uni'wissen 02-2012 ENG

25,000 nerve cells and measuring the interaction between the cells. This complicated undertaking can be achieved with the help of so-called micro- electrode arrays, which can be used to study ac- tive networks. Up to 1,000 external contact points are connected to micrometer-sized mea- suring points in the middle of the square recep- tacles, which are only a few centimeters in size. Each measuring point corresponds to an elec- trode that has contact with one or more neurons. Their signal is picked up by an amplifier. “We need a good sample of neurons that we can measure simultaneously,” says Egert. “The ar- rays give us a high spatial and temporal resolu- tion.” The scientists then observe the network to determine what forces are unleashed in it. “The neurons in the arrays spontaneously form groups and create local centers,” says Egert. “The role of the structure for the function is unclear. In the human brain, for instance, we have layers in which there are various types of cells, which are in turn located at very precise levels.” However, the example of bird brains illustrates that high performance is possible even without layered neurons. Simulation on the Computer Thus, when neuroscientists use computers to simulate such networks, they initially ignore the geometry. “There are various ways of building such neural networks,” says Egert. “We begin by concentrating on basic properties in order to study how the general rules of neural networks function, which we can then simulate in a realis- tic way on the computer.” When the simulations are successful, they serve as the starting point for new experiments. However, in order to create realistic simulations it is necessary to have a large mass of concrete data that must be deter- mined by way of experiments. What biophysical properties should the neurons have, which fre- quency should the stimulation be set at, where exactly should it be applied and in which phase, and how often and how long should the network be stimulated? The researchers are also investigating how an incoming stimulus spreads in the various net- work architectures. If it spreads evenly over the entire brain from a single point, the nerve cells in a particular area fire simultaneously and use up their resources all at once, much like in the case of an epileptic seizure. “We need areas that dif- fer from one another, that react to different kinds of excitation, that have a certain amount of ro- bustness, and whose nerve cells do not use up Brain slice on a microelectrode array: The research- ers stimulate various areas and measure the reaction of the nerve cells. “We concentrate initially on basic properties in order to study how the general rules of neural networks function” 6

uni'wissen 02-2012 ENG

Pages