This week @ShakinthatChalk tweeted the following poll:
It is quite clear to me that the pupil in question had learnt the topic but had then forgotten it. The only way I could conceive that they hadn’t was if they had fluked the test, but surely that possibility was excluded by the fact that they’d ‘demonstrated secure understanding’.
But I was clearly in the minority! Something odd seems to have happened to our definition of ‘learning’.
1. Definitions matter because how we define words determines how we conceive of education and our roles within it
How we define words is not an inconsequential academic parlour game; how we define words like learning matters. ‘To learn’ is a morally good thing, it is one of the main reasons why children go to school. So, how we define it determines (at least one of) our aims in school, and consequently, it determines the nature of accountability and assessment in education.
Without clarity, it is impossible to achieve any kind of justice. When Confucius was asked by one of his followers what his first priority would be as a government administrator, he replied that ‘Without question, it would be to ensure that words are used properly… If words are incorrect, then statements do not accord with facts; and when statements and facts do not accord, then business is not properly executed; when business is not properly executed, order and harmony do not flourish; when order and harmony do not flourish, then justice becomes arbitrary; and when justice becomes arbitrary the people do not know how to move hand or foot.’[i]
2. The assumption that the meaning of a word is what it signifies leads to a reductive analysis of the concept of learning
The problems regarding definitions in education stem from a mistaken conception of language, and the resulting confused notion of how we should go about analysing concepts. For the sake of argument, let’s call this mistaken view the representationalist view of language.
According to representationalism, the meaning of a word is what it signifies. We explain the meaning of words by pointing at things. Nouns are essentially names (for ideas, processes, entities). Language is fundamentally descriptive. We combine names to represent the world. Given this, to establish the meaning of a word, e.g. ‘learning’, we should describe what the word signifies. We need to come up with a checklist of features or conditions that we can tick-off and thereby know whether we can call that which is going on learning.
3. Reductive conceptual analysis leads to the fruitless game of hunt-the-counterexample, which in turn leads people to reject any kind of real analysis at all
Yet this process of finding necessary and sufficient conditions quickly becomes a tiresome game of hunt-the-counterexample: examples that either satisfy the checklist but aren’t learning or things that are clearly learning but don’t satisfy the checklist. Consider Kirschner, Sweller and Clark’s definition ‘learning… means that there has been a change made in one’s long-term memory’[ii]: A brain injury can cause change in long-term memory but isn’t learning (the condition isn’t sufficient); and a student can learn something and then forget it (i.e. @shakinthatchalk’s poll – the condition isn’t necessary).
Because this ‘reductive analysis’, this process of seeking necessary and sufficient conditions for calling something learning clearly bores people, and seems to be a thankless task, many people just try and ignore it, or stepover it.
4. It makes no sense to reject conceptual analysis of learning on the grounds that learning is invisible
Some claim that learning is somehow mysterious, hidden, invisible, goes on inside someone’s head, and so we can’t really know what it is. Therefore, we need to use a proxy to establish whether learning is really happening. E.g. Rob Coe says engagement is a bad proxy.[iii]
—But if learning is invisible, we have no way of confirming whether it’s a bad proxy, because we’d only be able to confirm/disconfirm it using another proxy which itself couldn’t be confirmed etc… our judgements about whether learning is taking place are all unfalsifiable.
5. It makes no sense to reject conceptual analysis of learning on the grounds that we don’t yet really know what learning is
Some claim that we don’t know what learning is yet, but in the future, we will be able to define learning in terms of brain processes.
—But we have the same problems as before: without first having a definition of learning we cannot know whether a brain process correlates with it.
6. It makes no sense to reject conceptual analysis of learning on the grounds that since we can’t find a real definition of learning, it must be an illusion
Some might think that learning cannot be directly observed and a definition cannot be agreed upon, that perhaps it is a nonsense term, and we should just stop using it.
—This would seem like a rather daft option for education researchers and educators to take. Perhaps the alternative might be a return to some kind of purely behaviouristic approach of Skinner, where education becomes merely a matter of ‘conditioning’. If this is the case then there is something of a vacuum when it comes to the values that drive education (see Tweets 1 and 9).
