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

uni'wissen 01-2015

Prof. Dr. Carsten Mehring has served as head of the neurobiology and neuro- technology working group at the Bernstein Center Freiburg and the Institute of Biology III of the Univer- sity of Freiburg since 2013. He earned his doctorate in Freiburg in 2003 and then worked in Freiburg and at the University College and the Imperial College in London/England. From 2002 to 2007 he was a member of the Junior Academy of the Heidelberg Academy of Science and Humanities. He is part of the University of Freiburg’s Cluster of Excellence BrainLinks– BrainTools. Photo: Thomas Kunz Further Reading McIntosh, J. / Mehring, C. (2013): The influence of transcranial alternating current on the timing of decision making. files/2013/0642_FI.pdf Kobak, D. / Mehring, C. (2012): Adaptation paths to novel motor tasks are shaped by prior structure learning. In: Journal of Neuroscience 32/29, pp. 9898–9908. doi: 10.1523/JNEUROSCI.0958-12.2012 Braun, D. A. / Aertsen, A. / Wolpert, D. M. / Mehring, C. (2009): Motor task variation induces structural learning. In: Current Biology 19/4, pp. 352–357. Exchange in a Network Mehring receives crucial input for his research from other scientists in “Neurex,” the trinational neurosciences research network on the Upper Rhine. The network includes around 100 labs at the Universities of Freiburg, Basel, and Stras- bourg. “Neurex brings together brain researchers from various backgrounds and areas of speciali- zation. In addition, the specialties of the neuro- sciences research groups at the universities in Freiburg, Basel, and Strasbourg complement each other very well. Mehring is a member of the external advisory board of the Erasmus Mundus doctoral program “Neurotime,” which includes the three previously mentioned institutions as well as the Universities of Amsterdam/Netherlands, Ban- galore/India, and Jerusalem/Israel. In a new study, Mehring and his team demon- strate how people improve movements and plan them in advance. The test subjects held a robot arm with their hands and moved it to control a mouse pointer. The task involved tracing a path on a monitor with the mouse pointer. The researchers tested how long it took the partici- pants to trace the path. The longer the test subjects had practiced beforehand, the faster they finished the task. In a second test, most of the path was hidden: The only visible section was the first few centimeters in front of the mouse pointer. “The test subjects who had prac- ticed beforehand made more efficient use of the information provided to them. Those who had learned the movement could include information on the future course of the path and thus adapt their movement early on.” At the moment, Mehring is working on a new technique for modifying the activity of particular neural networks. “We are developing a closed- loop simulation system that continually measures brain activity and adapts the electrical stimula- tion to it in real time.” The researchers hope it will help them find causal relationships between neu- ral activity and behavior. Despite these advance- ments in brain research, however, it will still be quite some time before robots learn how to ride bicycles like humans. Generalization, from a mathematical perspective: The brain doesn’t just learn which muscles it needs to activate to ride a particular kind of bicycle (points on the curve) – it also learns what movement patterns for riding any type of bicycle have in common (curves). This is why anyone who has ever ridden a bicycle can generally ride any model – whereas other move- ment patterns, such as those that are important for surfing (point next to the curve), require a new learning process. Photo: PRILL Mediendesign, apops, steamroller, joël BEHR (all Fotolia), Montage: Kathrin Jachmann 15uni wissen 01 2015 15uni wissen 012015