Machine Intelligence is not Artificial - Part 5: The Ratio Club and British Cybernetics

(This essay is Part 5 of an ongoing series, Machine Intelligence is not Artificial (MIINA) - MIINA Part 1: (Four Funerals & a Divorce), MIINA Part 2: (Allen Newell's Dissertation), MIINA Part 3: In the Beginning, MIINA Part 4: Cybernetics & Norbert Wiener, MIINA Part 5: The Ratio Club & British Cybernetics, MIINA Part 6: Dartmouth 1956, The Birth of AI and the Balkanization of Machine Intelligence, MIINA Part 7: Interlude)

Additionally, this essay series intersects with the wide-ranging thought work I've begun with Amicia D. Elliott, Ph.D. that we refer to as Neuroboros. We have begun producing a Neuroboros podcast, the first two episodes of which are available:

Episode 1 - Cybernetics and Wiener Walks roughly parallels Part 4 of the essay series

Episode 2 - The Ratio Club covers much of the same ground as this Part 5

The podcasts were recorded last year, and done somewhat extemporaneously, so while they follow the same path as this essay there will be divergence. This means each has its own gems and flaws. In the case of factual discrepancies, defer to the essay (and please let me know; this is an early and informal exercise, and I am learning, correcting, and improving as I go).

If all that seems a bit complex, it is. Over-simplification and misaligned incentives have left a lot of stuff we could use on the cutting room floor. It will get worse before it gets better. But it will be worth it. Now, let us continue...

There were a number of people in Britain who had Norbert Wiener's ideas before Wiener published his book, Cybernetics; at least 20 of them. Thinking about the science of control and communication in animals and machines, aka cybernetics, before Wiener's 1948 book came out was one of the prerequisites to being invited into the Ratio Club - the UK's answer to the Macy Conferences on Cybernetics in the US. The Ratio Club differed in form and function, and yielded no formal output unlike the American group, and yet its influence on the early days of machine intelligence may have been as great or greater.

A Cybernetics Dining Club

The Ratio Club was an informal gathering of physicians, biologists, psychologists, engineers and mathematicians who met 37 times in and around London between 1949 and 1955 (with a reunion in 1958). Meetings consisted of drinks, a couple lectures by members or guests, dinner, and finally discussion of the topic of the day over beers; what one attendee referred to as "a cybernetics dining club." It was organized by neurologist John Bates, who curated the invitees to provide professional balance and enough personality variation to keep the discussion lively. The only two criteria for membership were to have been known to be thinking and working on ideas related to cybernetics before Wiener's book came out in 1948, and to not yet be a professor (or the boss of anyone else in the group). The latter rule was intended to avoid stifling open discussion with the somewhat rigid academic hierarchy of the day.

There are several suggestions for where the name may have come from, including Wiener himself in the use of the term, "machina ratiocinatrix" in the introduction to Cybernetics (machina ratiocinatrix or the MR Club was the suggestion of member Donald Mackay). The fact that Bates is reported to have liked the versatility of the word "ratio" may point to multiple points of origin.

If I were to choose a patron saint for cybernetics out of the history of science, I shall have to choose Leibniz. . .. The calculus ratiocinator of Leibniz contains the germs of the machina ratiocinatrix, the reasoning machine. - Norbert Wiener, Cybernetics

The idea for the club came from Bates and a discussion he had with his colleagues at a meeting for "Animal Mechanisms of Behavior" in the summer of 1949. The meetings were initially held in the basement of the National Hospital for Nervous Disease in Bloomsbury, and later moved to different locations around and outside of London, sometimes coinciding with larger conferences on related topics. Membership was initially 15, but grew to 21 at its peak, as additional invitees were suggested. Not every meeting was attended by every member, and guests were common - including Warren McCulloch from the US, whose visit to London in September of 1949 used as driver to schedule the first Ratio Club meeting.

Many of the members went on to have significant impact on their respective fields, including Grey Walter and Ross Ashby who themselves were leaders in the interdisciplinary field of cybernetics. Alan Turing was invited along with Phillip Woodward after the first meeting in order to have some mathematicians who could "keep the biologists in order," according to John Pringle. It seems likely that the discussions Turing had at these Ratio Club meetings, which were almost monthly for the first year and a half, directly influenced his 1950 paper in the journal Mind, "Computing Machinery and Intelligence". Turing presented the ideas from his paper to the group during the 14th meeting on 7 December 1950 under the topic, "Educating a Digital Computer."

