I just finished one of the best books I’ve read on the science of learning. Daniel Willingham is a Harvard educated cognitive scientist who writes books [1] and articles [2] about how to learn and teach better.
The title of his book, Why Don’t Students Like School? [1], is a tad unfortunate, I think, because the book isn’t really about bored students. Instead, the book is divided into principles of learning. In order to make the cut, these principles needed to fulfill a strict set of scientific criteria:
- Robust scientific support. In Willingham’s words, “Each principle is based on a great deal of data, not only one or two studies. If any of these principles is wrong, something close to it is right.”
- Doesn’t depend on circumstances. These are facts about how human brains learn, so they don’t change whether you’re learning Spanish or mathematics.
- Ignoring it would be costly. Using the principles versus not using them showed a big difference in results. The principles aren’t just theoretical concerns but practically significant.
- Suggests non-obvious applications. The final criteria was that the implications of the principle should suggest new ways of teaching and learning.
The book is excellent, and I highly recommend getting a copy for yourself as Willingham explains many of the details and implications of each of these principles. I wanted to discuss each principle briefly, to share the implications it has for learning better.
Side note: The book lists nine principles, but two were more related to teaching, so I omitted them here.
1. Factual knowledge precedes skill.
Einstein was wrong [3]. Knowledge is more important than imagination, because knowledge is what allows us to imagine. There is considerable research showing the importance of background knowledge to how well we learn. Without background knowledge, the kinds of insights Einstein praised are impossible.
Careful studies show that having more background knowledge on a topic means we can read faster, understand more when we do and remember more of it later. This means knowledge is exponential growth [4], with past knowledge becoming a crucial factor in the speed at which more knowledge is acquired.
This means that you cannot teach someone “how” to think, without first teaching them a considerable amount of “what” to think. Thinking well first requires knowing a lot of stuff, and there’s no way around it.
2. Memory is the residue of thought.
You remember what you think about. Whatever aspect of what you’re learning your mind dwells on, will be the part that it is likely to be retained. If you, inadvertently, spend your studying time thinking about the wrong aspects of your studies you won’t remember much of use.
The problem with this principle is that knowing about it is not enough. We can’t constantly self-monitor our own cognition, noticing what we’re noticing. So even if you try to pay attention to the right things, it can be easy to accidentally focus on less important details which will take precedence in memory.
This is a reason why highlighting is often a lousy tactic. When you highlight, you’re not focusing on underlying meaning, but observing bolded words or particularly emphasized sentences. So you don’t remember much.
I recommend tactics like paraphrasing with sparse notes while reading, the Feynman technique [5] or taking pauses during a reading session to quickly recap what you just read. These are orienting tasks that encourage you to spend more time thinking about underlying meaning, which is almost always what you want to be learning.
This also shows one of the weaknesses I’ve seen in students who misuse analogies. If the analogy you make causes you to think about a surface detail of a concept, and not the underlying structure, you’ll only remember surface details on the test. A metaphor for voltage that uses volcanoes because they both start with “V” won’t help you with problems. The metaphor that voltage is analogous to height is useful because you’re forced to think about what voltage means (in this case the relation between gravitational and electric potential).
Interestingly, this also has implications for languages. The reason the “sounds like” method for memorizing vocabulary words can work is because it forces you to think about how a word sounds more exactly. Having to come up with an image that links to the sound forces you to spend a couple seconds thinking about what the word actually sounds like.
3. We understand new things in the context of what we already know.
Abstract subjects like math, physics, finance or law, can often be hard for people to learn. The reason why is that the we learn things by their relation to other things we already know (sound familiar? [6]). Willingham here suggests using many examples to ground a particular abstraction in concrete terms before moving on.
I would also add that I believe people overestimate their ability to learn abstract things. As such, we tell ourselves we understand an idea without first grounding it in numerous examples or analogies. Smart learners correctly understand the brains weakness for abstraction and build scaffolding to support new ideas before they fully set.
Occasionally when I recommend to students metaphors or analogies for learning a subject, they come up blank. I admit, it can be a tricky technique. But I believe part of the difficulty is that it points out when you don’t really understand a concept. If you understand a concept but can’t put it into a single example or analogy, you don’t really understand it at all (and should first do something like the Feynman technique to get that understanding).
4. Proficiency requires practice.
The only way to become good at skills is to practice them. Additionally, some basic skills require thorough practice in order to be successful at more complicated skills.
Math is an excellent example: you may have a conceptual understanding of calculus, but if you aren’t fully fluent with algebra, it will take you hours to do a simple problem. The only way to make algebra automatic is to practice a lot of problems.
