How We Learn - The Neuroscience Behind Education

Stanislas DeHaene is a cognitive neuroscientist who studies reading, number recognition, and learning in the brain. In his recent book How We Learn, he identifies four pillars of learning that are engaging and easy to understand.

These pillars simplify and expand our common language of learning. DeHaene is rigorously based in scientific research and he has done an excellent job of making the essential points easy to grasp for anyone working in education – all the way from pre-school to university, as well as parents and children themselves.

To begin, it helps to understand that DeHaene is taking us all the way down to the foundation here – the neurons. He’s trying to help us understand how learning is embodied in the physical structure of the brain. It’s helpful to remember that at this level anything a person learns is considered “learning.” We’re not merely talking about school and academics, though that’s certainly a major area of focus. We’re also talking about how eyes “learn” to see color, how babies “learn” to walk, and how curmudgeons “learn” to speak unkindly. All of it.

It All Starts with Attention

The brain receives sensory inputs during the day – sights, sounds, meaning, etc. Most of it is ignored, and that’s why attention is so important. It is pillar number one, and it helps to remember that our bodies and brains are picking up a huge range of sensations that we remain mostly unconscious of – like a giant satellite dish fielding a huge bandwidth of signals. When we give our attention to something it’s like narrowing our receiver down to a specific radio station.

But attention in the brain doesn’t just single something out, it also enhances, prolongs, and repeats some of those original signals – sort of like turning up the volume and putting our favorite song on repeat. This is accomplished because attention in the brain is initiated by “higher order” areas of the brain that cause a cascade of neural functions that repeat, amplify, and solidify the original sensory signals – making them more noticeable and memorable.

The essential point echoes what we’ve known for eons – if you’re not paying attention, you won’t learn much. Now we know why – because attention is the method by which our brains physically repeat important messages to itself. We can also restate the old adage with a little more clarity, and in the positive form – you’re only learning what you’re paying attention to. There’s no such thing as “not paying attention.” Rather, it’s simply a matter of what you’re paying attention to. Whether it’s the equation on the board or the squirrel outside the window, that’s what you will remember. That’s what your neurons are going to repeat to themselves.

So if your teacher thinks you’re not paying attention, she’s sort of missing the point. We’re always paying attention. The better question is – how do I help myself (or my students) direct attention to what is important in this moment? Good pursuit, that.

All or Nothing

DeHaene likens attention to a bottleneck. It would be wonderful if we had lots of attention to go around, but for whatever reason the brain seems to have limited resources here. In order to learn new tasks we have to give our attention to it, and that means that all other things must disappear for the time being. Multitasking is something we only do with tasks we’ve already mastered – like driving, talking on a hands-free device, and sipping coffee. Easy enough for most of us, and we’ll revisit this later in pillar four.

The important point is that when learning a completely new subject or skill, we have but one thread of attention to give. This is the bottleneck, and it’s crucial.

Active Engagement

To utilize the power of attention – something each and every one of us has – we have to be what DeHaene calls actively engaged. This is pillar two. The quick and dirty way to describe this is in opposition to a receptive or listening state. That’s not how we learn at all. There’s no such thing as pouring information or skill into a docile and receptive brain. It’s not possible.

That’s because the brain works by engaging two important cycles. The first is a “top-down” cycle of prediction (like a model or expectation), and the second is a “bottom-up” cycle of external stimuli. The brain works by matching the predictions of cycle one to the external data from cycle two – and both cycles are constantly refining and tuning each other. This is sometimes called a Bayesian process, and it’s worth understanding that a bit.

First, we have to abandon our satellite dish analogy. That’s not really how the brain works. Too receptive. A better analogy might be a baker creating a new recipe for a cake.

If he’s starting from scratch, the baker will require some trial and error. There’s simply no such thing as tasting a cake without first putting the ingredients together. So that’s what he does – he makes his first guess, his first recipe. This is the top-down cycle. When the cake is done, the baker tastes it – yuck! That’s the sense data coming in from the bottom up. Back to the drawing board he goes, adding a pinch of this, a little more sugar, two less eggs, a raise in baking temperature and a decrease in baking time.

Recipe. Bake. Taste. Recipe. Bake. Taste. Ideally, after several trials he has honed in on the winning recipe.

