The Cybernetic Brain
Page 7
GREY WALTER,"PRESENTATION" (1971 [1954], 29)
The tortoises were simple and comprehensible artifacts. Anyone could understand how their two-neuron brains worked—at least, anyone familiar with the relay and triode circuits of the time. But, as Walter argued, "the variation of behaviour patterns exhibited even with such economy of structure are complex and unpredictable" (1953, 126). He noted, for example, that he had been taken by surprise by the tortoises' mirror and mating dances (1953, 130). The tortoises engaged with their environments in unexpected ways, displaying emergent properties relative to what Walter had designed into them. After the fact, of course, Walter explained such performances in terms of the tortoises' running lights, as mentioned above. But it is worth recognizing that such interpretations were themselves not beyond dispute. On the basis of his own tortoise reconstructions, Owen Holland (2003, 2101–8) was led to challenge Walter's interpretation of the source of these dances, arguing that they are a function of the oscillatory behavior set in motion by physical contact, rather than anything to do with the running lights. Here it begins to become clear that the tortoises remained mini–Black Boxes. As Walter put it, "Even in the simple models of behaviour we have described, it is often quite impossible to decide whether what the model is doing is the result of its design or its experience" (1953, 271).16
The tortoise thus again appears as ontological theater, but in a different sense from that discussed above. As a piece of engineering, it displayed the fact that a reductive knowledge of components does not necessarily translate into a predictive understanding of aggregate performance—one still has to run the machine and find out what it will do. As I said in chapter 2, I find this ontologically instructive too. Many people, including me, tend to think that the world has some determinate structure that is, in principle, fully compre hensible. What the tortoise stages us for us is that, even if that were true, we might still have to find out about the world in real-time performative interaction. For such people, it might be helpful to start by imagining the world as full of tortoiselike entities—unknowable in any predictive sense and always capable of surprising us, as the tortoise proved to be. This is another way to begin getting the hang of the ontology of cybernetics.
In his first publication on the tortoises, in Scientific American in May 1950, Walter (1950a, 44) emphasized this discovery of complexity in a striking extrapolation beyond the two-neuron tortoise brain: "It is unlikely that the number of perceptible functional elements in the brain is anything like the total number of the nerve cells; it is more likely to be of the order of 1,000. But even if it were only 10, this number of elements could provide enough variety for a lifetime of experience for all the men who ever lived or will be born if mankind survives a thousand million years." At stake here are not Walter's precise numbers (see Walter 1953, 118–20, for the calculation)—though cybernetic combinatorics readily generates enormous numbers, as we will see later. Walter was notsuggesting that given ten elements he could predict the future of the human race in classically scientific fashion. His point concerned, rather, I think, the unimaginable richness of performance that could be generated by a few simple parts articulated with one another. Even if we knew what the ten functional elements of the brain are and how they are interconnected, we would not be able to "solve" the system and thus calculate and predict all possible forms of human behavior over the next billion years. We would just have to build the system and run it, like the tortoise, to see what it would do—or we could just let history run its course and find out. In general, even if we know all that there is to know about the primitive components of a Black Box, we might still not know anything about how the ensemble will perform. At this level of aggregation, the box remains black, and this is what Walter learned from his tortoises.
_ _ _ _ _
Thus my sense of the tortoise as ontological theater—as variously conjuring up and playing out an ontological vision of performance and unknowability. We will see this ontology elaborated in all sorts of ways in the pages to follow. But here I should note two qualifications concerning just how much the tortoise can enlighten us. First, the tortoise was indeed adaptive, but only to a degree. Especially, it had fixed goals hard wired into it, such as pursuing lights. The tortoise did not evolve new goals as it went along in the world. This fixity of goals was a common feature of early cybernetic engineering, going back all the way to the steam-engine governor (which sought to keep the engine speed constant) and beyond. As ontological theater this has to be seen as a shortcoming. There is no reason to think that human beings, for example, are characterized by a fixity of goals, and every reason, in fact, to argue against it (Pickering 1995). From this angle too, then, we should see the tortoise as staging a hybrid ontology, part adaptive and part not.17 As I have said before, the adaptive aspects of cybernetics are what I want most to get into focus here, as pointing toward the unfamiliar aspects of nonmodern ontology.
