Herbert A. Simon Papers

Herbert A. Simon (b. Milwaukee, 1916-d. Pittsburgh, 2001) received his A. B. in Political Science at the University of Chicago in 1936. He earned his Ph.D. in 1943, in that same department headed by Charles Merriam, whose faculty favored a dissolving of boundaries between departmental disciplines and methodologies, and a corresponding application of systematic observation and quantitative analysis to political science. The environment was a productive influence on Simon, who, even as a scholar of 17, was confirmed in “the idea that human behavior could be studied scientifically,” and had seen, “if dimly, the challenge of bringing to social science…the mathematical thinking that had been so powerful” in hard sciences.

Already extensively read in economics in high school, Simon’s undergraduate coursework in price theory gave him “a glimpse of the applications of rigor and mathematics to economics.” At Chicago, Simon pursued private study in higher mathematics, eventually reading the equivalent of a doctoral curriculum. Progressing into graduate work, Simon learned from faculty outside the Political Science department the significant advantages of constructing mathematical models of biological phenomena; of the uses of modern statistical theory in economics; and of the importance of logical foundations to the sciences. The latter knowledge would serve Simon well in his computer science work, first in writing programming languages based on syllogistic reasoning, and then in defining the new field of computer studies as a logically-based science -- as he did with Allen Newell in their 1975 joint A. M. Turing Award lecture, “Computer Science as Empirical Inquiry: Symbols and Search”

After several projects in city management analysis under the tutelage of Professor Clarence Ridley, Simon turned his attention to organizational decision making. His Chicago thesis, Administrative Behavior (1943; published, with revisions, 1947) eschewed previous classical theorizing on corporate decisions. Where the classical model postulated an informed, centralized executive choosing amid a wealth of information to maximize profit, Simon applied the behaviorism absorbed at Chicago’s interdisciplinary Political Science department to produce a more realistic model. Administrative Behavior defined modern, large-organization behavior as decentralized decision making, performed by middle managers that could not have available to them all the information (or the time and ability to analyze it) needed to make “maximizing” decisions. Further, decision making behavior -- within or without the organizational frame -- is “determined by the irrational and nonrational elements that bound the area of rationality.” As a result of “bounded rationality,” Simon later elaborated, decision makers opt for “satisficing” behavior -- decisions that are “good enough” within allowances and under influences. A byproduct of Simon’s empirical study of organizational behavior was the note that decision making is a process of achieving mental subgoals that lead to a larger decision. This process of “making successive choices along a branching path” became the pattern for Simon’s subsequent work in simulating human decision making by computer. It lead to his prediction that thoroughly programmed computers, designed to learn from data and past decisions, would recentralize and rationalize organizational decision making. Three years (1939-1942) as director of the Bureau of Public Administration at the University of California, Berkeley, led to a faculty position at the Illinois Institute of Technology (1942-1949), where Simon eventually chaired the department of political and social science. Returning to Chicago offered Simon the opportunity of joining the Cowles Commission for Research in Economics. In its weekly meetings, Simon interacted with the Midwest's leading economists, mathematicians, and early computer scientists -- beginning the effort “to see and to be seen” which increased his influence and advancement. Recommend by Cowles colleague Bill Cooper to the new Graduate School of Industrial Administration at Pittsburgh’s Carnegie Institute of Technology, Simon became a faculty member in 1949. With Cooper, dean Lee Bach, and provost Elliott Dunlap Smith, Simon was instrumental in developing a faculty and a curriculum influenced by social science methods, creating “settings in which we could compare the ways decisions were actually made in business firms with the ways that economic theory and textbooks said they were made.”

In 1952, Simon began a 24-year consulting relationship with RAND, the Santa Monica-based “think tank.” Previous years spent researching human problem solving behavior were to find maximized application when Simon, on his first RAND visit, encountered both digital computing and Allen Newell. Newell -- who became Simon’s long-time research partner and faculty colleague at Carnegie Tech (later CMU) -- shared with Simon the conviction that the human mind was an information-processing system that manipulated symbols to “think,” the apprehension that the digital computer could perform non-numeric symbol recognition, and the realization that a programmed computer could formally mimic the “branching path” of decision making employed in problem solving behavior. In January 1956, Simon astonished a Carnegie Tech classroom by announcing that, “Over Christmas, Al Newell and I invented a thinking machine.” The LT (Logic Theorist), program used heuristic instructions strategized by Simon and produced by Newell and RAND programmer J. C. “Cliff” Shaw to adapt the original concept - solving visually-patterned chess problems -- to proving, with speed and success, propositional calculus theorems in Whitehead and Russell’s Principia Mathematica. In 1957, the NSS (Newell-Shaw-Simon) chess machine followed -- a program designed to replicate human decision making in play rather than to win games.

