The Cybernetic Brain

by Andrew Pickering

  4

  _ _ _ _ _

  ROSS ASHBY

  psychiatry, synthetic brains,

  and cybernetics

  HAVING DECIDED (HEAVEN FORGIVE ME, BUT IT IS MY CONVICTION) TO FOLLOW IN DARWIN'S FOOTSTEPS, I BOUGHT HIS AUTOBIOGRAPHY TO GET SOME HINTS ON HOW TO DO IT.

  ROSS ASHBY,JOURNAL ENTRY, 29 JUNE 1945 (ASHBY 1951–57, P. 1956)

  William Ross Ashby (fig. 4.1), always known as Ross, was born in London on 6 September 1903.1 After failing the entrance exam for the City of London School, he finished his schooling at the Edinburgh Academy between 1917 and 1921 and then graduated from Sidney Sussex College, Cambridge, with a BA in zoology in 1924. He was an unhappy child, incapable of living up to the expectations of a demanding father, and this unhappiness remained with him for many years.2 Ashby's father wanted him to pursue a career in either medicine or the law and, opting for the former, on leaving Cambridge Ashby trained at St. Bartholomew's Hospital, receiving the M.B. and B.Ch. degrees in 1928 (qualifying him to practice as a doctor) and the M.D. degree in 1935, both from Cambridge. In 1931 he was awarded a diploma in psychological medicine by the Royal College of Physicians and Surgeons. From 1930 to 1936 he was employed by London County Council as a clinical psychiatrist at Leavesden Mental Hospital in Hertfordshire. In 1931 Ashby married Elsie Maud Thorne—known to her intimates as Rosebud; Mrs. Ashby to others; born in 1908; employed at that point in the Millinery Department at Liberty's on Regent Street—and between 1932 and 1935 they had three daughters, Jill, Sally, and Ruth.

  Figure 4.1.W. Ross Ashby. (By permission of Jill Ashby, Sally Bannister, and Ruth Pettit.)

  From 1936 to 1947 Ashby was a research pathologist at St. Andrew's mental hospital in Northampton, an appointment he continued to hold while serving from 1945 until 1947 as a specialist pathologist in the Royal Army Medical Corps with the rank of lieutenant and later major. From June 1945 until May 1946 he was posted to India, in Poona and Bangalore. Returning to England, he became director of research at another mental institution, Barnwood House in Gloucester, in 1947 and remained there until 1959, when he was appointed director of the Burden Neurological Institute in Bristol, succeeding Frederick Golla and becoming Grey Walter's boss. In January 1961, after just a year at the Burden, Ashby moved to the United States to join the University of Illinois (Urbana-Champaign) as a professor in the Department of Electrical Engineering, primarily associated with Heinz von Foerster's Biological Computer Laboratory (BCL) but with a joint appointment in biophysics. He remained at the BCL until his retirement as an emeritus professor in 1970, when he returned to Britain as an honorary professorial fellow at the University of Wales, Cardiff. He died of a brain tumor shortly afterward, on 15 November 1972, after five months' illness.

  Ashby's first recognizably cybernetic publication, avant la lettre, appeared in 1940. In the mid-1940s he began to make contact with other protocyberneticians, and in 1948 at Barnwood House he built the cybernetic machine for which he is best remembered, the homeostat, described by Norbert Wiener (1967 [1950], 54) as "one of the great philosophical contributions of the present day." The concept of adaptation staged by the homeostat, different from Walter's, will echo through the following chapters. Over the course of his career, Ashby published more than 150 technical papers as well as two enormously influential books: Design for a Brain in 1952 and An Introduction to Cybernetics in 1956, both translated into many languages. From the homeostat onward, Ashby was one of the leaders of the international cybernetics community—a founding member of the Ratio Club in Britain, an invitee to the 1952 Macy cybernetics conference in the United States, and, reflecting his stature in the wider world of scholarship, an invited fellow at the newly established Center for Advanced Study in the Behavioral Sciences in Palo Alto, California, in 1955–56. After moving to Illinois, he was awarded a Guggenheim Fellowship in 1964–65, which he spent back in England as a visiting research fellow at Bristol University.3

