The Cybernetic Brain

by Andrew Pickering

  We could start with issues of representation and performance. In the discussion that followed Beer's presentation at the 1960 Allerton conference, Beer made an interesting contrast between digital and biological computing in just these terms. When the subject of the former came up, he remarked that "this analogy with computers I do not like for two reasons." One had to do with the dynamics of memory and whether memory should be understood like the storage of "a parcel in a cloakroom" or as a "path of facilitation through phase space." The other went like this (1962a, 220–21):

  The big electronic machines . . . are preoccupied with digital access. Now why is this? It is always possible, given an output channel which you can fit on somewhere, to say what is happening just there, and to get an enormous printout. Now we [Beer and Pask] are not concerned with digital access, but with outcomes. Why do we pay so much money to make it [digital output] available? In the sort of machines that Gordon and I have been concerned with, you cannot get at the intermediate answer. If you take out [one?] of Gordon's dishes of colloid, you may be effectively inverting a matrix of the order 20,000. The cost of the computer is perhaps 10 cents. The only trouble is you do not know what the answer is. Now this sounds absurdly naïve, but it is not, you know, because you do not want the answer. What you do want is to use this answer. So why ever digitise it?

  We are back to the notion of representation as a detour away from performance. Digital computing, in this sense, is an enormous detour away from its object—the functioning of a factory for example—into and through a world of symbols. In the previous chapter we discussed the discovery at Kingsley Hall and Archway that this detour could be drastically shortened or even done away with in therapeutic practice. But Beer started from this realization: in a world of exceedingly complex systems, for which any representation can only be provisional, performance is what we need to care about. The important thing is that the firm adapts to its ever-changing environment, not that we find the right representation of either entity. As ontological theater, then, Beer and Pask's biological computers stage this performative ontology vividly for us, dispensing entirely with representation, both exemplifying an ontology of sheer performance and indicating how one might go on in computing if one took it seriously. I could note here that this concern for performance and a suspicion of representation per se is a theme that ran through all of Beer's work.13

  There is second and related sense of a detour that also deserves attention here. As Beer put it (1962a, 209, 215), "As a constructor of machines man has become accustomed to regard his materials as inert lumps of matter which have to be fashioned and assembled to make a useful system. He does not normally think first of materials as having an intrinsically high variety which has to be constrained. . . . [But] we do not want a lot of bits and pieces which we have got to put together. Because once we settle for [that], we have got to have a blueprint. We have got to design the damn thing; and that is just what we do not want to do." The echoes of Ashby on DAMS and the blueprint attitude are clear. We are back to the contrasting conceptions of design that go with the modern ontology of knowable systems and the cybernetic ontology of unknowable ones. Within the frame of modern science and engineering, design entails figuring out what needs to be done to achieve some result and then arranging "inert lumps of matter" to achieve those specifications. Digital computers depend on this sort of design, specifying material configurations right down to the molecular level of chemical elements on silicon chips. Beer's idea instead was, as we have seen, to find lively (not inert) chunks of matter and to try to enroll their agency directly into his projects. This gets us back to the discussion of the hylozoist quality of biofeedback music (chap. 3) and the idea that it's all there already in nature (as in the extraction of music from the material brain). We could say that the modern stance on design has no faith in matter and relies upon human representations and agency to achieve its effects. The cybernetic ontology, as Beer staged it, entailed a faith in the agency of matter: whatever ends we aim at, some chunk of nature probably already exists that can help us along the way. We don't need these long detours through modern design. We can explore Beer's hylozoism further later in the chapter in a broader discussion of his spirituality.

  There is, of course, yet a third sense of detour that comes to mind here. The mastery of matter, from the molecular level upward, required to build a digital computer has been painstakingly acquired over centuries of technscientific effort. Beer's argument was, in effect, that perhaps we didn't need to make the trek. Just to be able to suggest that is another striking manifestation of the difference that ontology makes.

