mental ability required to transfer between domains?

            You suggest that “inflexible knowledge” equates to “little knowledge” and the more facts you amass, ipso facto the better you become at applying them. I think you put the cart before the horse: I would say that the more you apply your memorized facts, the more meaningful and secure your knowledge becomes.

            I suggest that quantity is not enough and that there is also a quality of interlinkedness, which varies between cases. Many students will have memorized the facts required to solve a problem but will not be able to see their relevance or how to apply them. Yet (and this muddies the distinction between quantity over quality of knowledge), I also suspect that *without* good interlinking and modelling, knowledge becomes insecure. In other words, secure knowledge *is* the interlinkedness and not the isolated facts. Which doesn’t mean that there is not also insecure knowledge of isolated facts: Grandgrind’s sort of knowledge.

            I agree with your point that skill is just another form of knowledge and I have never attempted to argue otherwise. I prefer to talk about “knowing how” rather than “procedural knowledge”, because the latter seems to me to reduce “knowing how” to a form of explicit propositional knowledge (knowing that you have to do A then B then C) when “knowing how” is much more commonly tacit and unconscious. But one has bear in mind that people commonly use “knowledge” to mean propositional knowledge, in distinction to “skills” – so one has to be sensitive to what other people mean by the words that they use.

            On the heritability of human traits, again, you slightly mistake my position. I do not doubt this at all. But knowledge is not a trait. Which begs the question, is intelligence a sort of knowledge or a sort of trait? I suspect that it is the same sort of question that I remember dominating O level physics – is light a particle or a wave? – the answer being one that challenges our understanding of the distinction between the two.

            In suggesting that I am making up my position as I go along, you completely ignore the reference to Willingham to support my position that (a) general intelligence exists, and (b) that it is strongly influenced by environment and hard work.

            You are very confident that “The evidence from twin and adoption studies is really very robust and clearly shows that every trait is between 50-80% heritable”. Well, Willingham does not agree with you and reckons you (and maybe Kathryn Asbury) are 20 years out of date. Here is a summary of his argument from pages 175-78 of “Why children don’t like school”.

            1. “The results of these [identical twin studies] are startling. Genetics seems to play a huge role in general intelligence; that is, our genes seem to be responsible for something like 50% of our smarts…For young children, it’s more like 20%, then it goes up to 40% for older children, and its 60% or even higher later in life. [i.e the evidence for genetic inheritance, even at its strongest, is much less strong than you suppose]. This increase is the opposite of what you might expect.

            2. “A real turning point in this work came during the 1980s with the discovery that over the last half-century IQ scores have shown quite substantial gains.

            3. “How does this assessment fit with the studies of twins?… No one is completely sure, but Flynn has a pretty good suggestion. He claims that the effect of genetics is actually fairly modest. It looks large because the effect of genetics is to make the person likely to seek out particularly environments.

            4. “The key idea here is that genetics and the environment interact. Small differences in genetic inheritance can steer people to seek different experiences in their environments, and it is differences in these environmental experiences, especially over the long term, that have large cognitive consequences”.

            Two further insights could be thrown in here: one is the Matthew effect (to those who have more shall be given) and the second are recent discoveries regarding the plasticity of the brain: even our “hardware” responds to the environment and is not a genetic “given”, as one might assume from e.g. a computer analogy.

            By this, to my mind plausible account, you are wrong on three counts:
            * on the extent of the influence of genetics, expressed in percentage terms;
            * second on the way that you present genetics and environment as dichotomous;
            * in your certainty that you are right.

            We are nevertheless in violent agreement that:
            * we inherit genetic predispositions,
            * (contra Rose Luckin) we should not try and teach skill without propositional knowledge;
            * as educators, we should focus on the malleability of human capability.

            But this third point is not conveyed well in your post, with its (I think rather simplistic) theory of innate, biologically primary skills. I suspect that a more plausible explanation of the difference between what is difficult to learn and what is easy to learn is the availability of feedback – which is plentiful in the context of the physical world but very scarce in the context of logical reasoning, when it is dependent on interaction with an expert, who is often hard to come by – again, an environmental and not a genetic explanation.

