A fuzzy logic approach to neuronal modeling

W. Senn1;2, M. Jakob2, K. Wyler2, J. Kleinle2, H.-R. L uscher3, H. Mey2, L. M uller2 Abstract | Methods of fuzzy logic are commonly used to model a complex system by a set of rules provided by experts. But fuzzy methods can also be applied to the reverse problem: given the inputoutput behavior of a system, what are the rules which are governing that behavior. We introduce this reengineering method to neuronal modeling and, as an example, we analyze a biological neuron of the Hodgkin-Huxley type. Starting from measured data of the membrane potential and the sodiumand potassium-current through the membrane, we extract fuzzy rules which determine the time course of the membrane potential in terms of the actual values of the membrane current. These fuzzy rules allow not only to rebuild the behavior but also to understand the underlying membrane mechanisms. The approach o ers a new representation of neuronal mechanisms in terms of linguistic rules.