The tortoise's world also left something to be desired. It was a world that, to a first approximation, never changed, a fixed array of lights and obstacles. The tortoise adapted to its environment, but the environment did nothing in response.18 There was no place for a dance of agency between the tortoise and its world. This has to be regarded as another shortcoming of Walter's cybernetics as ontological theater, and we can see in later chapters how other cybernetic systems, beginning with Ashby's homeostat, transcended this limitation.
Tortoises as Not-Brains
IT IS UP TO M. WALTER TO EXPLAIN THE IMPORTANCE OF HIS MODELS FOR PHYSIOLOGY. THE ENGINEER IS INTERESTED IN THE MACHINE THAT IMITATES SENSE ORGANS AND THE MACHINE THAT LEARNS. ONE CAN IMAGINE A DAY WHEN MACHINES THAT LEARN WOULD HAVE A GENERAL IMPORTANCE IN INDUSTRY. THAT IS WHY WE HAVE REPEATED HIS APPROACH.
HEINZ ZEMANEK,"LA TORTUE DE VIENNE ET LES AUTRES TRAVAUX CYBERNÉTIQUES" (ZEMANEK 1958, 772, MY TRANSLATION)
In the opening chapter I mentioned the protean quality of cybernetics, that although the brain was its original referent, the brain could not contain it, and I can elaborate on that remark now. I have shown how the tortoise took shape as a model of the brain and as a contribution to brain science; I will shortly explore its specific connection to psychiatry. But one did not have to see a brain when contemplating a tortoise. One could simply see a machine, an interesting example of a particular style of adaptive engineering, a robot. Here is Walter's own account of the origins of the tortoise from The Living Brain (1953, 125): "The first notion of constructing a free goal-seeking mechanism goes back to a wartime talk with the psychologist, Kenneth Craik. . . . When he was engaged on a war job for the Government, he came to get the help of our automatic [EEG] analyser with some very complicated curves he had obtained, curves relating to the aiming errors of air gunners. Goal-seeking missiles were literally much in the air in those days; so, in our minds, were scanning mechanisms. . . . The two ideas, goal-seeking and scanning, . . . combined as the essential mechanical conception of a working model that would behave like a very simple animal." Craik was a young experimental psychologist and protocybernetican, who died at the age of thirty-one in a bicycle accident in Cambridge on 18 May 1945, the last day of World War II in Europe. He was very much the British Wiener, even more heavily involved in military research into gun aiming and the like during the war, and there are clear echoes of Wiener's wartime work on autonomous weapons systems in this quotation from Walter.19 And though there is no evidence that Walter ever sought to develop the tortoise for such purposes, if one wanted to find a usefor it, an obvious thing to do would be to fix a gun next to the guiding photocell or fill its body with explosives detonated by the contact switch. And Walter was certainly well aware of such possibilities. At the end of his technical description of tortoise construction, he stated that "the model may be made into a better 'self-directing missile' by using two photocells in the usual way" (1953, 291–92).20
Walter's contribution to brain science was thus also a contribution to the history of engineering and robotics (on which mo
re below). And beyond the technical realms of brain science and robotics, the tortoises also found a place in popular culture. They were not simply technical devices. Walter showed them off and people liked them. He demonstrated the first two tortoises, Elmer and Elsie, in public in 1949, though "they were rather unreliable and required frequent attention." Three of the tortoises built by Bunny Warren were exhibited at the Festival of Britain in 1951; others were demonstrated in public regularly throughout the 1950s. They appeared on BBC television (Holland 2003, 2090–91, gives an account and analysis of a 1950 BBC newsreel on the tortoises). Walter set them loose at a meeting of the British Association for the Advancement of Science, where they displayed a lively interest in women's legs (presumably attracted to the light-reflecting qualities of nylon stockings: Hayward, 2001b).