Content and Themes

While the initial several meetings had general introductory talks by the members getting to know each other, subsequent meetings had one or two short talks before dinner that allowed members and guests to highlight their work relating to a specific topic. The was followed by a deep dive discussion of that topic during and after dinner by the group. Ross Ashby had generated the groups' initial list of question and discussion areas and circulated it to the group in February of 1950. The refined version of the list from May of 1950 gives a sense of what was on the minds of Ashby and the other members (Ashby's list below comes from the unpublished Ratio Club papers of John Wescott, as reproduced in The Mechanical Mind in History, Husbands, Holland, and Wheeler, 2008):

1. What is known of "machines" that are defined only statistically? To what extent is this knowledge applicable to the brain?
2. What evidence is there that "noise" (a) does, (b) does not, play a part in brain function?
3. To what extent can the abnormalities of brains and machines be reduced to common terms?
4. The brain shows some indifference to the exact localisation of some of its processes: to what extent can this indifference be paralleled in physical systems? Can any general principle be deduced from them, suitable for application to the brain?
5. From what is known about present-day mechanical memories can any principle be deduced to which the brain must be subject?
6. To what extent do the sense-organs' known properties illustrate the principles of information-theory?
7. Consider the various well known optical illusions: what can information-theory deduce from them?
8. What are the general effects, in machines and brains[,] of delay in the transmission of information?
9. Can the members agree on definitions, applicable equally to all systems - biological, physiological, physical, sociological - cf: feedback, stability, servo-mechanism.
10. The physiologist observing the brain and the physicist observing an atomic system are each observing a system only partly accessible to observation: to what extent can they use common principles?
11. The two observers of 10, above, are also alike in that each can observe his system only by interfering with it: to what extent can they use common principles?
12. Is "mind" a physical "unobservable"? If so, what corollaries may be drawn?
13. What are the applications, to cerebral processes, of the thermodynamics of open systems?
14. To what extent can the phenomena of life be imitated by present-day machines?
15. To what extent have mechanisms been successful in imitating the conditioned reflex? What features of the C.R. have conspicuously not yet been imitated?
16. What principles must govern the design of a machine which, like the brain, has to work out its own formulae for prediction?
17. What cerebral processes are recognisably (a) analogical, (b) digital, in nature?
18. What conditions are necessary and sufficient that a machine built of integrated parts should be able, like the brain, to perform an action either quickly or slowly without becoming uncoordinated?
19. Steady states in economic systems.
20. What general methods are available for making systems stable, and what are their applications to physiology?
21. To what extent can information-theory be applied to communication in insect and similar communities?
22. To what extent are principles of discontinuous servo-mechanisms applicable to the brain?
23. What re-organisation of the Civil Service would improve it cybernetically?
24. What economic "vicious circles" can be explained cybernetically?
25. What re-organisation of the present economic system would improve it cybernetically?
26. To what extent can information-theory be applied to the control exerted genetically by one generation over the next?
27. Can the members agree on a conclusion about extra-sensory perception?
28. What would be the properties of a machine whose "time" was not a real but a complex variable? Has such a system any application to certain obscure, i.e. spiritualistic, properties of the brain?

There was an additional topic on the original list that was dropped from the final, more detailed list, specifically, "If all else fails: the effect of alcohol on control and communication, with practical work." It does seem to have been at play in all of the Ratio Club meetings. Guests were a regular part of the meetings, as both speakers and observers. Many guests were US leaders in cybernetics and information theory (McCulloch, Oliver Selfridge, J.C.R. Licklider and others). Other guests were personal colleagues of the members - including at one point the head of the US National Security Agency, who had worked with member Jack Good when he was with British Intelligence during World War 2, who was part of a deep dive discussion on extra-sensory perception as part of meeting on Probability.

Machine Intelligence in Action

In addition to the talks and discussions that took place at the Ratio Club meetings, there was also an amateur interest of several of the members in creating their own applications of many of the ideas they were discussing in the form of automata. Most notable among these were the Ross Ashby's Homeostat and Grey Walter's Tortoise.

Ross Ashby was one of the earliest and most influential of the cyberneticist. His earliest work related to the topic, his 1940 article "Adaptiveness and Equilibrium" in the Journal of Mental Sciences, pre-dates Norbert Wiener's own work in the early 1940's. He was also communicating with Kenneth Craik on these topics as early as 1944. Craik, whose 1943 book The Nature of Explanation inspired the cybernetics movement but died tragically in 1945, was invoked in Ratio Club founder Bates' notes from the first meeting. Ashby went on to continue this line of thinking and his work past the end of the war.