  Ashby's contributions to cybernetics were many and various, and I am not going to attempt to cover them all here. Speaking very crudely, one can distinguish three series of publications in Ashby's oeuvre: (1) publications relating to the brain that one can describe as distinctly cybernetic, running up to and beyond Design for a Brain; (2) distinctly medical publications in the same period having to do with mental pathology; and (3) more general publications on complex systems having no especial reference to the brain, running roughly from the publication of An Introduction to Cybernetics and characterizing Ashby's later work at Illinois. My principle of selection is to focus mostly on the first and second series and their intertwining, because I want to explore how Ashby's cybernetics, like Walter's, developed as brain science in a psychiatric milieu. I will explore the third series only as it relates to the "instability of the referent" of the first series: although Ashby's earlier work always aimed to elucidate the functioning of the brain, normal and pathological, he developed, almost despite himself, a very general theory of machines. My object here is thus to explore the way that Ashby's cybernetics erupted along this line into a whole variety of fields, but I am not going to follow in any detail his later articulation of cybernetics as a general science of complex systems. This later work is certainly interesting as theory, but, as I have said before, I am most interested in what cybernetics looked like when put into practice in realworld projects, and here the natural trajectory runs from Ashby's cybernetic brain not into his own work on systems but into Stafford Beer's management cybernetics—the topic of the next chapter.

  The skeleton of what follows is this. I begin with a brief discussion of Ashby's distinctly clinical research. Then I embark on a discussion of the development of his cybernetics, running through the homeostat and Design for a Brain up to the homeostat's failed successor, DAMS. Then I seek to reunite these two threads in an exploration of the relation between Ashby's cybernetics and his clinical work up the late 1950s. After that, we can pick up the third thread just mentioned, and look at the extensions of Ashby's research beyond the brain. Finally, I discuss echoes of Ashby's work up to the present, in fields as diverse as architecture, theoretical biology and cellular automata studies. Throughout, I draw heavily upon Ashby's handwritten private journal that he kept throughout his adult life and various notebooks, now available at the British Library in London.4

  The Pathological Brain

  When one reads Ashby's canonical works in cybernetics it is easy to imagine that they have little to do with his professional life in medicine and psychiatry. It is certainly the case that in following the trajectory of his distinctive contributions to cybernetics, psychiatry recedes into the shadows. Nevertheless, as I will try to show later, these two strands of Ashby's research were intimately connected, and, indeed, the concern with insanity came first. To emphasize this, I begin with some remarks on his medical career.

  Overall, it is important to remember that Ashby spent his entire working life in Britain in mental institutions; it would be surprising if that milieu had nothing to do with his cybernetic vision of the brain. More specifically, it is clear that Ashby, like Walter, belonged to a very materialist school of psychiatry led in Britain by Frederick Golla. Though I have been unable to determine when Ashby first met Golla and Walter, all three men moved in the same psychiatric circles in London in the mid-1930s, and it is probably best to think of them as a group.5 It is clear, in any event, that from an early date Ashby shared with the others a conviction that all mental phenomena have a physical basis in the brain and a concomitant concern to understand the go of the brain, just how the brain turned specific inputs into specific outputs. And this concern is manifest in Ashby's earliest publications. At the start of his career, in London between 1930 and 1936, he published seventeen research papers in medical journals, seeking in different ways to explore connections between mental problems and physical characteristics of the brain, often based on postmortem dissections. Such writings include his very first publication, "The Physiological Basis of the Neuroses," a
nd a three-part series, "The Brain of the Mental Defective," as well as his 1935 Cambridge MA thesis, "The Thickness of the Cerebral Cortex and Its Layers in the Mental Defective" (Ashby 1933, 1935; Ashby and Stewart 1934–35).