            Thank you for the book recommendations. I will put Geary, Asbury and your own recent book on psychology on my reading list to see if any of them challenge what I find to be the convincing summary of this topic to be found in Willingham.


          • David Didau says:

            Hi Crispin, thanks for the latest missive. If I’m reading you right, you appear to think that ‘transferable skills’ and ‘transfer’ are somehow synonymous. They’re not.

            Let me have another go at trying to explain what I mean. Propositional knowledge than can’t be applied outside of the context in which it was learned is ‘inflexible’. Increasing the quality and quantity of such knowledge makes it more flexible. To suggest there is an additional “a quality of interlinkedness” seems to unnecessarily complicate things. Of course additional practice at retrieving and applying propositional knowledge will create both more flexible (useful) knowledge and more ingrained procedural knowledge. So, the more you know, the more you practice applying what you know, the more flexible (and ‘interlinked’) what you know becomes. That is to say, whenever we add to our store of knowledge we create additional nodes on our schema and whenever we retrieve these schema we create new prompts and cues which enable further retrieval to become progressive easier. Suggesting there’s some other mystical quality is a bit like suggesting we need a concept like ether to understand physics. Occam’s razor is instructive here:

            You’re right to say that knowledge isn’t a trait. This is exactly the point `I was clumsily trying to make: knowledge basically accounts for the 20-50% which *isn’t* heritable. While there’s little doubt that some people will find it easier to acquire knowledge than others, no one comes pre-loaded with any propositional or much procedural knowledge. It is teaching which can enable equitable access to the most useful knowledge and thus prevent differences between individuals being *just* dependent on environmental advantages.

            None of this is to dispute your propositions that “(a) general intelligence exists, and (b) that it is strongly influenced by environment and hard work.” I wholeheartedly agree with this. Willingham’s position in WDCLS isn’t really in conflict with the position I have tried to outline, but invoking the Flynn Effect is problematic in this context. All it really shows is the cultural context of IQ testing. No one really thinks we’ve become clever over the past century but it’s certainly true that our environment better lends it self to the kind of abstract thinking measured in IQ tests. This probably just shows that it’s devilishly difficult to separate knowledge from g.

            On to your analysis of where I’m wrong:

            1. “on the extent of the influence of genetics, expressed in percentage terms” – I may be wrong about this, but the scientific consensus appears to support me. The percentage estimates I’ve quoted are those generally used bu those working in the fields of behaviour genetics and epigenetics. That said, I’m happy to concede that these are just estimates and that trying to attach a nice round percentage to such imponderables in naive at best.
            * “second on the way that you present genetics and environment as dichotomous” – I have never said and do not think this. Of course it’s not a dichotomy. The science of epigenetics is all about the complex ways the environment affects heritability.
            * “in your certainty that you are right.” I’m not at all certain I’m right. I’m ALWAYS open to the possibility that I’m wrong and have made something of a career out of unpicking my certainties and admitting my errors. The confusion may come from the fact that I think my position is stronger than yours. It’s not so much I’m certain that I’m right, it’s more that I’m fairly confident you’re wrong.

            So, let’s agree that we ought to focus as you put it on “on the malleability of human capability”. You may be right that my blog is overly simplistic. This is probably a fault of the form and of the inflexibility of my knowledge. I’m learning more and more about the fascinating science of evolutionary psychology and I think the conclusions are increasingly inescapable, no matter how they conflict with what we might wish were true. That said, no doubt they can be explain better than I’ve managed in this post. I also agree with your point about feedback. It is absolutely a feature of secondary knowledge that trial and error feedback is often counterproductive. To my mind, this just supports the idea that some things are inherently easier to learn.

            Best, David

          • Dear David.

            We certainly seem to be converging (in most respects apart from you being so convinced that I am still wrong – about what exactly I am not quite sure).

            I agree with what you say about practice and I am not sure there is very much to separate my idea of interlinkedness and your point about different aptitudes for knowledge retrieval. I think my language is helpful because retrieval is not a single act (like measuring how many seconds it takes to unpack and erect your tent) but is relative to many different starting positions i.e. different initial stimuli (sense data or requirements) traced through other memories. Every act of retrieval implies a transfer and both terms (retrieval and transfer) are analogies for the creation of links. I also agree that what I would call transferable or generalised skills depend on an ever wider network of knowledge.