  Now, Heidegger. It makes sense to see modern computer engineering as operating in the mode of enframing. It is not that semiconductor engineers, for example, have actually achieved some magical mastery over matter. For all their representational knowledge, they remain, like the rest of us, in medias res, obliged to struggle with the performance of obstinate stuff (Lécuyer and Brock 2006). Nevertheless, a successful chip is one that fits in with our preconceived plans: the chip either manipulates binary variables in a regular fashion, or it does not—in which case it is junk. Bending matter to our will like that is just what Heidegger meant by enframing. And then we can begin, at least, to see that the cybernetic ontology in this instance has more in common ;with a stance of revealing. Beer wanted to find out what the world—assemblages of mice, Daphnia, his local pond—could offer us. Against this, one might argue that Beer had some definite end in view: a replacement for the human manager of the factory. But the important point to note is that the pond was not envisaged as an identical substitute for the human. We will see in the next chapter that Pask, who thought this through in print further than Beer, was clear that biological computers would have their own management style, not identical to any human manager—and that we would, indeed, have to find out what that style was, and whether we could adapt to and live with it. This is the sense in which this form of cybernetic design in the thick of things is a stance of revealing rather than enframing.

  One last thought in this connection. Somewhere along the line when one tries to get grips with Beer on biological computing, an apparent paradox surfaces. Beer's goal, all along, was to improve management. The cybernetic factory was supposed to be an improvement on existing factories with their human managers. And yet the cybernetic brain of the factory was supposed to be a colony of insects, some dead leaves for them to feed on, the odd leech. Did Beer really think that his local pond was cleverer than he was? In a way, the answer has to be that he did, but we should be clear what way that was. Recall that Beer thought that the economic environment of the factory was itself an exceedingly complex system, ultimately unknowable and always becoming something new. He therefore felt that this environment would always be setting managers problems that our usual modes of cognition are simply unable to solve. This connects straight back to the above remarks on Beer's scepticism toward representational knowledge. On the other hand, according to Beer, biological systems can solve these problems that are beyond our cognitive capacity. They can adapt to unforeseeable fluctuations and changes. The pond survives. Our bodies maintain our temperatures close to constant whatever we eat, whatever we do, in all sorts of physical environments. It seems more than likely that if we were given conscious control over all the parameters that bear on our internal milieu, our cognitive abilities would not prove equal to the task of maintaining our essential variables within bounds and we would quickly die. This, then, is the sense in which Beer thought that ecosystems are smarter than we are—not in their representational cognitive abilities, which one might think are nonexistent, but in their performative ability to solve problems that exceed our cognitive ones. In biological computers, the hope was that "solutions to problems simply grow" (1962a, 211).

  The Social Basis of Beer's Cybernetics

  At United Steel, Beer was the director of a large operations research group, members of which he involved in the simulation of the cybernetic fa
ctory at the Templeborough Rolling Mills. This was a serious OR exercise, supported by his company. The key ingredient, however, in moving from the simulation to the cybernetic reality, was the U-machine, and, as Beer remarked in opening his 1962 status report on biological computing, "everything that follows is very much a spare time activity for me, although I am doing my best to keep the work alive—for I have a conviction that it will ultimately pay off. Ideally, an endowed project is required to finance my company's Cybernetic Research Unit in this fundamental work" (1962b, 25). I quoted Beer above on tinkering with tanks in the middle of the night, evidently at home, and Beer's daughter Vanilla has, in fact, fond childhood memories of weekend walks with her father to collect water from local ponds (conversation with the author, 22 June 2002). We are back once more on the terrain of amateurism, ten years after Walter had worked at home on his tortoises and Ashby on his homeostat.

  Again, then, a distinctive cybernetic initiative sprang up and flourished for some years in a private space, outside any established social institution. And, as usual, one can see why that was. Beer's work looked wrong. Tinkering with tanks full of pond water looked neither like OR nor like any plausible extension of OR. It was the kind of thing an academic biologist might do, but biologists are not concerned with managing factories. The other side of the protean quality of cybernetics meant that, in this instance, too, it had no obvious home, and the ontological mismatch found its parallel in the social world. I do not know whether Beer ever proposed to the higher management of United Steel or to the sponsors of his consulting company, SIGMA, that they should support his research on biological computing, but it is not surprising that he should be thinking wistfully of an endowed project in 1962, or that such was not forthcoming. We should, indeed, note that Beer failed to construct a working U-machine, or even a convincing prototype. This is, no doubt, part of the explanation for the collapse of Beer's (and Pask's) research in this area after 1962. But it is only part of the explanation. The electronic computer would not have got very far, either, if its development had been left solely to a handful of hobbyists.

