The von Neumann-Ortvay Connection

Annals Hist Comput (1989) 11:1&B-188 0 American Federation of Information Processing Societies

The von Neumann-Ortvay Connection DiNES NAGY, PETER HORV/iTH, AND FERENC NAGY

Three Hungarian scholars present a revised chronology of von Neumann’s life. The correspondence between Ortvay and von Neumann published here allows insight into the early development of von Neumann’s ideas about automata theory and “the computer and the brain.”

Categories and Subject Descriptor: K.2 [Computing Milleux]: History of Computing-people, hardware, theory. F. 1 [Theory of Computation]: Computation by abstract devices-automata, computability theory.

John von Neumann: A Revised Chronology

The chronology of John von Neumann’s career given by Bochner (1958; reprinted in Goldstine 1972) does not document the important events in von Neumann’s life prior to 1930. The present chronology adds to Bochner’s by supplying the dates for events in the early stages of von Neu- mann’s life.

1903 Born, Budapest, Hungary, De- cember 28

1913-1921 High school student, Luther- an Gymnasium, Budapest, Hungary @-year system). Graduated 1921 (Maturity Examination)

1921-1925 Student, University of Buda- pest, Hungary (mathemat- its major with physics and chemistry minors)

1921-1923 Student, University of Berlin, Germany (mathematics, physics, and chemistry)

1924-1926 Federal Institute of Technol- ogy, Zurich, Switzerland (chemical engineering)

1926 Ph.D. in Mathematics, Uni- versity of Budapest, Hun- wry

Privatdozent, University of Berlin, Germany

Privatdozent, University of Hamburg, Germany

Visiting Professor, Princeton University, Princeton, New Jersey, USA

Professor of Mathematics, In- stitute for Advanced Study, Princeton, New Jersey, USA

Gibbs Lecturer, Colloquium Lecturer, B&her Prize (American Mathematical Society)

Scientific Advisory Commit- tee, Ballistics Research Laboratories, Auberdeen Proving Ground, Maryland, USA

Navy Bureau of Ordnance, Washington, D.C., USA

Los Alamos Scientific Labo- ratory (AEC), Los Alamos, New Mexico, USA

Director, Electronic Computer Project, Institute for Ad- vanced Study, Princeton, New Jersey, USA

D.Sc. (Honorary), Princeton

Annals of the History of Computing, Volume 11, Number 3, 1989 l 183

D. Nagy et al. l The von Neumann-Ortvay Connection

1947-1955

1949-1953

1949-1954

1950

1950-1955

1950-1957

1951-1953

1951-1957

1952

1952-1954

1953-1957

1954

1955-1957

1956

University; Medal of Merit (Presidential Award); Dis- tinguished Civilian Service Award (U.S. Navy)

Naval Ordnance Laboratory, Silver Springs, Maryland, USA

Research and Development Board, Washington, D.C., USA

Oak Ridge National Labora- tory, Oak Ridge, Tennessee, USA

D.Sc. (Honorary), University of Pennsylvania and Har- vard University

Armed Forces Special Weap- ons Project, Washington, D.C.; and Weapons System Evaluation Group, Wash- ington, D.C., USA

Member, Board of Advisors, Universidad de 10s Andes, Columbia, South America

President, American Mathe- matical Society

Scientific Advisory Board, U.S. Air Force, Washington, D.C., USA

DSc. (Honorary), University of Istanbul, Case Institute of Technology, and Univer- sity of Maryland

Member, General Advisory Panel on Atomic Energy, Washington, D .C . , USA (Presidential appointment)

Technical Advisory Panel on Atomic Energy, Washing ton, D.C., USA,

DSc. (Honorary), Institute of Polytechnics, Munich, FRG; Vanuxem Lecturer, Prince- ton University, Princeton, New Jersey, USA

U.S. Atomic Energy Commis- sioner (Presidential appointment)

Medal of Freedom (Presiden- tial Award), Albert Einstein Commemorative Award, Enrico Fermi Award

Died, Washington, D.C., USA, February 8

Von Neumann, Ortvay, and Automata Theory

It is not surprising to find that von Neumann was interested in automata theory. As Goldstine (1972) points out, von Neumann’s interest may have been inspired by his early interest in logic and strongly enhanced by his newer interest in computers and neurophysiology. His interest in computers may have arisen in part from his reading of the very important work by McCulloch and Pitts on neural networks (McCulloch and Pitts 1943). Goldstine (1972) remarks: “Von Neumann worked on his theory of automata alone. This was in rather sharp distinction to most of his later work where his practice was almost always to work with a col- league. Very possibly he wanted his automata work to stand as a monument to himself-as in- deed it does.”