Simon and Newell’s interest in interpreting the LT program as a cognitive theory of problem solving was tested with empirical evidence -- protocols composed of the test subjects’ thinking-aloud descriptions of their behaviors in problem solving tasks. Subjects were clearly using means-end analysis, or comparing the task goal with their mental distance from it and noting the actions that could reduce the distance. The important new observation was introduced into the programming of the GPS, or General Problem Solver, in 1957. The GPS system performed a broad range of tasks using only one set of mechanisms. Edward Feigenbaum, then Simon and Newell’s student, notes that the performance precedents in generality set by GPS “commanded the attention of its generation of scientists.” Feigenbaum -- now at Stanford -- was among the students who developed, with Simon and Newell, the list-processing languages (IPLs or Information Processing Languages) that programmed the LT and GPS. Other students designed programs that used heuristic search to plan industrial productions, and expert systems to diagnose problems and offer solutions in specialized fields.

Simon’s books and publications since Administrative Behavior reflect the range of his research interests and their influence on economic, cognitive, and artificial intelligence studies. Papers such as “On the Definition of the Causal Relation” (1952) and “Causal Ordering and Identifiability” (1953) offered insights into the breaking down of complex equation systems -- cited in the announcement of his 1978 Nobel Prize as “of particular importance” to the economic sciences. Models of Man (1957) collected papers on causality written for the Cowles Commission with work from his Carnegie and RAND research; “A Behavioral Model of Rational Choice” (first published 1955) formalized the theory of bounded rationality and introduced the satisficing concept. Simon’s 1959 New York University lectures on management generated his 1960 book, The New Science of Management Decision; his 1968 Compton Lectures at MIT on artificial intelligence produced The Sciences of the Artificial (1969). Both works assessed and predicted the social effects of his researches to the chagrin of other theorists, as was frequently the case with Simon’s predictions. The long-deferred (for research interests) co-production with Allen Newell, Human Problem Solving (1972), summarized 30 years of research into cognition and its successful experimental translation to computer programming. Representation and Meaning (with Laurent Siklóssy, 1972) studied the cognitive import of the discovery that natural language computer systems could avoid semantic ambiguity by consulting combinations of pictures and sentences. The 1975 citation of the Association for Computing Machinery’s A. M. Turing award, awarded jointly to Simon and Allen Newell, cites their contributions “to the establishment of [artificial intelligence] as an area of scientific endeavor,” and to “the idea that human cognition can be described in terms of a symbol system.” The Royal Swedish Academy of Sciences’ press release on the occasion of awarding Simon the 1978 Nobel Prize in Economics praised him for “scientific output…far beyond the disciplines in which he has held professorships.” Whether in science theory, applied mathematical statistics, operations analysis, economics, or business administration, “Simon has had something of importance to say; and…has developed his ideas to such an extent that it has been possible to use them as a basis for empirical studies.”

Other of Dr. Simon’s honors include the Award for Distinguished Scientific Contributions of the American Psychological Association (1969); Distinguished Fellow of the American Economic Association (1976); the Frederick Mosher Award of the American Society of Public Administration (1974); the Procter Prize of Sigma Xi fraternity (1980); the Dow-Jones & Company Award (1983); the Award for Scholarly Contributions to Management of the Academy of Management (1983); the James Madison Award of the American Political Science Association (1984); the National Medal of Science (1986); the American Psychological Foundation Gold Medal Award for Life Achievement in Psychological Science (1988); the John von Neumann Theory Prize of ORSA/TIMS (1988); the American Psychological Association Award for Outstanding Lifetime Contribution to Psychology (1993); the International Joint Conferences on Artificial Intelligence Award for Research Excellence (1995) and the Dwight Waldo Award of the American Society of Public Administration (1995). Dr. Simon’s numerous science organization fellowships and honorary degrees are listed in the records history for Series X, Awards, below.

From 1966 until his death in 2001, Dr. Simon was the Richard King Mellon University Professor of Computer Science and Psychology at Carnegie Mellon University. He lived in Pittsburgh with his wife Dorothea (née Pye, d.2002). They have three children: Katherine, Peter and Barbara.