7. Theory specific definitions of learning are often conflated with ordinary usage and this is a great source of confusion
Daniel Willingham explains that educational researchers define learning ‘in the context of specific theories or… specific goals or projects.’ Education researchers are not trying to provide ‘a definition that claims reality independent of any particular theory, set of goals, or assumptions.’[iv] —This’d be fine if it was always clear that, e.g. ‘learning is a change in long-term memory’ was something theory specific; that it was not a redefinition, or an improved definition of learning as commonly understood but something quite different. But these theory specific definitions are frequently conflated with our ordinary usage, hence our confusions regarding the definition of learning. (And remember the furore over Ofsted’s use of the term?) Furthermore, if scientific definitions are theory-specific, how do they connect with ordinary language? This borrowing of words from ordinary language is fraught with danger. As Werner Heisenberg said, ‘words are…taken from our daily life situation where they have a good meaning, but when we use such terms we are usually extrapolating from our daily lives into an area very remote from it, where we cannot expect the words to have a meaning.’[v]
8. Arguing that we should just adopt the definitions associated with the theories that ‘work’ begs the question
One might defend this approach by arguing that if a theory works, then that’s the one we should use and those are the definitions we should adopt? —But then we have the same problem: whether a theory works depends upon how you define works – which depends on how you define learning.
9. The meaning of words is not a matter of choice or political ideology
Some might think that the response to this mess is to just accept that there are value-judgements to be made, and recognise that at some point, we just need to commit to one definition or another, to one theory, one proxy etc. Everybody is entitled to their views, that’s why education must be in the hands of a democratically elected government.
—But this is a more bizarre conclusion than it first appears. So, we should just accept whatever definitions of learning are the governments decide to use because they’ve been democratically elected? What a strange and confused kind of tyranny this would be where rules of language are imposed upon us!? What kind of legislation could ensure that words have a particular meaning? We cannot legislate for understanding. If a word has a particular meaning in ordinary language we cannot refuse or choose to not understand what someone is saying; we can only pretend not to understand. At best, such an approach leads to a proliferation of what David Graeber would’ve called bullshit, bullshit tasks and bullshit jobs[vi]. At worst, it becomes a very sinister act of political oppression.
10. To resolve these confusions, we must recognise that the meaning of a word is not determined by what it signifies but by its use according to the rules of language
So…how do we get out of this mess? We need to return to the beginning and recognise that the basic assumption of the representationalism is wrong. The meaning of a word is not what it signifies. Instead, let’s look at, what P. M. S. Hacker calls an Anthropological conception of language.[vii]
We learn the meaning of words by pointing at examples, e.g. ‘that’s a car’, and we can describe those features without which it wouldn’t be the thing that we say it is. But it is not those features which determine whether a thing is what we say it is. Whether we can say a student has learnt something isn’t determined by a particular feature of the behaviour or work of the student, nor of the student themselves (including any brain-states). It’s determined by rules of language governing the use of the concept learning.
11. To clarify the meaning of words we need to focus on the rules in accordance with which the words are used
So, instead a checklist of conditions, we need to look at how the word is used, and what the rules relating to that word are in different circumstance. Our explanations of meaning are expressions of these rules, so we simply need to ask, ‘what does the word mean in this case?’
We already know what the word ‘learning’ means. The difficulty is that we don’t recognise the common-or-garden explanations of meaning as what it is we are looking for. We want something more all-encompassing, we want the secret sauce, holy grail. (Why do we want such a thing? This is a psychological rather than a philosophical question, but perhaps the everyday explanations are too messy for us. I also suspect it is because a real definition of something gives us a sense of power.)
12. The methods of Connective Analysis should replace Reductive Analysis
Instead of reductive analysis (searching for necessary and sufficient conditions) we need to do, what P. F. Strawson calls, connectiveanalysis[viii]: collect the rules and explanations of meaning (that we already know) to remind us of what sort of things can make sense and what can’t and under what conditions.
We should heed Strawson’s suggestion:
‘Let us abandon the notion of perfect simplicity in concepts; let us abandon even the notion that analysis must always be in the direction of greater simplicity. Let us imagine, instead, the model of an elaborate network, a system, of connected items, concepts, such that the function of each item, each concept, could, from the philosophical point of view, be properly understood only by.grasping its connections with the others, its place in the system- perhaps better still, the picture of a set of interlocking systems of such a kind.’