            Let me wrap things up from my side by restating why I thought it was worth spilling so much digital ink on this conversation in the first place. I have you down as a “traditionalist”, a proponent of the knowledge curriculum, a member of a group of educational revisionists which has had a great deal of fun over the last six years shooting down quack progressive theories that have ruled for the last 50. I agree with this argument, which I have also been making. But I see a danger of the pendulum swinging too far in the opposite direction – the rebels against the last intellectual tyranny beginning to impose their own. The traditionalists have done a pretty good job of winning the commanding heights but having cleared the ground, I don’t see much evidence of them doing much reconstruction work – except for Robert Peal who has turned to writing school textbooks instead of (very good) polemics. There comes a time when the new occupying army needs to switch their focus from hunting down the few remaining insurgents, hiding in their cellars, to what should be done to restore the sewage system and getting the traffic flowing.

            That is why (if you follow my Mosul analogy) I think the polarisation of knowledge and skills is unhelpful. It is the polarisation which is what unites the fighters on both sides of the argument – e.g. Daisy Christodoulou and Geert Biesta, as I am arguing in my recent posts on the purpose of education ( and is in my view blocking the construction of a new and demonstrably effective approach to delivering education in the real world.

            Best, Crispin.

    • Chester Draws says:

      they don’t want people with content knowledge but with thinking skills. How could this advice be right unless higher-order thinking skills are transferable?

      But employers don’t actually want people with transferred higher-order skills, if they are sensible. They want ones with innately higher ability, because that gives them the best base to work from. (Would you rather train as an athlete a person who naturally could run 100 m in 12 seconds, or one that could run it in 12 seconds only after years of training?)

      Employers are also looking for people who don’t shy from a challenge and are prepared to slog it out.

      Hence those that fight their way through Latin or Calculus are often preferred, even when the transferred skills are negligible — the person is either clever or hard-working.

      • You are misrepresenting the point of disagreement. It is not about natural flair vs graft – its about whether “flair” is “natural” in the first place, or learned.

        It is worth pointing out that your position is pessimistic and unhelpful, from the point of view of the educator (see my argument 3 about eugenics in my reply to David of October 23, 2016 at 8:51 am., above).

        Perhaps more significant for those interested in what is true, and not merely what is useful to believe, is that pure genetic inheritance is not really consistent with the evidence of the plasticity of the brain being brought forward to modern neuroscience.

        I would accept Dan Willingham on this (Why students don’t like school), when he says (a) of the existence of innate general intelligence:

        “Our genetic inheritance does impact our intelligence, but it seems to do so mostly through the environment. There is no doubt that intelligence can be changed [through sustained hard work]”

        and that (p.173):

        Whenever the question “Is it nature or nurture?” is asked, the answer is almost always “both”…but there has been a significant shift in researchers’ points of view in the last twenty years, from thinking that the answer is “both, but probably mostly genetic” to thinking it’s “both, but probably mostly environmental”.

        Willingham also disagrees with @LeoToAcquarius’ position that skill is not transferrable because, if you test Maths and English (p.172):

        “Separate cognitive processes contribute to verbal and mathematical intelligences, but g [general intelligence] contributes something to each of them too”.

        Final point about the attitude of employers: beggars can’t be choosers. I think a lot of employers would be very pleased to employ anyone who could run the 100m in 12 seconds, if that was what the job required, and wouldn’t be too fussed with how that capability had been achieved.


    • jim sweetman says:

      Life might be easier if we viewed teaching and learning as activities, sometimes hot sometimes cold, rather than magical processes. Also, around 40% of children learn nothing in school (look at the outcomes), around 10% get mentally ill because of it and most of the experts just think that what they did (which made them experts) is the answer for everyone.

  3. Jen says:

    “we need a more systematic approach to education, the centralization of the production of instructional resources (a point on which I *do* agree with Tim Oates), a greater willingness to view education as an explicitly describable technological process and not just a matter for the private intuition of the individual teacher.”

