About 94 percent of American adults know how to ride a bike, according to an August 2013 YouGov survey. Once you learn how to ride one, you probably never forget (thanks, muscle memory!). But most of us probably don't think too hard about how we actually ride a bike -- how we hop on board and get our balance and start pedaling away.

If you were a professor of mechanical engineering, you might say that "the control strategies humans employ to balance a bicycle while riding are not well understood." That's the question that motivated Stephen Cain and his colleagues at the University of Michigan Ann-Arbor to figure out what separates novice bike riders -- those folks you see wobbling uncomfortably down the street on a CitiBike -- from the more skilled riders who go whizzing past them.

The researchers brought 14 cyclists into their lab. Half of them were experienced bicycle riders -- people who "identified themselves as skilled cyclists, went on regular training rides, belonged to a cycling club or team, [and] competed several times per year," according to the study. The other half were "non-cyclists" -- people who knew how to ride a bicycle, but only rode occasionally.

In a piece for The Conversation, Cain describes the experiment: "We conducted our experiments in a motion capture laboratory, where the riders rode a typical mountain bike on rollers." Think of rollers as a low-tech treadmill that you can put a bike on. "They require a bicycle rider to maintain balance by pedaling, steering and leaning, as one would outdoors," Cain writes.

Once the study participants familiarized themselves with the rollers (a more difficult task for the non-cyclists than for the pros), Cain and his colleagues had them ride while they measured all manner of things -- speed, steering angle, rider lean, pedaling force, etc.

Now, a word about the rollers before we move on to the results. Riding a bike on a roller set-up can be a tricky business. Even experienced riders can have trouble using them at first. "Rollers are viewed with suspicion, incomprehension or downright fear by many riders," a recent article on bikeradar.com states. YouTube is awash in videos of bikers crashing while on rollers, occasionally taking out their neighbors in the process. Here's a fun compilation video I found.

On the other hand, the researchers say this increased difficulty can be useful in a lab setting. "Riding a bicycle on rollers is also more challenging than riding overground, and therefore may be particularly useful in eliciting differences between skilled riders (cyclists) and less skilled riders (non-cyclists)," the study notes. And if you're in a laboratory situation and you need to make the kind of precise measurements that you can generalize out into a mathematical equation, rollers are the way to go.

Once Cain and his colleagues got all the riders accustomed to the rollers, they put them through their paces and started taking their measurements. In the end, they found that there was a surprisingly simple but significant difference in how newbies and pros stayed aloft on their bikes. "Expert riders achieve superior balance performance by employing smaller but more effective body movements and less steering," Cain wrote.

In short: Bad cyclists steer with the handlebars. Good cyclists steer with their bodies.

This chart from the study illustrates that pretty clearly. It measures steering power -- basically, how hard you're turning the handlebars -- for each of the 14 test subjects. The solid light grey lines are the non-cyclists, while the dotted black lines are the experienced bikers. As you can see, the non-cyclists are pushing on the handlebars a lot harder than the experienced riders.

Experienced riders, in other words, control their bikes by leaning to and fro, making slight adjustments to their center of mass. Less experienced riders try to do the same thing with the handlebars, but this can be counterproductive: One poorly-timed tug on the bars and you and your bike are tumbling.

"Expert riders are able to use body movements more effectively than novice riders, which results in reducing the demand for both large corrective steering and body movements," Cain concludes. But, he adds, "mysteries remain." Riding a bike on rollers in a laboratory is a fundamentally different process than weaving in and out of pedestrians and cars in the real world. Better ways to model cyclist behavior in the real world can help identify the riskiest riders and improve bike safety -- not just for the riders themselves, but also for everyone else around them.

More from Wonkblog: