Allen Newell

Overview

Allen Newell (1927–1992) was an American computer scientist and psychologist who spent his career trying to state, in the form of working programs, how human thinking happens. He was born in San Francisco, California, on March 19, 1927, and died in Pittsburgh, Pennsylvania, on July 19, 1992^[1]. From 1961 to 1992 he was Professor of Computer Science and Psychology at Carnegie-Mellon University, a double appointment that matched his view that artificial intelligence and the psychology of thinking are two routes into one problem^[1].

Education and early research

Newell earned a BS in Physics from Stanford University in 1949, and a PhD from the Graduate School of Industrial Administration at Carnegie Institute of Technology in 1957^[1]. From 1950 to 1961 he worked at the RAND Corporation^[1], at a time when digital computers had become available to researchers and the question of what such machines could be made to do was still open.

Programs that reason

With Herbert A. Simon and J. C. Shaw, Newell developed the first list-processing language, IPL, together with two heuristic problem-solving programs, Logic Theorist and General Problem Solver^[1]. The three pieces fit together. List processing organises computer memory as chains of symbols that can be lengthened, shortened and re-linked, rather than as a fixed block of numbers; a program can then hold objects, relations and goals as data it is able to take apart and rebuild. A heuristic is a rule of thumb that usually helps find a solution without guaranteeing one, and it is what makes a search through many possibilities manageable: instead of trying every step, the program tries the steps that look promising.

Logic Theorist applied this to reasoning in symbolic logic, treating a proof as a path through possible steps and using heuristics to decide which step to take next. General Problem Solver pushed for generality: rather than a program for one task, it was an attempt at a common core of problem-solving machinery that could be pointed at different problems. Its characteristic move was to compare what it had with what it wanted, measure the difference, and look for an operation that would reduce it — a strategy usually called means-ends analysis. The ambition behind it, that a small set of mechanisms might account for problem solving across many domains, became a lasting frame for the field.

Career

Newell's working life ran through two institutions. He was at the RAND Corporation from 1950 to 1961, and then, from 1961 until his death in 1992, at Carnegie-Mellon University, where he held the joint professorship in computer science and psychology^[1]. The same period saw the results of that work recognised in a long series of honours.

Recognition

He received the Harry Goode Memorial Award of the American Federation of Information Processing Societies in 1971 and was elected to the United States National Academy of Sciences in 1972^[1]. In 1975 he shared the A.M. Turing Award of the ACM with Herbert A. Simon^[1]. He was elected to the United States National Academy of Engineering in 1980, and in that year served as First President of the American Association for Artificial Intelligence^[1]. In 1981 he received the IEEE Computer Society Computer Pioneer Award as a charter recipient, and in 1989 the IJCAI Award for Research Excellence^[1]. He received the U.S. National Medal of Science in 1992, and the ACM/AAAI Allen Newell Award was named in his honour^[1].

The question behind the work

Newell described his driving question as "How can the human mind occur in the physical universe."^[1] The list-processing language, the heuristic search programs and the double appointment in computer science and psychology all belong to that one project: to say precisely enough how thinking works that a machine could be made to do it, and to learn from the machine when the description fails.