A Neuroscientist Looks at Destin Sandlin’s Backward Bicycle

Being a neuroscientist, a kinesiologist, a learning and training researcher and a cyclist, a lot of my friends have sent me a link to the remarkable video where a guy learns to ride a special bicycle. The Smarter Every Day podcast presents us with some cool problems, ideas, and challenges to the way we think. In an episode that has gone viral, our host Destin Sandlin tries to ride a bicycle that has been engineered so the turning the handlebar to the left results in the front wheel steering to the right. You can see the episode here.

After a few attempts, he states that “your brain cannot handle it”. He challenges others to ride the backwards bike and even offers a $200 reward all over the world to the person who can successfully ride it. Destin decides to spend a few minutes every day learning to ride the bike, and finally manages, only to find that he can no longer ride a regular bike. The episode is really amazing for a lot of reasons, but Destin says some things about learning, about brains, and about bicycles that are not quite correct. It is not my desire to be a kill-joy, because the overall thrust of the episode is beautiful. However, I study brains, movement and learning and so he is playing in my play-ground: my playground, my rules.

“Knowledge ≠ Understanding”

Sure, sure. This is kind of obvious, except for a very important thing: The video conflates two very different kinds of brain activity, and our host talks about cognitive activity and motor activity like they are the same thing. You see, knowledge and understanding are descriptors about learning in the cognitive domain, and they represent different levels of cognition. A great modern example of this difference can be found in Einstein’s famous equation describing special relativity: E=mc2. Some people know the equation and can repeat it back to you. Some people even know that E is energy, and m is mass and c is the constant, or speed of light in a vacuum. A relatively (ha! GET IT?) small percentage of the population understands that the equation explains the equivalency of mass and energy, or a speed limit to light, or any number of other things, and fewer still understand why or how this is so, and in what ways the theory might be limited or incorrect.  But these are all cognitive processes, and there is more than one kind of learning, and bike-riding belongs to a different kind, called sensori-motor learning or procedural learning. Procedural learning, by its very nature, is difficult to describe. In other words, his comments don’t relate to learning a motor skill: they relate to learning facts.

There was a very famous brain patient named H.M. (Henry Molaison) who was studied extensively by Brenda Milner from McGill University in Canada. He had severe epilepsy which doctors attempted to cure by removing part of a brain structure called the hippocampus.  The hippocampus is associated with memory. After the surgery, his epilepsy was cured, but H.M. lost all of his short-term memory. If you introduced yourself to H.M. in a room, then left the room for a moment, and returned, he would not remember that you had met. If this sounds like a cool idea for a movie, Christopher Nolan already beat you to it with Memento.

Still,  Suzanne Corkin found something fascinating about H.M.. She and her team taught him a new motor skill, and very importantly for the discussion here, the skill was drawing a picture while looking in a mirror.  This is a challenging thing to do, and involves reversing certain parts of the motor program in conjunction with vision. People who practice this skill over time get very good at it, and can eventually draw just as well looking in a mirror as they can without. For Henry, the interesting thing is that his skill improved in mirror drawing over time, but he never remembered having come to the room, or having met the researchers, or the names of certain things. This story illustrates how knowledge aspects of learning are different from motoric ones.  In fact this difference is so clear and important, in the learning research community, people talk about training, which is about procedural things, and education, which is about information and ways of thinking.

Ultimately, riding the backward bike is a motor learning problem, and we have to be careful about using words like “thinking” or “knowledge” or “understanding” because motor learning is not about those things.

“Your brain cannot handle this”

I think this comment points out something important about motor learning, and may indicate the difference between rationally or explicitly knowing what the rule is, and being able to perform the task. In studies where experimenters explicitly tell people what the secret is (“the mouse cursor goes left 90° when you move the mouse forward”), the subjects actually end up doing worse, then eventually give up on the explicit strategy before just figuring it out by doing. Knowing the rule actually hinders the learning, perhaps in part because cognitively considering the information introduces more delay in responding to disturbances in balance, which is catastrophic. It also consumes important attentional resources in this very complex task.

“Suddenly my brain clicked back into the old algorithm”

If you do something in a unique context often enough, your brain will build a forward model for that context. In the Henriques motor control lab at York University, they have found that the model for an altered mouse cursor will last up to 24 hours with only 240 training reaches. As another example, you can walk in the water then on the beach, and your brain can adjust the amount of force for each step pretty much instantly. There does not appear to be any limit to how many models you can have. With continued practice and switching back and forth, he would eventually be able to do both equally well.

Our brains become quite adept at resolving this kind of backward mapping problem, as when we shave or brush our hair in a mirror: we must reverse depth (forward and back) but side to side remains the same. The idea for the bicycle is very similar to learning to reverse a car pulling a trailer, only you are messing with balance, not just direction. Eventually your brain just swaps one map for another. In fact, there are a great number of studies about having people learn to throw darts while wearing special prism glasses that swap left and right or up and down, and in my own PhD lab, we rotated forces or directions when people reach for something using a mouse or experimental robot arm.

Bicycles, Balance and Brains

Part of the issue may be the bicycle itself, and I puzzled over this for some time. The bicycle is not simply reversed left for right. The axis of rotation of the handlebars is further away from Destin than the axis of rotation of the wheel fork, and this may be important in what is happening here.