Ashby developed a device, the Homeostat, that was capable of maintaining its homeostasis in the face of changes from the outside environment through habituation, learning and feedback. The Homeostat was novel and successful enough to be highlighted in the international media in the late 1940's as the most advanced artificial brain to date. Ashby also went to the US to present on the Homeostat to the Macy Conference on Cybernetics in 1952, along with publishing more about its design in his book, Design for a Brain, that same year. He later published a more general book, An Introduction to Cybernetics, in 1956 that helped to further codify the science captured in the first-order cybernetics.

Grey Walter, another of the British cybernetic pioneers, was also known for his robots, a pair of Tortoises - Elmer and Elsie. These were able to move around, react to external stimuli, learn and unlearn, and reconnect to their charger when their power was low. Other more robust versions of the tortoises were also constructed by Walter and his colleagues. Some of these live on in museums today. This trend of creating synthetic automata caught on with the US cyberneticists as well, with Claude Shannon, Norbert Wiener, and others picking up the hobby.

Influence and Cybernetics 2.0

While the Ratio Club served as a forum for the British cyberneticists in a similar fashion to the Macy Conferences for the American cyberneticists, with cross-pollination between the two, there were some key differences between them. In the US, the cybernetics movement may have been off-the-beaten academic path, but it was still largely government and non-profit funded work. The Macy Conferences produced transcripts and reports (though these were limited in the first few) to be shared with others and disseminate the ideas. In the UK, the cybernetics efforts were far more hobbyist, akin to today's DIY movement. The Ratio Club had a much more informal atmosphere intended to foster collaboration and inspiration among the participants, but without formal reports or dissemination of the discussions. Its members did, however, produce the majority of the papers published in Claude Shannon and John McCarthy's Automata Studies (1955), which served as a direct precursor to and likely impetus for the McCarthy led proposal to the Rockefeller Foundation in 1955 for the Dartmouth Conference on Artificial Intelligence - the birthplace of AI - in 1956. We'll touch on this more in the next essay (Part 6) in this series.

As the Ratio Club moved through its desired topics, with its members getting busier with careers and families (along with the tragic death of Turing in 1954), they decided to wind it down in 1955 (with a reunion in 1958, that was only partially attended). The mantle for European cybernetics passed to the International Congress on Cybernetics first held in Namur, Belgium in 1956 (sometimes referred to as the Namur Conferences) and the Ratio Club members went on to transform a number of brain related fields, with Ashby and Walter continuing to influence the core of cybernetics. As in the US, this time period saw an extension of cybernetics spread to non-STEM areas (the integration of cybernetics into the social sciences is called second-order cybernetics) and inspire a wider second generation of cyberneticists.

Here, the second generation of cyberneticists out of the UK (including Greg Bateson, R.D. Laing, Stafford Beer, Gordon Pask, and more) played a considerable role in advancing a move of this thinking into the psychological and social sciences and humanities - intersecting not only with the adoption of cybernetics by cultural movements in the US in the 60s, but also with the Soviet's newfound adoption of cybernetics in the late 50's and early 60's (and somewhat curiously the US intelligence community, who had shifted US funding focus to AI rather than cybernetics in early 60s, but by the late 60s helped start the American Cybernetics Society). As I said before, things got weird in the 60s (and worse in the 70s), but that is for a later discussion.

One side effect of the Ratio Club's informal, internal only dialogue is that while it may have been influential to some, there is little captured history of it to date. After someone discovered pictures and papers from the group in a desk in the early 80's, there was briefly and effort by Bates and others to write a history. This was not successful, but at least consolidated some of the information. Philip Husbands and Owen Holland began pulling together a history and interviewing those members that were still alive in the early 2000's. Together with sources like Ashby's journals available in the British National Library, they put together the wonderful article for the book they edited that I mentioned above ("The Ratio Club" Husbands & Holland in The Mechanical Mind in History (2008), Eds. Husbands, Holland, and Wheeler), along with some follow-on papers. A great debt is owed to them for many aspects of this phase in history not being lost. As we saw in Part 2 and will be looking at again in the next couple Parts of this series, there are key elements of machine learning that have been absent since the acceleration of AI, and these dormant pieces may place a critical role in future advances.

Cheers!