We provide below some important documents concerning the early development of von Neu- mann’s ideas about automata theory and the “computer and the brain.” Von Neumann’s ex- tended correspondence with the Hungarian phys- icist Rudolf Ortvay (Director of the Theoretical Physics Institute of the University of Budapest) offered him a stimulus and sounding board for working out many of his ideas on the subject of automata.

Ortvay was a major figure in Hungarian science. He devoted a great deal of energy to the organization of scientific life in Hungary, fol- lowed the main trends in a variety of branches of science, and helped many colleagues by his suggestions. The Ortvay Symposia became important international meeting places.

The Ortvay-von Neumann letters, written in Hungarian, are housed in the Manuscript Divi- sions of two libraries: the Library of Congress, Washington, D.C. (John von Neumann Papers) and the Library of the Hungarian Academy of Science, Budapest. Sixty letters from the Ort- vay-von Neumann correspondence have recently been published in Hungarian by Nagy (1987). The texts of the selected letters given below in En- glish translation are based on the Hungarian version printed in Nagy’s book.

The von Neumann-Ortvay Correspondence

Von Neumann to Ortvay 29 March 1939 Princeton, New Jersey

184 l Annals of the History of Computing, Volume 11, Number 3, 1989


Dear Rudolf,

. . . I am deeply interested in what you write about biology and quantum theory and the subject is very sympathetic to me, too. Your remarks about the anatomy of the brain are especially in- teresting. I too believe that the possibility should be taken seriously that processes inherently re- lated to life cannot be depicted geometrically, i.e., cannot be described in space, because of the spa- tial location of our physical body. Within this location, all processes can only be approximated. If we go beyond a certain point (which is perhaps very vaguely interpreted, as can be seen from the following examples), the same difficulties, con- tradictions and pseudo-problems arise as when we examine the simultaneity of remote events in the special relativity or when we attempt to measure p and q at the same time in quantum mechanics. NB. I have been thinking extensively, mainly from last year, about the nature of the “observer” fig- uring in quantum mechanics. This is a kind of quasi-physiological concept. I think I can de- scribe it in an abstract sense by getting rid of quasi-physiological complications . . . .

Ortvay to von Neumann 25 December 1939 Budapest

Dear Johnny,

. . . I would like to revert to the issue I have already brought forward many times before, namely to the theory of operation of the brain. Somehow I would like to justify my stubbornness. As I see the situation, the question, within certain limits, is ripe for a solution. Today we have the.men- tality of dealing successfully with the problem, and we are not alone. Currently physicians are not leaning towards comprehensive theories, and they almost entirely lack the capability which would allow them to see the simple and theoretical structure of a complicated complex. Physi- cists and mathematicians are better trained to achieve this. I think only an outsider could give an impetus in this regard, of course in co-operation with physicians and physiologists!

I would be most interested to have your further thoughts on the psychophysical parallelism. NB. Is there any literature relevant to this point of the anatomy of the brain? . . .

The situation is somewhat similar to the cir- cumstances surrounding hereditology at the time when I was a medical student; Mendel and the new explorers, such as Corens, were neglected and nobody paid attention to their work. The discipline was nothing more than a muddy, blurred jumble of data.

Sincerely,

Johnny * * *

Ortvay to von Neumann 3 May 1939 Budapest

Dear Johnny,

. . . a very loose relationship, which reminds me of the essential standpoint of classical physics or “realism”-that a “closed” physical system is possible. The present system of the quantum mechanics states, however, that such a system is not possible without an essential reference to the observer. Taken in the philosophical sense, this would be an idealistic concept of the world. It seems, on the other hand, that mathematics should deal with logical problems in depth. . . .

