Aspects of the Theory of Artificial Intelligence

ion When the growth of an operational structure in the material is viewed in this way, it seems only natural that the filters estabINFORMATION IN RANDOM NETWORKS 29 lished to begin with will res pond to a broad cross section of the environment. The stabilizing influence, which formed an essential part of the original model, impose an economy of the activity so that there is a tendency for activity corresponding to irrelevant background to be eliminated in the course of time. Thus, with experience, the filters gradually become modified to separate the relevant from the irrelevant with increasing efficiency. Eventually, particularly und er the influence of education, language, and society, highly selective filters will become established which abstract particular objects or details which are of importance. Even then the response will in effect still be a Gestalt response in that a detail of the environment which calls for one response in one set of circumstances may call for a completely different response on another occasion. The ability to concentrate on the important details, while ignoring a background which is usually unimportant, is useful in that it makes for economy of mental analysis. When such abstract features are taken as a guide for action, and the rest of the environment is temporarily ignored, we have generalization. Generalization contains an implied assumption that some feature which has been the important criterion on a number of occasions will continue to be important, irrespective of changes in the rest of the environment. This is often, but of course not always, true. Once it has become possible to abstract individual details, there arises the possibility of a second mass of cells receiving as its input the abstracted information which comprises the output of the first, and thus enabling a second order of abstraction to take pi ace. Then, in pi ace of objects and details abstracted from the Gestalt we may have relationships between objects. Further developments along these lines would make it possible to take into account more and more abstract and complex relationships in a cascaded hierarchy of cell· masses. Given appropriate mental control, the transition from one stage to the next may occur naturally as experience grows, but a far more powerful influence in insuring a smooth transition is deliberate tuition in which the environment is carefully controlled at each stage.

[1]  R. Burger,et al.  Application of the Ion Bombardment Cleaning Method to Titanium, Germanium, Silicon, and Nickel as Determined by Low‐Energy Electron Diffraction , 1958 .

[2]  A. A. Lumsdaine,et al.  TEACHING MACHINES AND PROGRAMMED LEARNING, A SOURCE BOOK. , 1960 .

[3]  Donald M. MacKay,et al.  Quantal aspects of scientific information , 1953, Trans. IRE Prof. Group Inf. Theory.

[4]  W. McCulloch,et al.  The limiting information capacity of a neuronal link , 1952 .

[5]  F. Rawlins Design for a Brain—2nd Edition , 1960 .

[6]  H. Kay Teaching Machines , 1961, Nature.

[7]  Howard Hunt Pattee,et al.  X‐Ray Microscopy and Microradiography , 1958 .

[8]  Kenneth L. Artis Design for a Brain , 1961 .

[9]  N. Goodman,et al.  The Structure of Appearance. , 1953 .

[10]  Martin D. Davis,et al.  Computability and Unsolvability , 1959, McGraw-Hill Series in Information Processing and Computers.

[11]  Roman Jakobson,et al.  Structure of Language and Its Mathematical Aspects , 1961 .

[12]  W. Pitts,et al.  What the Frog's Eye Tells the Frog's Brain , 1959, Proceedings of the IRE.

[13]  K. Gödel Die Vollständigkeit der Axiome des logischen Funktionenkalküls , 1930 .

[14]  Hajo Bruining,et al.  Physics and applications of secondary electron emission , 1954 .

[15]  Peter H. Greene,et al.  An approach to computers that perceive, learn, and reason , 1959, IRE-AIEE-ACM '59 (Western).

[16]  K. Gödel Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I , 1931 .

[17]  D. Mackay The Epistemological Problem for Automata , 1956 .

[18]  Gordon Pask,et al.  The natural history of networks , 1960 .

[19]  A. M. Skellett The Use of Secondary Electron Emission to Obtain Trigger or Relay Action , 1942 .

[20]  W. Walter The Living Brain , 1963 .