  Such research was by no means untypical of this period, but it appears to have led nowhere. No systematic physiological differences betwen normal and pathological brains were convincingly identified, and Ashby did not publish in this area after 1937.6 After his move to St. Andrew's Hospital in 1936, Ashby's research into insanity moved in several directions.7 The January 1937 annual report from the hospital mentions a survey of "the incidence of various mental and neurological abnormalities in the general population, so that this incidence could be compared with the incidence in the relatives of those suffering from mental or neurological disorders. . . . Dr. Ashby's work strongly suggests that heredity cannot be so important a factor as has sometimes been maintained" (Ashby 1937a). The report also mentions that Ashby and R. M. Stewart had studied the brain of one of Stewart's patients who had suffered from a rare form of brain disease (Ashby, Stewart, and Watkin 1937), and that Ashby had begun looking into tissue culture methods for the investigation of brain chemistry (Ashby 1937b). Ashby's pathological work continued to feature in the January 1938 report, as well as the fact that "Dr. Ashby has also commenced a study on the theory of organisation as applied to the nervous system. It appears to be likely to yield interesting information about the fundamental processes of the brain, and to give more information about the ways in which these processes may become deranged"—this was the beginning of Ashby's cybernetics, the topic of the next section.

  According to the St. Andrew's report from January 1941, "Various lines of research have been undertaken in connection with Hypoglycaemic Therapy. Drs. Ashby and Gibson have studied the effects of Insulin as a conditioned stimulus. Their results have been completed and form the basis of a paper awaiting publication. They are actively engaged also in studying various metabolic responses before and after treatment by Insulin and Cardiazol. The complications arising from treatment by these methods are being fully investigated and their subsequent effects, if any, carefully observed. It is hoped to publish our observations at an early date." Here we are back in the realm of the great and desperate psychiatric cures discussed in the previous chapter. Insulin and cardiazol were used to induce supposedly therapeutic convulsions in mental patients, and we can note that in this work Ashby had moved from his earlier interest in the pathological brain per se to the biological mechanisms of psychiatric treatment.

  Figure 4.2."The most important variables affected by E.C.T." Reproduced with permission from W. R. Ashby, "The Mode of Action of Electro-convulsive Therapy," Journal of Mental Science, 99 (1953), 203, fig. 1. (© 1953 The Royal College of Psychiatrists.)

  This shift in focus intensified after Ashby's move to Barnwood House in 1947. Not far from the Burden Neurological Institute, Barnwood House was at the epicenter of radical psychiatric cures in Britain. Its director, G. W. T. H. Fleming, was the first author listed, with Golla and Walter, on the first published report on the use of electroconvulsive therapy in Britain (Fleming, Golla, and Walter 1939, discussed in the previous chapter). Ashby had no doubts about the efficacy of ECT: "Electroshock therapy . . . has long passed its period of probation and is now universally accepted as active and effective." "Yet," he wrote, "its mode of action is still unknown." From its introduction there had been speculation that ECT achieved its ends not directly, via the shock itself, but by inducing some therapeutic change in the chemistry of the brain, and this was what Ashby sought to elucidate at Barnwood House, most notably in a long essay on his empirical research published in 1949, which won a prize—the £100 Burlingame Prize awarded by the Royal Medico- Psychological Association. There, Ashby reported on his own observations on fourteen mental patients who had been subjected to ECT and concluded, "The usual effect of convulsive therapy is to cause a brisk outpouring of adrenal chemical steroids during the first few days of the treatment. . . . There is evidence that [this] outpouring . . . is associated with a greater tendency to clinical recovery" (Ashby 1949a, 275, 321). Again, we see the characteristic concern to illuminate the material "go of it"—now to spell out the beginning of a chain of effects leading from the administration of electroshock to modified mental performances. And Ashby followed this up in, for example, a 1953 paper entitled "The Mode of Action of Electro-convulsive Therapy," in which he reported his own research on rats subjected to ECT, using an assay of his own devising to explore ECT's effects on the "adenohypophyseal-adrenocortical system" (Ashby 1953a; see also Ashby 1949b for earlier rat experiments on this topic).

  It is clear, then, that Ashby was actively involved in a certain kind of clinical psychiatric research well into his fifties, trying to understand the material peculiarities of pathological brains and how therapeutic interventions worked. This was his professional life until he left Britain in 1961, and I will come back to it. Now, however, we can move to a more rarefied plane and explore the development of Ashby's distinctive cybernetic understanding of the brain.