This popular appeal, in turn, manifested itself in at least two lines of subsequent development. One was an embryonic eruption into the toy market: a tortoise was sent to the United States after the Festival of Britain as the prototype for a line of transistorized children's toys—which never went into production, alas (Holland 1996, n.d.; Hayward 2001b). One can now, however, buy construction kits for devices which are clearly versions of the tortoise. Along another axis, the tortoise entered the world of science fiction and popular entertainment. In the BBC's long-running Doctor Who TV series, I find it hard to doubt that the tortoise was the model for K-9, the Doctor's robot dog (which looked just like a tortoise, with a small tail attached). One thinks also of the Daleks, with their sinister optical scanner, and my recollection is that the Daleks were first seen in an electronic readout from a human brain which itself took the form of a TV image—another imaginative version of the cybernetic notion of scanning. What should we make of this popular appeal? It derived, I assume, from the quasi-magical properties of tortoises I mentioned in chapter 1, as mechanical devices that behaved as if they were alive. We are back in the territory of the Golem and the Sorcerer's Apprentice, and a fascination with transgression of the boundary between the animate and the inanimate. This animation of the inanimate hangs together, of course, with the implementation of the cybernetic ontology just discussed: the tortoises appeared so lively just because of their autonomy and sensitivity to their environment.
Brain science, psychiatry, robotics, toys, TV sci-fi: these are some of the areas that the tortoises contributed to. This list starts to establish what I mean by the protean quality of cybernetics, and as the book goes on, we can extend it.
The Social Basis of Cybernetics
THE MECHANICAL DESIGN [OF A TORTOISE] IS USUALLY MORE OF A PROBLEM THAN THE ELECTRICAL. . . . THERE IS NOT A GREAT CHOICE OF MOTORS; THOSE USED FOR DRIVING SMALL HOME-CONSTRUCTED MODELS ARE ADEQUATE BUT NOT EFFICIENT. . . . IT IS OFTEN ADVISABLE TO RE-BUSH THE BEARINGS. . . . THE GEAR TRAINS TO THE DRIVING AND SCANNING SHAFTS ARE THE MOST AWKWARD PARTS FOR THE AMATEUR CONSTRUCTOR. THE FIRST MODEL OF THIS SPECIES WAS FURNISHED WITH PINIONS FROM OLD CLOCKS AND GASMETERS.
GREY WALTER,THE LIVING BRAIN (1953, 290–91)
SO MANY DISCOVERIES HAVE BEEN MADE BY AMATEURS THAT THERE MUST BE A SPECIAL STATE OF MIND AND A PHASE OF SCIENTIFIC EVOLUTION WHEN TOO MUCH KNOWLEDGE IS A DANGEROUS THING. COULD ONE SAY THAT AN AMATEUR IS ONE WHO DOES NOT KNOW HIS OWN IMPOTENCE?
GREY WALTER,"TRAPS, TRICKS AND TRIUMPHS IN E.E.G." (1966, 9)
I mentioned in the opening chapter that cybernetics had an unconventional social basis as well as an unfamiliar ontology, and here we can begin the investigation of the former. One point to bear in mind is that Walter did have a steady job throughout his working life, spending the thirty-one years prior to his scooter accident at the Burden Neurological Institute. As I said, however, his career there revolved around his EEG work and electrophysiological research more generally, and the question that I want to focus on here concerns the social basis for his cybernetics as exemplified by the tortoises.
In the quotation above on Craik and the origins of the tortoise, I skipped over a phrase, "long before the home study was turned into a workshop," which precedes "the two ideas, goal-seeking and scanning, had combined." Walter built the first tortoises at home, in his spare time.21 Hence, for example, the practical advice to readers on tortoise construction just quoted. Walter's key contribution to cybernetics was, then, the work of an amateur, a hobbyist. And, as we will see, this was true of all four of our principals. In this sense, then, we can say that at its origins British cybernetics had no social basis.It emerged from nowhere as far as established fields and career paths were concerned. The cyberneticians and their projects were outsiders to established fields of endeavor.
Some discussion is appropriate here. First, it is worth emphasizing that the amateur and hobbyist roots of British cybernetics are a marker of its oddity: there was no obvious field for it to grow from. Perhaps the most likely matrix would have been experimental psychology (one thinks of Kenneth Craik) but in fact cybernetics did not originate there. Second, we should go back to the standard origin story of cybernetics, connecting it to Norbert Wiener's military research. There is, as I said in chapter 1, a contrast here between British and American cybernetics. As I have already indicated, the primary referent of Walter's tortoise work was not some piece of military technology such as Wiener's antiaircraft predictor; it was the brain. Walter always presented the tortoise precisely as a model brain, and though I just quoted him on the tortoise as a self-guided missile, this was a passing remark. And, of course, it makes sense that a brain researcher working at a neurological institute would have the brain rather than weapons systems on his mind.22
This, then, is the other origin story of cybernetics that I can develop further as we go on, the story of cybernetics as emerging from and as brain science rather than military research. This story requires some nuance, needless to say. Little research in the 1940s and 1950s was immune to military influence, and it was Craik, the British Wiener, who gave Walter the idea of scanning. Nevertheless, it would be misleading to try to center the story of British cybernetics on war; it is much more illuminating to focus on the brain.23 That said, there is another connection to warfare that is worth mentioning, which in fact deepens the contrast with Wiener.
