Potential Significance to Neurophysiology of Design Algorithms for Digital Computers
by R. D. Traub, J. P. Roth
A set of algorithms has been developed within the computer industry that aids in the design and comparison of large switching circuits. To apply these algorithms to the nervous system in a realistic way, it is necessary to find neurons where (a) the electrophysiological data exist to allow construction of a plausible model; and where (b) the behavior of the neuron is not dependent on its distant past, so that the cell may be approximated by a switching circuit. It is also necessary to find areas of the nervous system whose structure resembles that of so-called regular designs; a regular design is a logic circuit in which feedback loops are constrained to take certain simple forms. In cases where local feedback loops are an intricate part of the design (as is true in virtually all areas of the mammalian central nervous system), this constrains one to analyze the behavior of the area in question only over time intervals sufficiently short so that such feedback loops can be ignored. Certain areas, such as the retina, are excluded because the individual neurons behave nondiscretely and are intimately coupled together (i.e., the retina does not resemble a regular design).