This is sort of what the brain is doing when it learns – testing its models or predictions against the sense data coming back in. Now we can understand what DeHaene means by active engagement. There’s no such thing, when it comes to learning in nerve cells, as simply eating a delicious cake. Your brain has to “cook it up” first. It has to guess at what it’s doing, what it’s seeing, in order to refine the sense data coming in and arrive at a finely tuned piece of learning.

It's a little counterintuitive, but everything works this way – eyesight, touch, hearing, algebra, even emotions. Our brains are never merely receiving signals. They are constantly predicting what those signals mean, then listening closer to see if they’re right. DeHaene calls this active engagement.

The commonsense lesson is this – in order to learn something, students should be actively engaged in trying to figure it out for themselves. Lectures might be okay, so long as students are actively trying to comprehend or question the material, but as a whole the rote transfer of information orally or textually is far inferior to tasking students with trying to make sense of it themselves.

DeHaene is clear that this does not mean you give a child a few textbooks and then leave them alone to figure it out by themselves. Neither does it mean kids need to be physically active as they learn (though that may be fine). It just means that children need to be mentally active in searching for the recipe of truth, not just blithely listening.

Error Feedback

Because our brains learn based on these two cycles, top-down and bottom-up, error feedback is crucial. That’s the bottom-up portion. It’s what tells our brain the prediction is off, or needs refinement. The cake tastes bad – do it over. We didn’t move our hands into the right place to catch the ball – try again. Missouri is not the capital of Tennessee.

Error feedback should be as immediate and emotionally neutral as possible. DeHaene, for all his conventional love of school, is very clear on this point. He dismisses grades as useless to the purpose of learning. They may have a purpose, but it isn’t related to learning.

The reason is that the brain cannot wait a week or a month to get feedback. It needs the feedback as close to the encounter as possible, otherwise the neurons are no longer engaged in this task and the feedback is pointless. Imagine trying to catch a ball, initiating the complex series of motor controls to put your hands in the right place, and…!

A week later you get a letter in the mail telling you whether you caught it or not. Ideally, error correction is in real time. That’s why grades are ineffective, DeHaene says. They do not refine learning. It may very well be that they accomplish other things, worthy things, but they have nothing to do with maximizing learning in the nervous tissue.

DeHaene is adamant about this. Teachers, children, and parents should be trained on this subject, so that error feedback is not an emotional experience. Why? Because the emotions tend to…duhn, duhn, duhn…get our attention! Since attention is a bottleneck, once it goes to the emotional content of the experience, or god forbid the conflict that erupts, then that’s what we’re learning. We have no more attention left for the actual lesson we started with in the first place.

That’s not to say that emotional lessons aren’t important. Of course they are! When we’re talking about learning, we’re talking about every possible thing we might encounter or experience in life. Obviously our emotions around grades, parents, teachers, etc., are important. DeHaene’s point isn’t that emotions ruin some sort of stoic search for rational exploration, but that poorly timed or delivered feedback tends to produce difficult emotions that distract from one’s original purpose. Therefore it’s essential, he says, to teach children, parents, and teachers how to give and receive error feedback in sensitive and timely ways.

You Can’t Learn Without Making Mistakes

The last piece of error feedback is mind-blowing: there’s no such thing as learning without mistakes. Just like the baker can’t “learn” without first coming up with a recipe, or top-down model, neither can he learn without tasting the cake. But it turns out that nerve cells have a very unique way of “tasting” or recording information.

Put simply, NEURONS DO NOT RECORD CORRECT ANSWERS. They record refinements in predictions and errors. If the very first recipe is a hit (correct), it doesn’t get recorded in the brain. The recipe may have been a success, but nothing triggers the neuron to remember the incident. Thus, it’s not repeatable.

The only way neurons fix a memory is by “fixing” the errors. Therefore they have to have errors to fix.

Consolidation

The final pillar of DeHaene is sleep. If you like brain science as much as I do, there is another great book by sleep scientist Matthew Walker called Why We Sleep that covers this topic thoroughly. It absolutely rewrites the concept of sleep. Far from being at rest, it is one of the most active times in the brain.