In the second Grey Walter Memorial Lecture, the veteran EEG researcher W. A. Cobb told a story of wartime shortages of equipment and of how he eventually obtained a special timer from the wreckage of a crashed Spitfire (Cobb 1981, 61). We can take this as iconic of the conditions under which British cybernetics developed. Wiener worked on a well-funded military project at the cutting edge of research at MIT, the very heart of the U.S. militaryacademic complex; like Cobb, Walter and the other British cyberneticians cobbled together their creations from the detritus of war and a couple of centuries of industrialization.24 The electronic components of machines like the tortoise were availably cheaply as war surplus (Hayward 2001b, 300), and, as Walter said, other parts were salvaged from old clocks and gas meters. If Wiener's cybernetics grew directly out of a military project, Walter's was instead improvised in a material culture left over from the war.
One last remark on the origins of British cybernetics. Inescapably associated with the notions of the amateur and the hobbyist are notions of sheer pleasure and fun. Just as there is no reason to doubt that Walter intended the tortoises as a serious contribution to brain science, there is no reason to doubt that he had fun building them and watching them perform. This theme of having fun is another that runs through the history of British cybernetics and again presents a stark contrast with that of cybernetics in the United States, where the only fun one senses in reading the proceedings of the Macy Conferences is the familiar and rather grim academic pleasure of the cut and thrust of scholarly debate. The chairman of the meetings, Warren McCulloch (2004, 356), recalled: "We were unable to behave in a familiar, friendly or even civil manner. The first five meetings were intolerable. Some participants left in tears, never to return. We tried some sessions with and some without recording, but nothing was printable. The smoke, the n
oise, the smell of battle are not printable." Of the many conventional boundaries and dichotomies that British cybernetics undermined, that between work and fun was not the least.
We can turn from the origins of British cybernetics to its propagation. Walter made no secret of his hobby; quite the reverse: he publicized the tortoises widely, engaging with at least three rather distinct audiences which we can discuss in turn. The first audience was the general public. According to Owen Holland (2003, 2090), "by late 1949, Grey Walter was demonstrating Elmer and Elsie, the first two tortoises, to the press, with all the showmanship that some held against him," and the first major press report appeared in the Daily Expresson 13 December 1949, written by Chapman Pincher. The BBC TV newsreel mentioned above followed in 1950, and so on. Outside the world of journalism, Walter himself wrote for a popular readership. The principal published technical sources on the tortoises are Walter's two articles in Scientific American in May 1950 and August 1951 and his 1953 popular book The Living Brain.These contributed greatly to the public visibility of Walter and the tortoises, but let me postpone discussion of substantive outcomes of this publicity for a while.
As an academic myself, I have tended to assume that the proper readership and home for a field like cybernetics would be a scholarly one. Walter did not publish any detailed accounts of the tortoises alone in the scholarly literature, but in the early 1950s they often featured as parts of his emerging account of the brain as otherwise explored in EEG research. A lecture delivered to a psychiatric audience published in January 1950, for example, began with a discussion of the tortoises (not named as such), their complexity of behavior, and the significance of scanning, before plunging into the details of EEG findings and their interpretation (Walter 1950b, 3–6). But it is also safe to say that the major impact of cybernetics was not centered on any established field. Historical overviews of twentieth-century psychiatry, for example (on which more shortly), make little or no mention of cybernetics (e.g., Valenstein 1986; Shorter 1997).25And one can see why this should have been. The combination of brain science and engineering made concrete in the tortoises was a strange one, both to the sciences of the brain (neurophysiology, EEG research, psychology, psychiatry) and, from the other direction, to engineering. To do any of these disciplines on the model of Walter and the tortoises would have required drastic shifts in practice, which are much harder to make than any simple shift in the realm of ideas.