REVISITING WARREN S. McCULLOCH

Stuart A. Umpleby
Department of Management, George Washington University

Warren McCulloch was one of the founders of the field of cybernetics. To understand the field of cybernetics, it is essential to understand his interests and career, even though his contributions are not currently as well-known as those of Norbert Wiener and John von Neumann. McCulloch’s work was more in philosophy and neurophysiology than in math and engineering. However, his approach to cybernetics continues to generate new ideas. McCulloch began his study of the brain using a realist epistemology (the world is primary, perception is secondary), but his research created the foundations for a constructivist epistemology (perception is primary, the world is constructed). Constructivist epistemology is now influencing many fields, particularly the social sciences. Indeed, cybernetics is changing our conception of science itself, (Mueller 2016; Riegler & Mueller, 2014).

McCulloch’s background is surprising for a person associated with what is thought to be a technical field. He attended Haverford College and studied philosophy and psychology at Yale. He received an MA in psychology from Columbia and his MD from Columbia University. He interned at Bellevue Hospital in New York and worked at the Laboratory for Neurophysiology at Yale University from 1934 to 1941 before moving to the Department of Psychiatry at the University of Illinois at Chicago.

McCulloch wanted to understand how the human brain works. He used questions to explain his research. For example, “What is a number that a man may know it and a man that he may know a number?” “Where is fancy bred?” “How do we know universals?” “What is in the brain that ink may character?” And how does perception occur? (McCulloch, 1965)

Previously our knowledge of the brain was based on reflection, logic and criticism. McCulloch used a method he called experimental epistemology (Abraham, 2016). He took ideas from philosophy and tested them using experiments in neurophysiology. For example, experiments with the nervous system of frogs showed that the eyes of a frog did not transmit an image of the world to the brain. Rather, different layers of cells on the retina of the frog sent different types of messages to the brain. Some cells would respond to a shadow passing over the frog. Some would track the movement of a black speck across the frog’s visual field. The eyes did not deliver data to the brain to use in decision-making. Rather, the data that was delivered to the brain initiated an action. With knowledge of how the brain functions, a physiologically based conception of knowledge and brain activity could be developed.

McCulloch combined philosophy, theory and practice and used each to reinforce the others. Theories guide practice and philosophy guides the construction of theories. In addition McCulloch used science to test ideas from philosophy. Most academic work is limited both by the field of study and by types of knowledgephilosophy, theory and practice. Using all three types of knowledge aids in overcoming obstacles in the development of science and helps in understanding how knowledge fits together, how it grows, and hence how to make a contribution.

Two very important articles were published in 1943. McCulloch and Walter Pitts published “A logical calculus of the ideas immanent in nervous activity.” The article began with the observation that the network of neurons in the brain acts in a manner that we experience as an idea. How does that happen? We would like to create a description of the process in the form of a formal axiomatic theory, hence a logical calculus (McCulloch, 1965).

The same year Arturo Rosenblueth, Norbert Wiener and Julian Bigelow published “Behavior, Purpose and Teleology” (Rosenblueth, 1943). When we observe behavior, it often seems purposeful. How do purposes arise? A theory of teleological mechanisms was needed.

These articles attracted considerable interest. There seemed to be the possibility of creating a theory of how the brain functions. A series of conferences sponsored by the Josiah Macy Jr. Foundation were held in New York City between 1946 and 1953. McCulloch recruited the participants, arranged the order of presentations and chaired the meetings. The first five conferences were characterized by conflicts, hurt feelings and misunderstandings. McCulloch said he learned that outside their fields, scientists need to be addressed as lay persons. There were no proceedings from the first five meetings. The proceedings of the second five meetings were edited by Heinz von Foerster, Margaret Mead and Hans Lukas Teuber (Pias, 2003).

After the Macy conferences both Warren McCulloch and Norbert Wiener were at MIT. Engineers were becoming interested in research on the brain. John McCarthy organized a workshop at Dartmouth College in the summer of 1956 (Wikipedia). The field of Artificial Intelligence (AI) was invented, and it separated from cybernetics. The biologists, psychologists and philosophers, including McCulloch, wanted to study the brain in order to understand cognition. The engineers, including Marvin Minsky, McCarthy and Oliver Selfridge sought to advance the field by building machines, such as a computer that could play a game of checkers.