Yours truly,

Rudolf * * *

After having discovered that certain units were relevant to genetics, one of the nicest chapters in the history of biology began. It is still obscure what a gene is and we are only beginning to discover how it exerts its effects. (Russenkaninchen, Ephestia, see the Naturwissenschaften and Na- ture reports on the genetics congress in Edin- burgh.) But the proper connection between units, genes and detectable characters made possible the emergence of a new discipline. This happened in an area where the crucial scientific problem was at the time, and not many issues remained in the dark. I think a similar statement can be made about the problem of the operation of the brain. On the one hand, we have the psychical operation-the sensations with their many complex subjects, the mental concepts, and the relation- ships among them, such as judgments. These are arranged somehow, even if in a coarse manner. On the other hand, there is the brain with its characteristic set-up. I think you have your own picture of it: the simplest nervous system con- sists of two neurons; one of them receives an ex- ternal effect, the other is connected to a motory organ. These two neurons are capable of affect-



ing each other. The more highly developed nervous system is different from this simple reflex arch. First of all, there are numerous receptory and motory neurons; on the other hand, there is also a complete system of transmitting neurons between them. These transmitting neurons are connected to more than one receptory and motory neuron so that they can receive effects from more than one cell and can transmit actions to several cells. Obviously this is a highly sophisticated switchboard system, the scheme of which we do not know. The very task would be to propose theorems about this or to devise an adequate model for certain substantial features. What is the sensation of a nerve? How does it relate to the quality of the senses? For the time being, such questions should be excluded from the investi- gation. The simple element of a sensation, e.g., a focussed sensation of light, corresponds to a certain state of neuron connected to the first sen- soring cell. The following formulas-table, chair, man, etc.-are the simple (concrete) concept which we denote by nouns. In fact, it is these concepts and not the elements of sensation which form part of our everyday thinking; the elements of sensation are the results of a somewhat forced ab- straction. I personally think that these concepts could be adapted to the sensational state of the first neurons in a way in which a complex of neurons-a single cell within the second system-is activated. If the first complex is broad enough, but varies within the limits of one class, we can assume that always the same cell is activated. A physical analogy to this would be, e.g., the count- ing equipment of Forro and Bartnoty. This de- vice is triggered only if an electron passes through two or three counter valves consecutively. This could serve as a model for the concepts of “two” and “three.” Similarly, more abstract conceptual systems would be based on the system of concrete concepts in various stages. I want to mention here that more abstract concepts can be applied only if they are connected ‘lwords,” obviously simple formulas in the second system. I don’t want to exclude the possibility that in this way we could arrive at a sensible theory of the operation of the brain in the near future. Perhaps we can even arrive at a very important theory which would correspond more or less to Mendelism. A very similar theory is the combinatorial arrangement of spectra, inhibition rules, and vectory theory, while eliminating the quantitative statements. Of course all this will require very deep thought

and the details can only be elaborated in connection with empirical material and with the development of the relevant theory. For the time being the situation resembles the days of Darwin and Weissman; they suspected certain units to be relevant for genetics, but they could not make them concrete.

Yours truly,

Rudolf * * *

Ortvay to von Neumann 30 March 1940 Budapest

Dear Johnny,

. . . I do not deal with the physiological problems in detail, since I don’t see a chance to proceed with them. Of course I don’t think that the mechanism of the brain is very similar to computing equipment consisting of electronic valves, but I do believe that the brain is a complicated and very specialized piece of equipment which may be very simple in terms of basic principles! The atom also is a rather complicated system. A histological examination clearly shows that the brain is a highly differentiated organ. Its complexity stands in sharp contrast to the truly primitive ideas physiologists and physicians have about it. We see elsewhere, too, that there are other very highly differentiated systems such as the chromosome, for example, or the gene system which governs hereditology, or the chemical processes that oc- cur in the organism. . . .

When it comes to the “mechanism” one is in- clined to think of a machine consisting of rigid parts, which is quite obviously too narrow an approach. . . .

Yours truly,

Rudolf * * *

Von Neumann to Ortvay 13 May 1940 Winslow, Arizona

Dear Rudolf,

. . . To the subject of physiology-chemistry- quantum mechanics I wish to add the following:



You stressed that our modern-day vision and concepts are too complicated when it comes to the description of the brain. I cannot accept that a theory of prime importance which describes processes everybody believes to be elementary, can be right if it is too complicated, i.e., if it describes these elementary processes as horribly complex and sophisticated. Thus, I am horrified by the present-day quantum chemistry and especially biochemistry. I cannot rationalize this view any better than with my opening statement.

social sciences. Besides, we already touched upon this issue in our correspondence. . . .