13. Connective Analysis reminds us that the word ‘learn’ has more than one meaning
Connective Analysis reminds us, for example, that the word ‘learn’ has a variety of meanings:
- To learn a poem is to commit it to memory;
- to learn about a poem is to acquire knowledge and understanding of the poem;
- to learn to ride a bike is to acquire the ability to ride a bike
- to learn of his arrival is to become aware of his arrival;
- to learn the truth is to realise the truth;
- to learn French is to study French;
- to learn a lesson is to improve as a result of an experience;
Whilst these (and the many, many other) uses of the word learn may overlap, there isn’t and doesn’t need to be one single meaning which encompasses them all. It is enough to say that there is, as Wittgenstein would say, a family resemblance between the various uses.[ix]
14. Connective Analysis reminds us that there is nothing essentially private about learning
Connective Analysis also reminds us that by correctly identifying the meaning of ‘to learn’ in a particular case, it becomes obvious that there is nothing essentially invisible or private about learning. In some cases, to have learnt something is to be able to explain it, use it and respond to it correctly. In such cases, these are not proxies they are what constitutes having learnt. So long as the student is cooperating and answering questions etc. the teacher need not guess as to whether they have learnt or not.
I am certain that there are many other things which a rigorous connective analysis of the concept of learning would yield which would have a huge impact on how we understand assessment, values in education, curriculum planning and a wide range of other issues.
15. The objection that we want to know about learning, not language misconstrues the nature of the relationship between the two.
Some people may object to the arguments that I’ve put forward here by saying, ‘I want to know about learning, not language; things not words’. —But to care about the meaning of words is to care about things. Consider the following questions:
- What does ‘learning’ mean?
- What is the concept of learning?
- What is learning?
They’re all asking the same thing; the answer to one gives the answer to the others. In cases when ‘learning’ means ‘the acquisition of knowledge’, the concept of learning is the concept of the acquisition of knowledge, and learning is the acquisition of knowledge.
16. It makes no sense to think that our ordinary definitions of learning are in someway inadequate and should be replaced by new scientific definitions
Others might think that our everyday explanations of ‘learning’ are informed by folk science and this has been surpassed and improved upon by Cognitive Science. Therefore, we should replace our ordinary language definitions with these new scientific ones. —But this confuses a definition, or explanation of meaning for an empirical statement, something that is true by virtue of observation (which would be true if the meaning of a word was what it signifies). But a definition is not an empirical statement, it is a rule. The sentence ‘learning is the acquisition of knowledge’ is not a description of anything, but a rule for the use of the word ‘learning’. Rules cannot be true or false, they are either followed or not, and so no amount of science can disprove it. Experimental psychology increases our understanding of human capacities and their relation to learning, or the underlying neurophysiological structures; and whilst scientists may invent new concepts of learning, they cannot redefine learning or discover what it really is.
17. Conclusion: Education debate requires conceptual clarity and that can be achieved with Connective but not Reductive Analysis
We in education need, not only experimental and observational data, but also conceptual clarity. To achieve that kind of clarity we need to engage in Connective Analysis to find the place of concepts like ‘learning’ in the web of language.
[ii] Paul A. Kirschner, John Sweller, and Richard E. Clark, ‘Why Minimal Guidance during Instruction Does Not Work: An Analysis of the Failure of Constructivist, Discovery, Problem-Based, Experiential, and Inquiry-Based Teaching’, Educational Psychologist, 41.2 (2006), 75–86 <https://doi.org/10.1207/s15326985ep4102_1>.
[iv] Daniel Willingham, ‘On the Definition of Learning’, Daniel Willingham–Science and Education, 2017.
[v] F David Peat, ‘Interview with Walter Heisenberg’, Fdavidpeat.Com, 1970.
[viii] P. F. Strawson, Analysis and Metaphysics (Oxford: Oxford University Press, 1992) <http://weekly.cnbnews.com/news/article.html?no=124000>.
[ix] Ludwig Wittgenstein, Philosophical Investigations (Blackwell, 1967).