Two strategies were considered for creating an intelligent machine. One strategy, favored by the engineers, was to program a computer to follow a procedure created by a human being. The second strategy was to create a large network of elements that would organize themselves to accomplish the task, by learning from interactions with its environment. At the time no one knew how to implement the second strategy, so people drawn to AI chose the first strategy.

In the late 1950s Heinz von Foerster, who was the youngest participant in the Macy meetings, took a sabbatical to study biology with Arturo Rosenblueth in Mexico City. He returned to the University of Illinois in Urbana-Champaign (UIUC) and established the Biological Computer Laboratory (BCL) in 1958. Several conferences on self-organizing systems were held in the early 1960s (Yovits & Cameron, 1960; Foerster & Zopf, 1962) in an effort to understand networks and adaptation as part of the second, learning strategy in AI. Recent work in AI has used a multi-layered or “deep learning” strategy.

Wiener died in 1964, and McCulloch died in 1969. Due to rapid growth of computer technology, interest in solving practical problems, and science fiction, cybernetics had become associated in the popular imagination with engineering. So, von Foerster created the term “second order cybernetics” in 1974 to refocus attention on the McCulloch and Macy agenda of biology, cognition, epistemology and purposeful behavior. First order cybernetics told researchers how to create control mechanisms, second order cybernetics used the theory of control mechanisms to understand how an observer functions (Foerster, 1973, 2003; Mueller, 2016; Umpleby, 2016).

The new movement within cybernetics did not catch on initially in the US. Faster computers were being manufactured. Computer networks were being developed. Many useful applications were being created. Businesses were increasingly using computers for engineering and management tasks. Research grants were available for computer-related projects. But doing research on cognition depended on a philosophical and neurophysiological perspective unfamiliar to most US researchers and funding agencies (Riegler & Mueller, 2016; Kline, 2015).

BCL and its study of cognition encountered two unanticipated obstacles. The first was the Mansfield Amendment, which Congress passed in an attempt to cool the campuses in the late 1960s, when students were protesting military research on campus due to opposition to the Viet Nam war (Umpleby, 2003). The Amendment required military grant recipients to explain how their research was related to a military mission. Foerster said the research on cognition was not applicable to a military mission, so BCL was no longer eligible to receive research support from the Pentagon. There was hope that a new program, Research Applied to National Needs (RANN) would support the research that was no longer supported by the Pentagon. But the people at RANN did not know the history of cybernetics research or how to continue it.

The second obstacle to the continuation of research on McCulloch’s agenda was the lack of familiarity of American scientists with the history of philosophy. In continental Europe students study the history of philosophy for two years in high school. They are required to pass an exam on the ideas of the leading European philosophers. Americans have no similar training in philosophy. In his book Cybernetics (1948) Wiener suggested that the first industrial revolution used machines to replace human muscle power. The second industrial revolution would use machines to replace human mental activities. (The first industrial revolution concerned matter and energy. The second concerns communication and regulation.) However, there are deep philosophical and theoretical differences between these two revolutions. The history of the philosophy of knowledge can be interpreted as a debate between the follows of Plato and the followers of Aristotle. The followers of Aristotle include physicists and engineers. The followers of Plato include psychologists and philosophers (Gregory, 1986). Hence, Americans were better prepared by their education to deal with the first industrial revolution than the second industrial revolution.

Before, during and after World War II many European scientists moved to the US, UK and Australia. Their interest in philosophy and theory, when combined with the US interest in pragmatism led to a very productive period of post war research. But in the 1980s the European professors in the US began to retire and die. The US lost scientists with a broad philosophical background. People who understood the philosophical foundations of the second industrial revolution were missing. Considerable effort in the US has been devoted recently to improving STEM (science, technology, engineering and math) education. But the philosophical foundation underlying meaning, narratives, analogies, metaphors, purpose and understanding is missing in the US. American scientists tend to think that knowledge of philosophy and the humanities is not necessary for scientists. However, philosophy is considered an important part of European education. Philosophical differences and national cultural differences are related. In multi-disciplinary conversations in Europe references to the works of leading philosophers are common.

