Synergetic Computers for Pattern Recognition and Associative Memory

The purpose of my contribution is twofold: In this part I wish to present a brief preview on the contents of this volume and how its individual contributions are connected with each other. Then in the main part I wish to present my own model on the synergetic computer which allows for pattern recognition and associative memory. As I mentioned in the preface, it has been the goal of synergetics from the very beginning to bring scientists from neurobiology, physics, chemistry, computers, and other fields together and to provide a forum for interdisciplinary discussions. One of the main subjects of synergetics has been the relationship between pattern formation and pattern recognition. This tradition is followed up in the present proceedings where the first chapter deals with natural computational systems. Here we shall deal with the visual system (Koenderingk et al.), and (Baumgaertner et al.). (Here and in the following the quoted names refer to those authors who presented the paper.) It is wellknown from the cognitive sciences that we may complement an incomplete image, for instance by inserting automatically contour lines which are not present at all, but which are for instance indicated only by pieces of their ends. In the experiments by Baumgaertner et al. it is shown that under the above-mentioned circumstances quite evidently neurons start to fire which have been quiescent otherwise. Thus by means of the property of the total network of the visual cortex, there is a material substrate for this complementation effect. I do not doubt that these findings are of great importance for the construction of neural computers. An important problem in this context is figure/ground discrimination, on which a model is developed by Reitboeck et al. An abstract mathematical approach employing a variational approach is presented by Ingarden.