ABSTRACT

Although the current intense interest in the properties of networks of simple computing units, as models of the brain or of specific cognitive functions, or as brain-like parallel hardware, has emerged relatively recently, development of the general theory underlying this type of parallel computation has been in train for many years. By far the largest contribution to this subject has been by those primarily interested in the brain. Dating from, perhaps, the days of mathematical biophysics (Rashevsky, 1938), the subject has taken from many other disciplines, such as symbolic logic (McCulloch & Pitts, 1943), cybernetics (Wiener, 1948), automata theory (see Arbib, 1964) and information theory (Shannon & Weaver, 1949). What I reviewed in this chapter is the work done in Edinburgh by Christopher Longuet-Higgins, Peter Buneman, and myself on parallel computation, stimulated by the analogy with holography.