The problem with habits
Sometime around 2015, an engineer and YouTube personality named Destin Sandlin was given a bike that had been welded such that any turn of the handlebar moved the wheel in the opposite direction. It was meant as a joke. Sandlin, who said he had been riding bikes since age six, thought he could conquer it easily. Spoiler: He could not. When he tried, he looked like a small kid again, wobbling violently for a meter or so before spilling over. Sandlin grew frustrated. He took the bike home, and for five minutes every day, he practiced in his driveway. He expected he could learn it in a few days. It took him eight months. Then later, when he tried to ride a normal bike again, he could not do that either. The learning curve was much shorter—about 20 embarrassing minutes—but Sandlin still needed to concentrate hard on the task in order to accomplish it.
If it was so difficult to unlearn a motor skill, what should that tell you about learning it? For one thing, the complexity of riding a bicycle is often vastly underestimated. This is probably because almost all children ride bikes. Monkeys can ride bikes. The Moscow circus formerly employed bears that could ride bikes. Anyone who wants to learn to ride a bike can eventually learn to ride a bike. And then we never forget how. The skill can always boomerang back to us, as if we had evolved to do it, as if it was in our DNA.
Of course, the ability to ride a bike is not something we inherited. We’ve just pretty much got the learning of it down to a science. When you are just beginning, you carefully follow instructions. You concentrate hard on every movement, focusing on how to maintain balance while coordinating the churning of your legs. Eventually, as you get better at it, you can phase out those explicit strategies for a more automatized one. The instructions are called “declarative knowledge”, while automatization is said to be “procedural”. When your bike-riding gets to be procedural, you don’t have to think about; you just do it. The skill sinks in. It becomes, well, like riding a bike. Motor researchers have a phrase for this stepwise pattern of learning, which they call “scaffolding”. The early instructions support the construction of the skill until it is strong enough to support itself on its own. “The importance of repetition until automaticity cannot be overstated,” John Wooden once said, and coaches have adhered to this dogma ever since. Indeed, automaticity can be great: movements might come more quickly and easily, you are less fallible to distraction and your brain can perhaps be more efficiently used to solve other problems.
But there can also be a cost to automatization. That cost is the thing we call habit. In (Johns Hopkins neuroscientist) John Krakauer’s view, the reason Sandlin had such a difficult time with his experiment is not because he was following the wrong instructions or had never acquired the skill to ride a bike to begin with. It’s because he had also acquired a habit.
Habit is “the price you pay” for becoming automatic, Krakauer’s laboratory codirector, Adrian Haith, explained to me. They have found that habit is binary; once a motor activity gets to be habitual, additional practice doesn’t make it more habitual. There is no gradation. You either fall into the habit, or you don’t. So what is the problem? The problem, Haith says, is that habit is “an automatic retrieval of an action, even when you don’t really want it.” Even when it is wrong. Obviously, this can have a deleterious effect on skill. In Sandlin’s case, as soon as the handlebars were reversed, the habit he relied upon to ride a regular bike with ease became the wrong one for riding a bike with reversed handlebars. To adjust, he needed not only to relearn how to ride the new bike (by following a new set of declarative instructions). He also had to contend with the intrusion of a habit.
This, Krakauer argues, is not typically considered in investigations of motor learning or skill acquisition. The most famous model of the skill-learning curve, proposed by Paul Fitts and Michael Posner in 1967, involved three stages: (1) cognitive, or the declarative stage; (2) associative, or the exploratory stage; and (3) autonomous, which is self-explanatory. A fourth, offshoot stage might be habit, which could cast any frustrated performer, of any skill level, back to the drawing board. Where coaching can often miss the mark involves the need to proceduralize the right things for the right tasks. If baseball players spend their time hitting straight fastballs in batting practice, would the first sinker they face in a game not induce the sinking feeling of the reversion of handlebars on a bike?
So what does it actually mean to be skilled? Let’s say you are asked to answer a simple question in arithmetic. A straightforward question—what is 7 × 9?—can be solved a few ways. You could add nine to itself seven times. Or you could have rote memorized that 7 × 9 = 63. The latter method is taking advantage of the cached or automatized knowledge from weeks or years of flash cards and practice. You no longer need to follow the instructions; you just hop on the bike and do it. The answer arrives quickly to the tongue because your brain has consolidated those multiplication tables into an easily retrievable, yet inflexible, memory. This is a cognitive memory just as you might have a muscle memory. Krakauer remarked to me once that he has forgotten the numerals for his six-digit ATM code. This doesn’t leave him destitute—whenever he needs cash, he can simply go to the machine and his fingers will immediately, automatically punch the right formation on the keypad.
This sort of rapid, seemingly reflexive retrieval system works fine for some calculations. Krakauer and Haith, though, would contend that a mathematician is more likely to rely on both, procedural and declarative knowledge, in answering a long and intensive math problem, a math problem that would arguably result in observers remarking about her skill. They argue that a motor skill should be viewed the same way. The hallmark of any skill, they say, is being able to do the right thing, and quickly. But getting there is likely to require a more hierarchical representation, involving both forms of knowledge, braided for both quick retrieval and for the nimbleness that skill so often requires.
Adapted from the book The Performance Cortex: How Neuroscience is Redefining Athletic Genius by Zach Schonbrun. Copyright © 2018 by Zach Schonbrun. Reprinted with permission of Dutton, an imprint of Penguin Publishing Group, a division of Penguin Random House, LLC.