    While I do see your point, “the centralized production of instructional resources” has some potentially scary consequences. There would be a LOT of money for the person/company that produced those materials (they’d have a complete monopoly) and therefore a lot of politics involved in getting a particular message/point of view into the official texts. I have yet (24 years) to find a Spanish textbook that I like enough to find it worth the exorbitant price. Things are presented in a choppy manner with far too many major, yet unrelated concepts presented in one chapter. I question the ability of a national level committee to be able to “collaborate effectively” on designing a curriculum worthy of being forced on everyone.

    • Hi Jen,

      I agree with you (a) that our current learning resources are inadequate, (b) that we need to avoid monopolies, (c) and that that includes national level committees, (d) or a central curriculum that is forced on everyone.

      What I don’t agree with is that any of that follows from what I am saying.

      I *am* arguing that we need to devote more resources to the production of central resources – but this would represent a massive saving when viewed across the whole system, for which the main cost at the moment is staff wages. It is not good economics to spend this money on paying teachers to create poor quality resources, with a massive and inefficient duplication of effort that it involves.

      Why does (a), (b), (c) and (d) not follow from my position?

      (a) Mainly because we place too much emphasis on paper-based resources, which do not encode activity and feedback (the real “stuff” of instructional content – see my What do we mean by content But just because we do something badly, doesn’t mean we can’t do it well. Tim Oates is good on this with reference to Finland and Singapore. The management of relationships between different learning objectives (which, as you say, are complex) and between learning objectives and current student capabilities, is also something that digital systems will be able to manage much more effectively than passive, linear textbooks. Digital systems are also much more flexible, allowing “parameterization” of abstract processes, supporting the classroom teacher’s own strategy, rather than imposing a one-size-fits-all straight-jacket.

      (b) Our current system is highly monopolistic. The two main reasons for this are (i) the privileged position of the exam boards and the influence that they are allowed to exert over the production of textbooks; (ii) in the realm of digital technology, the lack of interoperability of educational data and the monopoly that Capita SIMS has been allowed to establish as a result.

      (c & d) By “centralize” I do not mean that we should go all the way, establishing a national monopoly. I mean that publishers (traditional and digital), operating in a competitive environment, should do well and once what teachers at the moment do badly and duplicate tens of thousands of times.

      It follows that teachers should have the freedom to choose what resources to use, supported by an open & competitive market for the supply of instructional resources. But that does not mean that teachers should not themselves be accountable for their educational objectives to a diverse range of accountability “channels”, which include parental influence (through a market in admissions), central government regulation (with respect to a core curriculum), and the influence of various societal stakeholders, such as employers, charities, subject academics etc – this influence may be exerted both (as at present) through central regulation and through influence on the market (e.g. collaborations with different suppliers, certification regimes etc).

      Hope this helps answer your concerns. Crispin.

  4. Hi Crispin you’ve misunderstood my position on abstract thinking skills 🙂

    “@LeoToAquarius argues not only that you need context to teach these things, but that there is no such thing as common, abstract skills”

    I said the exist but cannot be taught

    I agree with Geary generic abstract ‘skills’ exist and they cannot be taught and therefore are minimal in importance in education in planning, teaching a lesson or in assessment – the key being generic thinking i.e the pseudoscience of Bloom’s or the SOLO taxonomy

    This is because they are evolutionarily innate

    They do not fit the dictionary or Cog Sci definitions of the word skill, namely a process (series of steps) to achieve an outcome

    For example in English at present there is an emphasis on ‘Skimming & scanning’ skills and teaching these to the students.

    ‘Skimming & scanning’ is a innate ”skill” it cannot be taught in abstract – we do this automatically when we look around a room, forest, or in a book we can’t be taught look at those trees now move your eyes like this etc

    Further English teachers have been seduced into thinking that ‘Skimming & scanning’ can be taught through English by practicing doing it on pieces of text, sometimes under a time constraint. The idea being that by practicing ‘Skimming & scanning’ in English this will make them better at it.