In 2015 Ronald Kline wrote The Cybernetics Moment or Why we call our age the information age. He said that cybernetics in the U.S. ended in 1975, a year after second order cybernetics was introduced. Terms like information systems and information management were commonly used; more complicated terms like circularity and autonomy and neurophysiological research were discussed less frequently. A major obstacle to the adoption of second order cybernetics in the US was von Foerster’s desire to include the observer in science. American scientists thought that science must be objective, which seemed to require deliberately not paying attention to the observer. But by studying the brain cyberneticians realized that doing scienceformulating hypotheses and conducting experimentsrequires a scientist or observer.

In addition to minimal education in philosophy a second obstacle to acceptance of cybernetics in the US was Russell’s Paradox. When writing his book, Principles of Mathematics, Bertrand Russell realized that formulating a set that contains itself would result in paradox, a form of inconsistency (Russell, 1903). Subsequently, many scientists knew that a description of a social system formulated by a member of that system would be undecidable: was the description intended to be accurate or to bring benefits to the author of the description? In a 1971 article, “Computing in the Semantic Domain,” von Foerster pointed out that describing a large system with both an observed part and an observer was mathematically possible and had been done. Hence, scientific work with one or more observers in the system of interest was not an obstacle (Foerster, 1971). But this work was not generally known among social scientists in the US.

Research on cybernetics now has two branches: a branch devoted to technical issues, including Artificial Intelligence, and a philosophical and theoretical branch that continues the research traditions initiated by Warren McCulloch, Heinz von Foerster, Ross Ashby, Humberto Maturana and others. There is currently a large gap between theoretical cybernetics research in the US and applied cybernetics research. The people doing applied cybernetics research with rare exceptions are not familiar with the work being done in theoretical cybernetics.

Given the obstacles to research on cybernetics in the USno university educational programs and no government research fundingresearch on theoretical cybernetics tended to move to Europe in the 1980s and 1990s. Since the 1950s there has been gradual but steady growth of cybernetics societies, conferences, and journals, but primarily in Europe (Umpleby, Wu & Hughes, 2017). Understanding cybernetics requires an interest in cognition and creativity and ideally academic training in philosophy. But philosophical discussions occur much more commonly in Europe than in the US.

In 2000 Karl Mueller and Albert Mueller (no relation) established the Heinz von Foerster Society in Vienna which held conferences and published books for several years. Alexander Riegler created the journal Constructivist Foundations in 2005. A special issue on second order science was published in 2014. A special issue on second order cybernetics was published in 2016 as a reply to Ronald Kline’s book (2015). Clearly work in cybernetics had not ended in 1975. Recent work has been concerned less with neurophysiology and more with philosophy of science (Foerster, 2003), social science (Mueller, 2011; Umpleby, 2014) and design (Glanville, 2009).

In Europe people who had contributed to advancing systems and cybernetics were not being selected for their national academies of science. Candidates were selected from the older, more established disciplines such as physics, chemistry and biology. So, the International Federation for Systems Research created an honor society for those who have made outstanding contributions to systems and cybernetics. The International Academy for Systems and Cybernetic Sciences (www.iascys.org) was founded in 2010 and now has over 70 academicians from countries in Europe, the Americas and Asia. The leaders of this effort were Matjas Mulej from Slovenia and Jifa Gu from China with Pierre Bricage in France serving as secretary general.

The goals of the Academy are to create more general theories to aid communication among disciplines; to show how the different disciplines are related; to use McCulloch’s idea of experimental epistemology to advance a scientific foundation for philosophy; to combine practice, theory and philosophy to provide more depth of understanding to theories; to include the observer in science in order to explore purposes and motivations; and to study and design the relationships among science, society and the environment in order to advance sustainability and the improvement of society (Mueller, 2011; Umpleby, 2014).