On your bicycle, there is a part called the stem: it is the part that juts forward from the headset, which is the part that turns the forks. If you have ever changed your stem, especially for a longer one, you know how important it is for steering.  I put a longer stem on my road bike and had trouble steering at first: I kept over-steering. The arrangement on the backwards bike creates a longer stem, which is the first strike against Destin, but not the most important.

In normal bicycle riding, to steer to the right, you extend your left elbow while flexing your right, but you also perform an action at each shoulder. On the left the action is called horizontal flexion, and involves bringing the more or less horizontal humerus bone closer to the midline of the body. On the right, you perform a corresponding horizontal extension at the shoulder, moving the humerus further away from the body midline. But to steer successfully to the right, you musn’t go too far, so the antagonist muscle groups work to do the opposite things to recover from over-steering: Left elbow flexion, left shoulder horizontal extension, right elbow extension, right shoulder horizontal flexion. Over the course of your life, your brain develops a model of how this linkage between the actions of the two arms work together, and a lot of this coupled action is handled by a brain area called the supplementary motor area (SMA). The patterns of movement in kinesiology are called kinematics.

To complicate things, how much force you require to flex or extend your elbow changes at each degree of flexion or extension. Gym equipment manufacturers know this, and they put cam-shaped pulleys in their machines so that the amount of force required at each degree of movement remains more or less the same.  Again, your brain makes a model of how much force is required and how fast it must be delivered. In kinesiology this is called dynamics. Over your life, you develop very clear relationships between the pattern of movement and the amount of force required to achieve the movement under certain conditions.

The backward bicycle creates a situation where the torque required  at the elbow and shoulder to turn the offset handlebar does not match the torque required to turn the stem.  Simply put, we need to push harder to turn, and regulating movements must also be harder. This is not a really big deal for your brain, since power steering reduces the amount of force you must apply in your car, and you can learn that pretty quickly, but all the same it is yet another thing.

One more complication. The human motor system is noisy and slow. Yes, we get feedback on movement using a sense called proprioception, but that information must compete with all kinds of other sensory information, including information from the inner ear, which senses linear and angular accelerations. All of this information must make it to your brain, be interpreted, and then generate a compensatory movement command.  The information often arrives too late to make a difference, so over time your brain generates a forward model.  This is like a plan that your motor system creates to anticipate the kinematics and dynamics of a movement and allows us to compare the actual consequences of movement with the expected consequences.  This is why it feels so weird when you take the extra step when you are carrying laundry up the stairs: the forward model generated forces and movements to accomplish that task. You have one model for walking across the floor, another for across the carpet and another for up the stairs.

Which brings us back to the backward bicycle.

If you point the wheel of a two wheel vehicle to the right, the inertia will cause the center of mass to fall to the left of the front wheel UNLESS you project your center of mass more to the right. This normally happens when we turn the handle bars to the right: the bent right elbow and straight left elbow means that your mass is now to the right. The issue here is that there is a basic motor pattern that even extends to walking: greater rightward pressure will result in rightward movement, whether walking, running or riding a bike. Equally, shortening the right side is also the way to turn to the right. Smaller steps on the right side is the way we turn to the right. The inner ear is facing a dire, fundamental conflict. Shortening to the right and center of mass to the right mean a rightward turn. Equally, if the vestibular system of the inner ear sense a fall to the left, it will project our body mass to the right, often by bending the right elbow slightly and horizontally extending the shoulder (think about balancing on a curb or parking block). On the backward bicycle, this fundamental, neurologically determined behaviour is denied and this is not bicycle learning, but fundamental neurological synergies. The challenge of riding the bike is bigger and more basic than Destin assumes, but it has nothing to do with the “way you think”. It has to do with basic physics, biomechanics and neural behaviour.

The Bottom Line

Destin comments early in the video: “It is a complicated algorithm: affect one part, it wrecks the whole thing.” With the axis of rotation of the handlebars in front of the axis of rotation of the headset, a change in the torque required at each joint, a conflict between the inner ear and the motor system, a reversed mapping for steering, altered physics for steering, the welder did not just change one thing, he changed everything.

He claims over and over that it is “a pattern” or “an algorithm” but it is more than that: it is the way physics works. It is the way basic neurophysiology works. It is the way biomechanics work. Still, as he demonstrates we can overcome patterns, even natural ones, and I think that is a reasonable message, and perhaps an encouraging one in a world where people do bad things and claim it is normal. We can still define the way we behave, even in the face of incredible pressure, and even if it seems to go against the way things are. And that is worth thinking about.


He says a couple additional things that are also worth considering:

“Once you have a rigid way of thinking in your brain, you can’t change it.” Well not with that attitude you can’t! But seriously folks, he demonstrated that he can, and we all know that it is possible.  Sometimes it takes remarkable effort.

“Any small distraction at all, even a cell phone ringing in my pocket would throw my brain back into the old control algorithm”. This highlights two things: the difficulty of the task, and the finite nature of attention.  When people stand still, they sway a little. This is the dynamic nature of balance. When people solve math problems while standing, the amount of sway increases. Imagine how fragile the control is when you are learning such a difficult task!


Leave a Reply

Your email address will not be published. Required fields are marked *