“I can do a bit of mind reading when I teach people to juggle,” Wolf said, “and I can tell you’re thinking of this as a pattern.” Just throw to the corners, she reiterated. Don’t think of the overall pattern you’re throwing; just throw. Wolf didn’t want me to think of the catches, either; if I just kept throwing to the corners, my hands would move to where they needed to be for the catch.

“The key to learning juggling,” she said, “is not thinking.”

The problem with beginners is that they’re always thinking about themselves doing the skill. When we do try to think about an “overlearned” skill like walking, we’re likely to perform worse, under the theory of “reinvestment,” as proposed by the motor-learning expert Rich Masters.

People who have had a stroke, for instance, often suffer from an “asymmetrical gait,” or a limp. They must relearn how to walk, but because they’re self-conscious of how they now walk, they think about the mechanics of walking, which only makes it look more mechanical. To learn to walk well, they’re going to have to learn implicitly. “The trick,” as Masters has described it, “is getting people to learn to move without knowing that they’re learning.”

When we become skilled at something, it becomes automatic. We don’t have to think much about it, because our brain, running on virtual autopilot, is constantly making predictions—and most of its predictions are true.

As Pablo Celnik, the genial Argentine-born director of Johns Hopkins University’s Human Brain Physiology and Stimulation Lab, told me, the brain does this for efficiency’s sake, but also because of an inherent time lag. “Your brain receives feedback about what you’re doing, and that takes time—about 80 to 100 milliseconds,” he told me. “We live in the past. Whatever we see now is actually about 100 milliseconds ago for the motor domain.”

These predictions help us get through daily life. When they fail, we look for explanations. We trip on the sidewalk, our brain gets this news 100 milliseconds later, and we accusingly stare at the offending crack. The surprise violated our model. But when we try to tickle ourselves, nothing happens, because we already know what it’s going to feel like. Our cerebellum has “canceled” the sensory input, suppressed neurons. There’s no surprise; the model is intact.

When you first get on an escalator that has stopped working, you gingerly take a few steps. You may even “feel” motion. That’s because your brain has trained itself, through many repetitions. It’s ready for the escalator; it’s predicted it. We know, in our heads, that it’s broken, but we can’t help thinking, in our bodies, that it’s not.

Slowing Time

Juggling, I soon learned, wasn’t really the skill I thought it was. Like many beginners, my mind’s eye of juggling was what’s known as a shower pattern—three objects being passed in a clockwise semicircle. But the shower pattern is much harder than the “cascade,” the most common form of multi-object juggling. In the cascade, objects cross each other and land in the opposite hand. Traced out, it looks like a figure eight tipped on its side.

I’d also envisioned that jugglers were tracking each object in flight, which is precisely what beginners try to do. When my daughter gave it a go, her head was wildly snapping as she tried to monitor each scarf.

Courtesy of Tom Vanderbilt

But, as Heather Wolf had shown me, juggling is less about throwing individual objects than throwing to a pattern, like tossing to a little algorithm in the sky. It’s little wonder so many noteworthy mathematicians, from Claude Shannon* to Ronald Graham, were drawn to juggling.

In juggling, unlike most sports, you don’t actually want to keep your eyes on the balls. Jugglers look to the apex of where things are thrown—that external focus again—and only ever have a peripheral sense of all those objects in flight. This has been confirmed by studies in which most of a juggler’s vision was blocked, except for a thin slice up near the parabola of the throwing arc, and they juggled just fine. Good jugglers can do it blindfolded.