  Of course, Beer did not carry on his cybernetics in total isolation. As mentioned above, having read Wiener's Cybernetics in 1950, he sought out and got to know many of the leading cyberneticians in the United States as well as Britain. In the process, he quickly became a highly respected member of the cybernetics community which existed transversely to the conventional institutions to which its members also belonged. It was Beer who first brought Ashby and Pask together, by inviting both of them to a lecture he gave in the city hall in Sheffield in 1956, and his recollection of the meeting sheds some light on the characters of both (S. Beer 2001, 553): "Gordon was speaking in his familiar style—evocative, mercurial, allusory. He would wave his arms about and try to capture some fleeting insight or to give expression to some half-formed thought. I was used to this—as I was to Ross's rather punctilious manner. So Ashby would constantly interrupt Gordon's stream of consciousness to say, 'Excuse me, what exactly do you mean by that?' or 'Would you define that term?' Both were somewhat frustrated, and the evening was close to disaster." Beyond his personal involvement in the cybernetics community, Beer appreciated the importance of establishing a reliable social basis for the transmission and elaboration of cybernetics more than the other British cyberneticians. Ross Ashby also presented his work at the 1960 conference at which Beer presented "Towards the Cybernetic Factory," and while there Beer conspired with Heinz von Foerster to offer Ashby the position that took him to the University of Illinois (Beer 1994 [1960], 299–301). In the second half of the 1960s, when Beer was development director of the International Publishing Corporation, he conceived the idea of establishing a National Institute of Cybernetics at the new Brunel University in Uxbridge, London, aiming to create academic positions for both Gordon Pask and Frank George. Beer persuaded the chairman of IPC, Cecil King, to fund part of the endowment for the institute and a fund-raising dinner for the great and good of the British establishment was planned (with Lord Mountbatten, the queen's uncle, and Angus Ogilvy, the husband of Princess Alexandra, among the guests). Unfortunately, before the dinner could take place there was a palace coup at IPC—"in which, ironically, I [Beer] was involved"—which resulted in the replacement of King by Hugh Cudlipp as chairman.

  I had never managed to explain even the rudiments of cybernetics to him [Cudlipp]. Moreover, it is probably fair to say that he was not one of my greatest fans. . . . At any rate the dinner broke up in some disorder, without a single donation forthcoming. Dr Topping [the vice-chancellor at Brunel] went ahead with the plan insofar as he was able, based on the solitary commitment that Cecil King had made which the new Chairman was too late to withdraw. Gordon was greatly disappointed, and he could not bring his own operation (as he had intended) [System Research, discussed in the next chapter] into the ambit of the diminished Institute which soon became a simple department at Brunel. The funding was just not there. However, both he and Frank George used their Chairs on the diminished scale. (S. Beer 2001, 557)

  Though Beer had not fully achieved his ambition, the establishment of the Department of Cybernetics at Brunel was the zenith of the institutional career of cybernetics in Britain, and we shall see in the next chapter that Pask made good use of his position there in training a third generation of cyberneticians. Characteristically, the trajectory of cybernetics in Britain was further refracted at Brunel, with Pask's PhD students focusing on such topics as teaching machines and architecture. The Brunel department closed down in the early 1980s, and, given the lack of other institutional initiatives, these students were once more left to improvise a basis for their careers.14

  In the 1960s, then, Beer helped find academic positions for three of Britain's leading cyberneticians and played a major role in establishing an academic department of cybernetics. Conversely, as remarked already, in 1974 Beer effectively deinstitutionalized himself in moving to a cottage in Wales. Partly, as I said, this was an aspect of an overall shift in lifestyle; partly it was a response to events in Chile. Partly, too, I think, it was a reflection of his failure in the later 1960s to persuade Britain's Labour government of the importance of cybernetics. He wrote of his "disappointment in the performance of Harold Wilson's 'white heat of technology' government. This was operating at a barely perceptible glow, and the ministers with whom I had been trying to design a whole new strategy for national computing failed to exert any real clout. There were five ministers involved—the Postmaster General himself (John Stonehouse) 'did a runner' and was discovered much later in Australia" (S. Beer 2001, 556). Beer was an exceptionally well connected spokesman for cybernetics in the 1960s, but the fruits of his efforts were relatively few. As he once put it to me, speaking of the sixties, "the Establishment beat us" (phone conversation, 3 June 1999).15

  The Afterlife of Biological Computing

  Neither Beer nor Pask ever repudiated his biological computer work; both continued to mention it favorably after the 1960s. In his 1982 popular book, Micro Man, Pask discusses a variety of "maverick machines," including his electrochemical systems, which he describes as "dendritic." He mentions that improved versions of them have been built by R. M. Stewart in California and comments that "there is now a demand for such devices, which are appropriate to non-logical forms of computation, but dendrites . . . are physically too cumbersome for such demand to be met practically. It now seems that biological media may perform in similar fashion but on a more manageable scale" (Pask and Curran 1982, 135). A few pages later he actually reproduces a picture of a pond, with the caption "A real-life modular processor?" Likewise, Beer in the text he wrote for a popular book on the history of computing, Pebbles to Computers: "Some thirty years ago, some scientists began to think that biological computers might be constructed to outpace even electronic achievement. At that time it was not clear that transistors themselves would become reliable! Attempts were made to implicate living cells—microorganisms—in computations. In England in th
e 'fifties, one such computer solved an equation in four hours that a bright school girl or boy could solve in (maximum) four minutes. Its time had not yet come!" (Blohm, Beer, and Suzuki 1986, 13).

  Biological computing enjoyed a happier fate in science fiction, making its way into the popular imagination. With Beer's experiments on mice with cheese as a "reward function" we are surely in the presence of the mouse-computer that turns up in both Douglas Adams's The Hitchhiker's Guide to the Galaxy (1979) and Terry Pratchett's Discworld series of fantasy novels.16 The most convincing representations of biological computing that I have come across include the obviously organic control systems of alien space ships that featured in various episodes of Doctor Whoand, more recently, in Greg Bear's novel Slant(1997), which includes a biological computer called Roddy (recombinant optimized DNA device) that is an entire ecosystem of bees, wasps, ants, peas, and bacteria (and which succeeds in subverting the world's most sophisticated conventional AI, Jill).

  And back in the material world biological computing has, in fact, recently been experiencing a resurgence. Figure 6.9 shows a cockroach-controlled robot, recently built by Garnet Hertz in the Arts, Computing, Engineering Masters Program at the University of California, Irvine. A giant Madagascan cockroach stands on the white trackball at the top of the assembly, attached by Velcro on its back to the arm which loops above the other components. Motions of the cockroach's legs rotate the trackball, which in turn controls the motions of the cart (much as a trackball can be used to control the motion of the cursor on a computer screen). Infrared sensors detect when the cart is approaching an obstacle and trigger the appropriate light from an array that surrounds the roach. Since roaches tend to avoid light, this causes the roach to head off in another direction. The entire assemblage thus explores its environment without hitting anything or getting stuck—ideally, at least. The cybernetic filiations of this robot are obvious. From one angle, it is a version of Grey Walter's tortoise, five decades on. From the other, a lively biological agent replaces the precisely designed electronic circuitry of the tortoise's brain, exemplifying nicely the sense of "biological computing."17 Figure 6.10 shows another biorobot, this one built by Eduardo Kac as part of his installation The Eighth Day. This time, the robot is controlled by a slime mold. These machines have no functional purpose. They are artworks, staging for the viewer a cybernetic ontology of entrained lively nonhuman agency. We can return to the topic of cybernetic art at the end of this chapter. For now, we might note that back in the 1950s and early 1960s Beer and Pask were aiming at something much more ambitious than Hertz and Kac, to latch onto the adaptive properties of biological systems, rather than their basic tropic tendencies.18