    However ‘Skimming & scanning’ is innate and happens automatically – but knowledge transforms this ability by allowing it to be applied automatically in different contexts.

    You and I will automatically skim and scan a broken car engine, however a mechanic will do the same and be able to fix it depending on the quality & relevance of their prior memorised knowledge

    Tircot discuss in the paper below the biologist Geary which Didau has elaborated in this blog post but also the experimental evidence to support this view.

    You also think that I think that:
    “There is no synergy achieved by the uomo universale, but just a serendipitous aggregation of discrete bits of isolated knowledge”

    No you’ve misunderstood – I said that being good at thinking in one subject (domain) does not transfer to another – knowledge cannot be learnt in isolation it’s scientifically impossible & we have no experimental evidence to support the view that knowledge can be learnt in a isolated way.

    Any new knowledge will be linked to prior knowledge hence science and engineering advance & are communicated by the use of analogies and metaphors

    So being good at thinking in English, understanding poetry,writing a piece of literature does not transfer to being good at thinking in Science

    However a knowledge of English, and of it’s literature helps e.g ‘ouroboros’ being used by Kekule

    So knowledge transfers but not thinking.

    Hence my comments that ‘generic abstract thinking skills’ i.e Bloom’s/ SOLO cannot be taught and lots of practice using them does not improve (transferred from once subject to another) their use in another subject

    You don’t get better at analysis by doing lots of analysis in different subjects.

    Being good in analysis does not make you better at analysis in a new subject that you have never learnt e.g mastering make-up

    However ‘knowledge of science’ does transfer to using make-up in a limited way

    If I’m good at communicating in English I know how to write a book / essay etc then that knowledge (not thinking skills) will help to a limited extent in writing a science article.

    However it will be more important that I gain the knowledge of how to summarise science research e.g in a paper/dissertation – which is the expectation in any Science honours degree course

    The US senate at present thinks that Global Warming is not real because of several cold winters –

    Therefore thinking ability has not transferred, and in this case a lack of science knowledge means they can’t even think about the issue rationally

    Your post:
    “The chief point being made by Bill Schmidt when he talked about the “coherent curriculum” (a phrase recently much mis-quoted in the UK, notably by Tim Oates) was that knowledge is interlinked and that it is the networking of discreet items of information that turns information into knowledge.”

    I’ve put a link to his ideas here
    by W Schmidt – ‎2002

    He’s arguing that “teachers in the highest achieving countries have coherent guidelines in the form of a national curriculum”.

    This applies to a subject & a whole curriculum in terms of ensuring the core knowledge a society wishes to pass on

    What it can’t do is plan a curriculum with cross curricular links

    Because where the knowledge does transfer the subjects fall into the same domain for example Geography has been subsumed into science as an applied science – especially physical geography

    So critical thinking about science is different from critical thinking in ”human geography” (does not transfer) but similar – it does transfer in ”physical geography” as it’s the same domain

    The issue in the UK is Science & maths curricula are a “mile-wide-inch-deep” we’ve too much stuff to teach with too little time to practice it at depth – not that the concepts taught are lacking depth – Schmidt seems to have made the mistake that because topics are taught yearly – or appear to be started ‘late’ they don’t have depth – this is not supported by examining the content of those topics

    “Our intended content is highly repetitive. We introduce topics early and then repeat them year after year. To make matters worse, very little depth is added each time the topic
    is addressed because each year we devote much of the time to reviewing the topic.” Schmidt

    This is simply wrong, some topics need to be covered yearly to ensure they are memorised through practice especially in maths / science and secondly he assumes that depth can or should be added to them without bothering to examine the topics

    He is correct that

    “Our intended content is not focused. If you look at state standards, you’ll find more topics at each grade level than in any other nation. If you look at U.S. textbooks, you’ll find there is no textbook in the world that has as many topics as our mathematics textbooks, bar none.

    In fact, according to TIMSS data, eighth-grade mathematics textbooks in Japan have around 10 topics, but U.S. eighth-grade textbooks have over 30 topics. (See photo on page 20.) And finally, if you look in the classroom, you’ll find that U.S. teachers cover
    more topics than teachers in any other country.” Schmidt

    I agree with you that:

    “we need a more systematic approach to education, the centralization of the production of instructional resources (a point on which I *do* agree with Tim Oates), a greater willingness to view education as an explicitly describable technological process and not just a matter for the private intuition of the individual teacher.

