W. Ross Ashby

W. Ross Ashby: The Psychiatrist Who Engineered Adaptation

For forty-four years, from 1928 until the year he died, Ross Ashby kept a journal. It runs to twenty-five volumes and 7,189 pages, and it now sits in the British Library. Very little of it is diary. It is a working notebook — the sustained record of one man asking a single question for his entire adult life: what would it actually take to build something that adapts? He was a hospital psychiatrist by trade. He answered the question with a machine.

Cambridge and a Doctor's Training

Ashby was born in London on 6 September 1903, the son of a man who worked at an advertising agency. He went up to Sidney Sussex College, Cambridge, in 1921 and stayed until 1928, taking a B.A. in 1924 and qualifying in medicine, M.B. and B.Ch., in 1928, with clinical training at St Bartholomew's Hospital. Cambridge kept awarding him things afterwards: an M.A. in 1930, a Diploma in Psychological Medicine in 1931, an M.D. in 1935.

The medical formation matters. Ashby was never an engineer who wandered into biology; he was a clinician who spent his working days among people whose capacity to regulate themselves had broken down, and who wanted to know what regulation *was*. He began as a clinical psychiatrist with the London County Council in 1930, and from 1936 to 1947 served as research pathologist at St Andrew's Hospital in Northampton — a private mental hospital, not a physics faculty, which is where the foundations of his theory were laid. The war interrupted with a commission: he served as a Major in the Royal Army Medical Corps in India from 1945 to 1947.

The Homeostat

In 1947 he became Director of Research at Barnwood House Hospital in Gloucester, a post he held until 1959. There, in 1948, he built the homeostat: an electro-mechanical machine capable of adaptive behaviour, and the first mechanical demonstration of what he called *ultrastability*.

The idea behind it is subtler than the popular description of a machine that seeks equilibrium. An ordinary self-regulating device — a thermostat — corrects deviations using a control law its designer installed. The homeostat could do something else: when disturbed beyond the range its current organisation could handle, it changed that organisation, searching essentially at random through its own internal parameters until it stumbled into a configuration that held stable again. It did not have the answer built in. It had a procedure for finding answers to problems its designer had never anticipated, and it found them by trial that was genuinely blind.

This was a philosophical claim disguised as hardware. Adaptation, Ashby was arguing, does not require foresight, intention or intelligence in any special sense. It requires only a system that can reorganise itself when its essential variables leave their survivable range. Grey Walter's tortoises got more press; Ashby's homeostat carried the harder idea.

Design for a Brain

He set it all out in *Design for a Brain* in 1952. The title is deliberately provoking, and the book is not a work of neuroscience — there is very little brain tissue in it. It is an argument that the essential properties of a nervous system can be specified abstractly, as a dynamical system whose stability conditions can be written down, and that once you have specified them the biological substrate becomes a detail of implementation.

By then he was inside the small international circle inventing cybernetics. He co-founded the Ratio Club, the British dining-and-arguing society whose orbit included Alan Turing, and attended the Macy Conference in 1952, where the American branch of the field — Norbert Wiener, Claude Shannon — was assembling itself.

Only Variety Can Absorb Variety

*An Introduction to Cybernetics* followed in 1956, and it contains the result that outlived everything else he did. The Law of Requisite Variety states, in his formulation, that "only variety in [the regulator] can force down the variety due to [disturbances]."

Unpacked: a controller can only handle as many distinct situations as it has distinct responses. If the world can throw a thousand different disturbances at you and you have fifty available reactions, nine hundred and fifty of those disturbances will get through, and no amount of cleverness in deploying your fifty will change the count. Regulation has a hard capacity limit, and the limit is countable.

The law is one of the few genuinely quantitative results in the social and management sciences. Stafford Beer built an entire theory of organisational design on it. It reappears in the design of everything from immune-system models to corporate structures, and it explains, in a single line, why simple institutions cannot govern complicated ones.

Illinois

Ashby directed the Burden Neurological Institute in Bristol briefly, and then in 1960, at the invitation and arrangement of Heinz von Foerster, moved to the University of Illinois at Urbana-Champaign as Professor of Biophysics and Electrical Engineering — a psychiatrist appointed to an engineering faculty, which tells you how completely he had crossed over. He remained there until retiring in 1970. He served as president of the Society for General Systems Research from 1962 to 1964, was made an Honorary Professorial Fellow of the University of Wales in 1970 and a Fellow of the Royal College of Psychiatrists in 1971.

Why Ross Is Called a Genius

Ashby has been called the major theoretician of cybernetics after Wiener, and the specific quality that earns the description is a talent for radical abstraction — stripping a phenomenon of every feature except the one that makes it work. Confronted with brains, he removed the neurons. Confronted with control, he removed the controller's contents and kept only a count of its possible states. What remained were results so general they applied to a thermostat, a mental hospital and a national economy without modification, and so austere that they could be proved rather than merely argued. Very few people can operate at that altitude without producing vacuity; Ashby produced theorems.

The counter-case is that this same abstraction is why he is not famous. He is markedly less well known than many of the people his work shaped — Herbert Simon, Stafford Beer, Stuart Kauffman among them — partly because generality without application is easy to admire and hard to use. The homeostat searched for stability by random trial, which is a demonstration of a principle rather than a plausible model of how any real nervous system learns; it does not scale, and Ashby never showed that it could. Cybernetics as a discipline dispersed in the 1970s, its questions absorbed by control theory, AI and complexity science, and Ashby's name went with it. A hard-headed reading makes him a superb formalist whose central machine was a toy and whose field dissolved. What that reading misses is the Law of Requisite Variety, which is not a toy, is not vague, and has never been refuted.

Legacy

An inoperable brain tumour was diagnosed in June 1972. Ashby died on 15 November of that year, and the journal stopped after forty-four years.

Its survival is fitting. Ashby's conviction was that intelligence is not a substance but an organisation — that what matters about a mind is the shape of its dynamics, not the material it happens to run on. Every argument now conducted about whether machines can think proceeds on that assumption, usually without knowing whose assumption it was. The whole 7,189 pages are a demonstration of the same thing he built into the homeostat: keep varying until something holds.

Achievements

• Held posts at University of Illinois Urbana-Champaign
• Fields: cybernetics, neurology and systems theory