I’ve certainly been guilty [7] of downplaying the importance of repetitive practice in some of my early writing. But there’s no way I could have completed the MIT Challenge [8] or this language project [9] without extensive time spent practicing the basic tools for each subject. Merely understanding isn’t enough.
Willinham suggests an alternative to repetitive practice which can be painfully dull: learn harder subjects that require practicing earlier material. One study showed that those who took an algebra class showed rapid and predictable decline of their skills. The one group that didn’t? Those who learned calculus.
5. Cognition is fundamentally different early and late in training.
Should you learn physics like Newton? [10] For that matter, should you learn science like a scientist, making hypothesis, testing experiments, revising your theory to fit the data? Willingham offers substantial evidence that the answer is no.
I think there’s merit in understanding how scientists perform their work, but it’s also clear that knowledge creation and knowledge acquisition are very different. Because they are different, the learner needs to weigh them against each other. For most disciplines, understanding scientific facts is more important than scientific process, for the simple reason that scientific facts will inform our lives, but few of us will ever do scientific research. The same applies to history, philosophy and nearly any other discipline of knowledge.
Another implication of this is that the ideal method for learning a subject and creating knowledge within a subject will be different. Learning calculus and inventing calculus bear little resemblance, so don’t worry if you can’t learn calculus the way Newton did. You don’t have to.
6. People are more alike than different in how we learn.
Learning styles are bunk [11]. There is no such thing as visual, auditory or kinesthetic learners. This is also true for every serious theory of different cognitive styles for learning.
Defending this conclusion takes a bit of thought, because to most people the idea that people learn differently is obviously true, even though research says otherwise.
Part of the confusion stems from the fact that different abilities can exist while styles do not. Meaning Johnny might be really good at processing visual information and Mary might be good at processing auditory information. Show Johnny a map and he’ll remember where everything is better than Mary. Play Mary a tune, and she can hum it back a week later.
But this isn’t what a theory of learning styles suggests. It suggests that if you taught the same subject to both Johnny and Mary, and played Johnny a slideshow and Mary an audiobook, they would learn better than if Johnny had listened and Mary had watched. The experiments simply don’t find that.
This suggests that the ways we learn are more similar than different. Some people might be better at learning certain types of things than others, but given a particular subject, science hasn’t different ways of learning it that are consistently better for some people but not others.
Side note: Willingham also debunks holistic versus linear thinkers. However the only thing it shares with my idea of “holistic” learning is the name. My version of holistic learning is not a learning style in the sense Willingham debunks here, but a strategy and one that happens to closely correspond with the third cognitive principle listed above. The nomenclature is my mistake, owing to my being unaware of the other learning theory that used the same name at the time. I’ve since used tried to use the word less, preferring “learning by connections” to avoid confusion.
7. Intelligence can be changed through sustained hard work.
This was probably my favorite part of the entire book because it validates much of what I said here [12]. Intelligence is partially genetic and partially environmental. Innate differences do matter and some people are born with more talent than others.
However, Willingham argues that intelligence is malleable. Psychologists used to believe that intelligence was mostly genes. Twin studies and other natural experiments seemed to bear that out. Adopted children turn out more like their biological parents than their adoptive parents in many dimensions.
However, now the consensus has turned far more towards nurture, rather than nature. One of the biggest pieces of evidence is the Flynn Effect [13], which is the observation that people, over the last century, have gotten smarter (and the effect is too large to be from natural selection). Genes may have an important role in intelligence, but most of that role is played out through the environment, not independent of it.
If you re-read the first principle I listed, that shouldn’t be surprising. Knowledge being exponential growth means that a small initial advantage can quickly compound. If genes gave you a 5% headstart in math in kindergarten, there may not be much difference between you and a similar child. However, expand that small initial advantage over thirty years and you may have someone who has done a PhD in physics and someone who stopped at high-school.
From a population standpoint the difference between these two people may be “explained” by differences in genes. However, genes only created a small headstart. Sustained hard work can help set off your own exponential growth of learning in a domain as well.
Concluding Thoughts
I thoroughly enjoyed this book, and don’t let my brief summary and insights spoil it for you. It’s a fairly easy read while still being smart and insightful. What’s more, the book is based on robust research and science.
In terms of my own, more informal, writing about learning, I was happy that most of the principles discussed in the book reflected my own thinking. It’s comforting to see when the experience I’ve gained from my own learning challenges converges on the serious work scientists are doing to understand the brain and how we learn.