Studies in Living and Automatic Systems
The common bond between the student of living systems and the engineer in the study of pattern recognition is the general theme that binds this collection of papers together. The two types of systems – living and automatic – certainly share certain underlying principles, and it is these that should be uncovered. Without making simplistic or mechanistic analogies or building crude models relating one type of system to the other, underlying the basic principles they share will remain a source of insight of high potential.
In order to achieve the broadest cross-communication between the two groups, “pattern” is here defined in the most general way recognizable by the two groups. More than this, such a definition is likely to lead to the discovery of more general principles and, in the long run, to more practical results. For example, the limited success obtained in the narrowly defined field of two- dimensional visual recognition by automatic means may be due to its very conceptual narrowness – a test as to whether a “thing” fits an idealized templet largely ignores the intrinsic interrelationships of its parts. A wider study of these interrelationships – the general identification of patters as a series of relationships – not only is likely to lead to broader principles but will eventually lead to finer solutions of particular problems as well, including that of two-dimensional visual recognition.
However general the outlook and assumptions of the contributors, they report here on their own specific research interests and firmly base themselves on empirical results. Although they do not pretend to cover the whole area of this fast-growing field, among them they are able to skate out its present boundaries.
Their thesis – that engineers could build better machines if they used more of the principles of living systems and that students of living systems could benefit by using certain concepts of the engineers – is illustrated by the unity of approach that emerges from the variety of researchers reported on in the individual chapters. The first chapter, “Some Psychological Aspects of Pattern Recognition,” by Paul A. Kolers, reviews some of the concepts germane to human pattern recognition and shows why some typical models cannot be correct. “What We Do When We Speak,” by Samuel Jay Keyser and Morris Halle, describes the rules of syntax and phonology in the recognition of speech. The other chapters are “Neurophysiology of the Visual System,” by Shin-Ho Chung; “Stimulus Transformations in the Peripheral Auditory System,” by William M. Siebert; “Handwriting Generation and Recognition,” by Murray Eden; “Character Recognition in an Experimental Reading Machine for the Blind,: by Samuel J. Mason and Jon K. Clemens; “Contextual Understanding by Computers,” by Joseph Weizenbaum; and a final summarizing chapter by Murray Eden, “Other Pattern-Recognition Problems and Some Generalizations.”