From time to time human beings, often not deliberately, conduct a large-scale social experiment. As an example, the Cold War was an effort determine whether capitalism or communism would lead to a more moral, productive and sustainable economic system We now seem to be conducting another large-scale social experiment to determine which approach to the construction of knowledge is most desirable and most productive. Several versions of philosophy, theory and methods exist around the world. Warren McCulloch initiated a unique approach to creating knowledge and advancing science. Experimental epistemology has made fundamental contributions to philosophy, cognition, physiology, psychology and ethics.

In his retirement Heinz von Foerster explained the significance of second order cybernetics by saying it was a Copernican revolution in science. Why did he choose Copernicus? Why not Galileo, Newton, Einstein or Darwin? Well, how did Copernicus change science? Before Copernicus people lived in a geocentric world. They imagined the earth was the center of the universe. After Copernicus they lived in a heliocentric world. They believed the sun was the center of the solar system. Before second order cybernetics the object of investigation was the center of scientific inquiry. After second order cybernetics the observer became a focus of attention, because all scientific theories, methods, experiments and results are created by observers. Also, the social and environmental implications of science should be attended to. The background, education, experience and motivations of observers are all relevant. They are not considerations to be disregarded. This is definitely a change in the way science is done.

References

Abraham, T.H. (2016). Rebel Genius: Warren S. McCulloch’s Transdisciplinary Life in Science, Cambridge, MA: MIT Press.

Foerster, H. von and G.W. Zopf, Jr. (eds.) (1962). Principles of Self-Organization: Transactions of the University of Illinois Symposium, London, UK: Pergamon Press.

Foerster, H. von. (1971). "Computing in the semantic domain," Annals of the New York Academy of Sciences, 184: 239–241.

Foerster, H. von. (1973). "On constructing a reality," in Preiser F.E. (ed.) Environmental Design Research, 2: 35-46.

Foerster, H. von (2003). Understanding Understanding: Essays on Cybernetics and Cognition, New York: Springer-Verlag.

Glanville, R. (2009). The Black Box: The Collected Works of Ranulph Glanville, Vienna: Edition Ecoraum.

Gregory, D. (1986). "Philosophy and Practice in Knowledge Representation," in Joseph Zeidner (ed.). Human Productivity Enhancement: Training and Human Factors in Systems Design, 1: 165-203.

Kline, R. (2015). The Cybernetics Moment or Why we call our age the information age, Baltimore, MD: Johns Hopkins University Press.

McCulloch, W.S. (1965). Embodiments of Mind, Cambridge, MA: MIT Press.

Müller, K.H. (2011). The New Science of Cybernetics: The Evolution of Living Research Designs, Vol. 2. Theory. Vienna: Ed. Echoraum, 277–316.

Mueller, K.H. (2016). Second Order Science: The Revolution of Scientific Structures, Vienna: Edition Ecoraum.

Pias, C. (2003). Cybernetics: The Macy-Conferences 1946-1953, Zürich: Diaphanes.

Riegler, A., and Mueller, K.H. (2014). "A Special Issue on Second Order Science," Constructivist Foundations, 10(1).

Rosenblueth, A., Wiener, N. and Bigelow, J. (1943). “Behavior, purpose and teleology,” in Philosophy of Science, 10: 18–24.

Russell, B. (1903). Principles of Mathematics, London: Routledge.

Umpleby, S.A. (2003). “Heinz von Foerster and the Mansfield Amendment,” Cybernetics and Human Knowing, 10(3-4): 161-163.

Umpleby, S.A. (2016). "Second order cybernetics as a fundamental revolution in science," Constructivist Foundations, 11(3): 455-81, link.

Umpleby, S.A. (2014). "Second order science: Logic, strategies, methods," Constructivist Foundations, 10(1): 16-23, link.

Umpleby, S.A., Wu, X.-H. and Hughes, E. (2017). "Advances in cybernetics provide a foundation for the future," in International Journal of Systems and Society, Special Issue on the Future of Systems, 4(1): 29-36.

Yovits, M.C. and Cameron, S. (eds.) (1960). Self-organizing Systems, Oxford, UK: Pergamon Press.

Wiener, N. (1948). Cybernetics or Control and Communication in the Animal and the Machine, Cambridge, MA: MIT Press.