Yours truly,

Rudolf * * *

Ortvay to von Neumann 29 January 1941 Budapest

I cannot propose a better concept, but I believe that we must find a new terminology and new formulas (i.e., models) in all these fields before we can proceed any further.

Dear Johnny,

. . . Once again I read through your paper on

The Mendel theory, to which you refer as an example, is rather typical of this situation.

I suppose that I can interpret your remarks about the complicated operation of the brain in the light of your wish to underline the difficulties of the present situation. There is absolutely no reason that this must ultimately be so. A bad vision (model) can easily create the impression of a terribly complicated situation, when later on, with the aid of a smarter approach everything can be settled in a simple manner. . . .

Sincerely,

Johnny * * *

Ortvay to von Neumann 30 May 1940 Budapest

Dear Johnny,

. . . I think that it will be possible to find a proper and simple theoretical approach to the operation of the brain. The teleological concept of variation principles was widely spread in the eighteenth century and it was used in a very naive way at times so that it became discredited later. See F. Klein, Entwicklung der Mathematik im XIX Jahrhundert, p. 199. . . .

Another remark: don’t you think that classical mathematics developed along with the rational- ization of the physical world and made this development possible. By now, mathematics far ex- ceeds this framework (set, theory, topology, mathematical logic), and can be expected to con- tribute to the rationalizaton of psychology and

games and I-would be interested in having your new results published, preferably in a popular manner. I liked your paper very much at the time, and it gave me hope that you might succeed in formulating the problem of switching of brain cells if I succeed in drawing your attention to it. I think only an outsider, not a physician, can perform this task-or at least provide the decisive impulse. The problem seems to be this: the brain can be con- ceived as a network with brain cells in its nodes. These are connected in a way that every individ- ual cell can receive impulses from more than one other cell and can transmit impulses to several cells. Which of these impulses are received from or passed on to other cells may depend on the state of the cell, which in turn depends on the effects of anything that previously affected this partic- ular cell. It may perhaps be sufficient that a cell has a limited number of potential states. (Al- though genetics shows us how differentiated a cell can be, since we know that there are hundreds of genes within a chromosome.) The actual state

of the cells (which I conceive as being numbered) would characterize the state of the brain. There would be a certain distribution corresponding to every spiritual state and that state would be relevant to every reaction, e.g. the way in which a stimulus is transmitted from a nerve. This model may resemble an automatic telephone switchboard; there is, however, a change in the connec- tions after every communication. Perhaps the ever-refining technologies in the sw‘itchboard equipment would provide a facile analogy.

The big trouble is that the physicians who know the facts are so hostile towards a more abstract way of thinking. . . .

Yours truly,

Rudolf



* * *

Ortvay to von Neumann 16 February 1941 Budapest

Dear Johnny,

. . . Now I wish to mention another matter: these days everybody is talking about organization and totality. Today’s computing machines, automatic telehone switchboards, high-voltage equipment such as the cascade transformer, radio transmit- ter and receiver equipment, and also industrial plants or offices are technical examples of such organizations. I think there is a common element in all of these organizatons which can be the ba- sis for an axiom. I don’t know if there have been any attempts in this direction. This interests me for the following reason. I believe that once it is possible to identify clearly the essential elements of an organization such as these, this would give

us a survey of possible forms of organization. This, then, would facilitate our understanding of such systems as the brain.

Yours truly,

Rudolf * * *

References

Bochner, S. 1957. “John von Neumann.” National Academy of Sciences Biographical Memoirs 32, pp. 438-457.

Goldstine, H. H. 1972. The Computer from Pascal to von Neumann, Princeton, Princeton University Press.

McCulloch, W. S., and W. Pitts. 1943. “A Logical Cal- culus of the Ideas Immanent in Nervous Activity.” Bulletin of Mathematical Biophysics 5, pp. 115-133.

Nagy, F. 1987. Neumann Jknos e’s a r’magyar titok” [John von Neumann and the “Hungarian Secret”], Budapest, OMIKK.


Documents

The von Neumann-Ortvay Connection