    If that is even to be conceivable, we need to start with a clear description of our instructional objectives.

    Everyone at the moment has jumped to the conclusion that this cannot be done because our 25 year experiment with criterion referencing failed so badly. But the fact that something was done badly is not a reason why it cannot be done well.”

    I would suggest that the knowledge components specified by cognitive science listed by me in this thread,

    would allow the creation of a coherent curriculum for each subject and connections if they form part of the same domain but only if at secondary

    As sci/”physical geography”/geology are the same domain at university & post university but in a limited way at secondary & A level

    As for scalability if we had a national curriculum with less topics (not more knowledge [depth] science has more depth at UK GCSE secondary than other countries & too many topics as well as at A level) then we would have the time for them to learn the material properly

    It maybe the case that other non-science subject have too little depth or difficulty & that may include bits of maths

    But science has too many topics with too little time

    Kind regards


    • Thanks for your long reply. I agree with many of your attacks on the teaching of bogus skills. I agree with you on the importance of teaching any skill in the context of knowledge – that knowledge and skill are often two sides of the same coin.

      Where our difference lies, I think, is the extent to which skills learnt in different concrete contexts feed off each other. It is certainly a well established, common sense view that e.g. what employers are looking for when recruiting is not specific content knowledge but good attitudes and intellectual disciplines. This is summed up in what is probably the mis-attributed quotation from Einstein that what is valuable in our education is what stays with us after everything we have learnt is forgotten – i.e. the higher order skills and attitudes.

      You say that analysis in one subject does not transfer to analysis in another subject. But I would not expect this to be the case – because the content knowledge in that second domain is also necessary and if you are not familiar with that domain, then you are nowhere. What I would expect (and believe that experience strongly confirms) is that someone who has learnt an analytical approach in one domain then finds it much easier to learn an analytical approach in the second domain – the facts as you learn them fall into familiar relationships with one another, because we already have the abstract understandings and patterns of thought.

      I think much of this argument comes down to an understanding of abstract entities. I am saying that general thinking skills are abstract. They are learnt *through* concrete contexts and need to be instantiated back to concrete contexts in order to be useful. It seems to me that you and David are mistaking the abstract nature of these abilities for something concrete – here is a particular case that requires you to exercise your general thinking skills. That is the mistake made by those who try to teach these things directly. You absorb some of this mode of thought by saying that yes, we have this mythical thinking skill, but that it is easy and is a matter of biological disposition.

      As for biological dispositions, we have a biological disposition to hoover up knowledge. Does that make learning advanced maths easy? No, it is very hard. The general approach taken by most psychologists to the nature-nurture debate is that these are not alternatives (as you are proposing) but that most of what we learn depends on the interaction between the two. That is where I am at.

      As for Bill Schmidt, I am familar with your reference to “A coherent curriculum”. The reason that Tim Oates mis-quotes Schmidt is that he references the later 2006 article in which he examines whether the sort of curriculum coherence that he advocates is achieved by the publication of national curricula or guidelines, and concludes that this is not necessarily the case. But Oates assumes that curriculum coherence is all about the alignment of policy instruments and not, as the 2002 and early papers suggest, about the relationship between different topics.

      I think that the essential characteristic of Schmidt’s coherent curriculum is not depth or number of topics, but the interlinked nature of those topics. So in maths there might be a few “big ideas”, but many different ways in which those ideas are applied. It is the characteristic of a very wide puddle (“a mile wide and an inch deep”) is that it becomes disconnected, while a solid body of water is connected, with the different parts of such a curriculum being heavily cross referenced. Part of our ability to model such a curriculum depends on our ability to distinguish between more abstract concepts and skills against more concrete applications of those concepts and skills.

      As for describing the different “knowledge components”, my view is that it is not really very helpful to distinguish between knowledge, understanding, skills etc – all that is observable of any of these internal states and dispositions is capability, which reduces to performance.