For our purposes here, it’s enough to mention that sleep is the brain’s way of cleaning house. This is when it gets rid of old synapses that aren’t being used and consolidates and strengthens the ones that are.

Sleep is also the process whereby the brain takes the learning we’ve given to our attention and brings it to other areas of the brain that make it more repeatable and automatic – freeing up the bottleneck for other things. This is why it’s so hard to learn to drive at 16, but at 35 you have no trouble driving, talking, and sipping a latte.

It's as if we are acquiring new information and skills via this spotlight of attention, then constantly shuttling them to the backseat, where they become easier to recall and more automatic – like riding a bike, or walking. Have you ever had that sensation where you feel as if you’re walking without hardly thinking or even willing the action? That’s because it’s so deeply seated in this automatic part of the brain that the entire operation flows almost effortlessly. Not so for a baby. It took you several months to a year of constant attention and dedication to get that motion right. Today, it’s like pressing the Walk app on your phone. It runs itself. Almost all your bandwidth of attention is still available for other tasks.

The important point here is that sleep is essential to learning. We neglect this crucial brain function to our detriment. Lack of sleep, whether in our children or ourselves, means not only that we’re a little more tired or grouchy, but that we have less automaticity in our functions and therefore have to give more of our attention to tasks we’ve already learned. Chronic lack of sleep is a vicious cycle that will dumb down even the brightest.

The Parfait Looks Like This

The best way to learn is to alternate regular periods of explicit instruction or study with periods of testing and consolidation (sleep, not just rest). That is the parfait. Testing here means opportunities for error feedback, not some standardized test. This is the brain’s easiest road to learning. Regular and repeated encounters with the same material (like routine math practice) is better than a prolonged and in-depth class on a specific subject that gets dropped afterward. Why? Because our attention tends to lag.

Periods of repetition should vary with increasing time intervals. Thus, at first the periods are frequent (the parfait is thin). This might be every day. Over time, the intervals should increase – weekly, then monthly, and even 4-6 months later. These long periods of review (and testing) are crucial to the error feedback and consolidation processes in the brain that lead to increased refinement and automaticity.

Even at the expert level, there is no reason to ever stop the long-term cycles of study and testing. Juggling is a good example. A juggler’s hands need to land in precise locations, with precise movements, in rapid succession. Learning to juggle takes a huge amount of attention, but like all things eventually becomes automatic.

But to be a professional juggler, one requires constant periods of practice to fine tune and refine the muscle and eye data to a level of perfection that will wow an audience. Thus jugglers, like every professional of every sort – surgeon, teacher, biologist, baker – must be in a constant state of practice and refinement to keep their skill sharp.

The 4 Pillars in Summary

Attention – Recruiting the higher order areas of our brains to amplify, prolong, and repeat learning in the initial sensory circuits.

Active Engagement – Listening isn’t enough. We must be actively engaged with trying to figure something out. Active means mentally active, not physically active – though physical participation may be important for certain skills. This does not mean a child is best left to figure things out on her own, as some suggest. Guided learning is by far superior, but we cannot merely pass information on to children. We have to encourage them to participate in their own education and self-discovery by engaging their “top-down” cycles of prediction and modeling.

Error Feedback – This is so crucial to learning that we cannot learn merely by being right. We MUST make mistakes in order to signal to the brain that this is something to learn, refine, and consolidate. Error feedback should be as immediate as possible (not a day or a week later), and emotionally neutral. Children, parents, and teachers should be taught that this is an essential part of learning, and how to give and receive it skillfully.

Consolidation – Consolidation is the process whereby the brain re-fires neurons that fired (and learned) throughout the day. This happens during sleep. Neurons that fired once or twice during periods of learning during the day are re-fired at rapid speed – increasing learning at the neuronal level. As new info and methods are learned, the resource-heavy cortex increasingly moves this learning to deeper, more unconscious circuits – increasing automaticity and freeing up executive functions for other tasks. These deep circuits can learn and refine themselves almost forever, increasing the thickness and myelination of dendrites, their length and connections, and even the available blood flow and resources to those neurons. Regular practice at increasing intervals is key.

Thanks to Stanislas DeHaene for such an excellent presentation of this material.

One of my favorite writers!