  Ashby's Hobby

  Shortly after Ashby's death, his wife wrote to Mai von Foerster, Heinz's wife and a family friend at the University of Illinois:

  I came across a very private notebook the other day written in 1951. In it Ross wrote: After I qualified, work on the brain, of the type recorded in my notebooks, was to me merely a delightful amusement, a hobby I could retreat to, a world where I could weave complex and delightful patterns of pure thought, untroubled by social, financial or other distractions. So the work which I had treated for years only as a hobby began to arouse interest. I was asked to broadcast about it in March, 1949. My fear is now that I may become conspicuous, for a book of mine is in the press. For this sort of success I have no liking. My ambitions are vague—someday to produce something faultless.8

  The notebook in question is "Passing through Nature," Ashby's biographical notebook, written between 1951 and 1957 (see note 4).9 The broadcast Ashby referred to was a thirty-minute program on BBC radio, "Imitating the Brain," transmitted on 8 March 1949, for which he was paid twenty-six pounds and five shillings (i.e., twenty-five guineas) plus fifteen shillings and threepence rail fare; the book is Design for a Brain, which appeared in 1952.10 My aim now is to trace out the evolution of the strand of Ashby's early work that led up to and included Design. I am interested in its substance and how it emerged from the hobbyist shadows to establish Ashby's reputation as one of the world's leading cyberneticians. In a biographical note from 1962 Ashby wrote that "since 1928 Ashby has given most of his attention to the problem: How can the brain be at once mechanistic and adaptive? He obtained the solution in 1941, but it was not until 1948 that the Homeostat was built to embody the special process. . . . Since then he has worked to make the theory of brainlike mechanisms clearer" (Ashby 1962, 452). I will not try to trace out the evolution of his thinking from 1928 onward; instead, I want to pick up the historical story with Ashby's first protocybernetic publication. As I said, Ashby's clinical concerns are very much marginalized in his key cybernetic works, which focus on the normal rather than the pathological brain, but we can explore the interconnections later.

  Ashby's first step in translating his hobbyist concerns into public discourse was a 1940 essay entitled "Adaptiveness and Equilibrium" published in the Journal of Mental Science. In a journal normally devoted to reports of mental illness and therapies, this paper introduced in very general terms a dynamic notion of equilibrium drawn from physics and engineering. A cube lying on one of its faces, to mention Ashby's simplest example, is in a state of dynamic equilibrium inasmuch as if one tilts it, it will fall back to its initial position. Likewise, Ashby noted, if the temperature of a chicken incubator is perturbed, its thermostat will tend to return it to its desired value. In both cases, any disturbance from the equilibrium position calls forth opposing forces that restore the system to its initial state. One can thus say that these systems are able to
adapt to fluctuations in their environment, in the sense of being able to cope with them, whatever they turn out to be. Much elaborated, this notion of adaptation ran through all of Ashby's later work on cybernetics as brain science, and we can note here that it is a different notion from the one I associated with Walter and the tortoise in the previous chapter. There "adaptation" referred to a sensitive spatial engagement with the environment, while for Ashby the defining feature of adaptation was finding and maintaining a relation of dynamic equilibrium with the world. This divergence lay at the heart of their different contributions to cybernetics.

  Why should the readers of the Journal of Mental Science be interested in all this? Ashby's idea unfolded in two steps. One was to explain that dynamic equilibrium was a key feature of life. A tendency for certain "essential variables" to remain close to some constant equilibrium value in the face of environmental fluctuations was recognized to be a feature of many organisms; Ashby referred to the pH and sugar levels of the blood and the diameter of the pupil of the eye as familiar examples. Tilted cubes and thermostats could thus be seen as formal models for real organic adaptive processes—the mechanisms of homeostasis, as it was called, though Ashby did not use that word at this point. And Ashby's second step was to assert that "in psychiatry its importance [i.e., the importance of adaptiveness] is central, for it is precisely the loss of this 'adaptiveness' which is the reason for certification [i.e., forcible confinement to a mental institution]" (478). Here he tied his essay into a venerable tradition in psychiatry going back at least to the early twentieth century, namely, that madness and mental illness pointed to a failure to adapt—an inappropriate mental fixity in the face of the flux of events (Pressman 1998, chap. 2). As we saw, Walter's M. docilis likewise lost its adaptivity when driven mad.