  5. KenS says:

    First, Mr. Didau, thank you for all your thoughtful writing on education. I thoroughly enjoyed your book, “What if Everything You Knew About Education Was Wrong?” You’ve pushed me to be a better, more reflective teacher. My copy of “What Every Teacher Needs to Know About . . . Psychology” is sitting on a table at home waiting to be read right after the book I’m currently working on. More on that in a moment.

    Second, thanks to Mr. Weston for his thoughtful and thought-provoking contributions to this important topic.

    I wish I had more time this morning to make my own puny contribution, but I have to teach shortly. I’ll take the risk of having my brevity being taken as superficiality and add this:

    I think many otherwise intelligent people – in both education and business – mistakenly espouse the need for the teaching of skills like creativity, problem-solving, and collaboration because of their ignorance (strong word, I know) of the role long-term memory plays in cognition. Many of my colleagues pay lip service to the power of unconscious cognition, but I have the sense they don’t really understand how little of that iceberg is visible. Greg Ashman repeatedly, and with good reason, cites Kirschner, Sweller, and Clark’s paper on the problems with minimal guidance that states, ” . . . long-term memory is now viewed as the central, dominant structure of human cognition. Everything we see, hear, and think about is critically dependent on and influenced by our long-term memory.”

    The book that has delayed my starting “What Every Teacher Needs to Know About . . . Psychology” is Fauconnier and Turner’s “The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities.” I’m only three chapters in, but it is a fascinating exploration of the role of the unconscious cognitive mechanisms active in our construction of meaning. The writers state, “We rarely realize the extent of background knowledge and structure that we bring into a blend unconsciously. Blends [combinations of ] recruit great ranges of such background meaning. Pattern completion is the most basic kind of recruitment: We see some parts of a familiar frame of meaning, and much more of the frame is recruited silently but effectively to the blend.”

    I take that to mean that, while we perceive “problem solving” to be a mental operation that involves manipulating information in our consciousness, most of the heavy lifting is actually done behind the curtain and out of sight.

    I might be stretching here, but I have the sense that Fauconnier and Turner’s ideas on conceptual blending might provide at least a partial explanation of the difficulties of transfer. But I’ll have to give that one more thought. Time to teach!

    • Chester Draws says:

      I think many otherwise intelligent people – in both education and business – mistakenly espouse the need for the teaching of skills like creativity, problem-solving, and collaboration because of their ignorance (strong word, I know) of the role long-term memory plays in cognition.

      Weirdly, when outsiders comment on the particular area of knowledge of those business people it is not well received. If I spouted off about AI to Prof Luckin I have no doubt I would be told that I didn’t know what I was talking about.

      At that point it should be clear, contrary to pronouncements, that detailed knowledge of a topic is much more important than they proclaim. That critical thinking is useless without it.

  6. […] recent posts by Carl Hendrick and David Didau, and this oft-cited article by Daniel Willingham are representative of the arguments against […]

  7. […] It should come as little surprise to hear that some of what human beings can do is innate. That is to say, we are born with various capacities and abilities which appear to be ‘hardwired’ into our brains. The evolutionary psychologist David Geary talks about such capacities as being either biologically primary or secondary adaptations.  […]

  8. […] other point to make is that we all have an innate capacity to do many of these things already. We’re born with the ability to organise environmental stimuli into schemas which then form […]

  9. Bill Allen says:

    Brilliant. Particularly appreciated the point about teachers not telling others what to do yet everyone tells teachers what to do.
    I first read about Garry in your psychology book and it clarified so many things. Really helpful

  10. […] critical thinking, problem solving, resilience and whatever else you care to mention are evolutionary adaptations that we pick up from the environment as very young children, and, to be meaningful in school and […]

  11. […] In fact, what is known about traits such as critical thinking suggests that they are ‘biologically primary’ and don’t even need to be learned [see an accessible explanation here]. […]

  12. Matt says:

    Is there a body of evidence suggesting/demonstrating you can’t teach creativity, problem solving and critical thinking?

Constructive feedback is always appreciated

%d bloggers like this: