Please activate JavaScript!
Please install Adobe Flash Player, click here for download

uni'wissen 01-2015

amount of time. That’s what the mathematical analysis shows.” The solution to the problem is Mehring’s concept of structural learning: The brain doesn’t just learn the right muscle activation pattern for riding the mountain bike – it also learns what movement patterns that are important for riding any type of bicycle have in common. In the mathematical representation, all the movement patterns one needs to learn how to ride bicycles lie on a so-called manifold. A manifold may be thought of as being analogous to a plane or a curve in a three-dimensional space, but in reality these processes occur in a high-dimensional space. In order to switch from a mountain bike to a racing bike, a cyclist does not – mathematically speaking – need to search everywhere for the right movement pattern but only within the space or along the curve. “We and other research teams have found evidence to support this concept in a number of experiments.” corresponds to a point in a parameter space. The muscle activation pattern for riding a racing bike, which is of a different size and shape, corre- sponds to another point, because the brain needs to control the muscles of the cyclist differ- ently. “If I can ride the one bicycle and want to ride the other, that means I know the one point and want to find the other one.” But how does the cyclist learn how to ride the new bicycle? How does he find the right parameter point? It cannot be that he has to try out all pos- sible muscle activation patterns until he has found the right one: “Then it wouldn’t be possible to learn a new movement task within a plausible “When it comes to flexible movements, even a child can beat the best robots.” What’s going on inside my head?: In the method of electroencephalography, electrodes measure the neural activity on the brain’s surface. Photo: Thomas Kunz 1414 Specialtopic:ResearchintheUpperRhineregion